ABSTRACT

In design flood computation procedure, selection of appropriate design rainfall distribution is

necessary. If a hydraulic structure is under-designed, the floods may cause serious damages

to the structures. On the other hand, if in case of over-designed structures, the cost of the

project may increase substantially. Hyetographs and depth-duration-frequency relationships

are important components of the rainfall runoff models. Design hyetographs are used in

conjunction with the unit hydrographs to obtain design flood. Methods are proposed to

establish regional design hyetographs for determination of design floods for ungauged basins

in Swat region. Hyetographs have been constructed according to regional methods proposed

by Khan in 1980, as well as conventional methods like Huff quartile curves and triangular

hyetograph. The computation and comparisons of design floods have been carried out for

developed curves and the Soil Conservation Service curves. The results reveal that significant

underestimation of the peak flood is resulted when Huff curves and SCS type-1A curves are

used in computation. Quite the opposite, the SCS type-2 and SCS type-3 curves produce

overestimated design flood. The results of design flood with Khan hyetograph, triangular

hyetograph and SCS type-1 methods are similar. It was concluded that as well as storm

i
distribution as witnessed in 2010 was the most critical and can be adopted as design

hyetograph for hydrological analysis of structures within the region in the future. The 100

year return period was considered appropriate in the present study for flood computation in

the Swat River Basin. The rainfall frequency analysis reveals that 100 year return period

maximum 24 hour rainfall at Kalam, Daggar, Bisham Qila and Mardan is 147, 199, 150 and

199 mm, respectively. Peak floods were estimated with the help of calibrated HEC HMS

model using various rainfall distributions. The 100 year return period floods in the Swat

region at the confluence of the rivers Swat and Panjkora is worked out as 12960, 10300,

16300, 16650, 12020, 11500 and 8650 m3/sec for SCS Type 1, SCS Type 1-A, SCS Type 2,

SCS Type 3, triangular hyetograph, Khan hyetograph and Huff curves, respectively.

ii
 ACKNOWLEDGEMENT

Thanks to Allah, with whose will and grace this thesis has been made possible.

I take this opportunity to express my gratitude to all those people who have been instrumental

in completion of this thesis research.

I pay special thanks to my advisor Dr. Muhammad Tousif Bhatti, without his continuous

guidance and moral support the thesis would not have been possible. He was always there to

help me and provide me guidance. He always remained a great source of inspiration and

motivation to me, and provided me with all the resources I needed.

I am obliged to Prof. Dr. Muhammad Latif, Director, CEWRE, for providing academic and

research environment throughout the course of study.

I am also very thankful to respected Prof. Dr. Iftikhar Ahmad for providing guidance and

valuable suggestions throughout this study in managing the research. He was always very

helping and kind to me.

I am also very thankful to Dr. Saleem Sarwar for providing supervision and idea of this

research.

During this work my parents, family, my friends Adnan Maqbool Niazi, Ehsan Khan,

Zeeshan Shahzad and colleagues cooperated, for whom I have great regard, and I wish to

extend my warmest thanks to all those who have helped me with my work.

iii
 TABLE OF CONTENTS

Chapter Description
Page
#

ABSTRACT................................................................................................................................i
ACKNOWLEDGEMENT........................................................................................................iii
TABLE OF CONTENTS……………………………………………………………………..iv
LIST OF TABLES...................................................................................................................vii
LIST OF FIGURES..................................................................................................................ix
LIST OF ABBREVATIONS……………………………………………………………….. xi

I INTRODUCTION.........................................................................................................1

1.1. GENRAL........................................................................................................................1
1.2. RESEARCH BACKGROUND......................................................................................1
1.3. OBJECTIVE..................................................................................................................2
1.4. SCOPE OF RESEARCH...............................................................................................2

II REVIEW OF LITERATURE.........................................................................................4

2.1. INTRODUCTION..........................................................................................................4
2.2. DEPTH DURATION FREQUENCY RELATIONSHIP..............................................4
2.3. DEVELOPMENT OF HYETOGRAPH........................................................................6

III THEORATICAL BACKGROUND...............................................................................9

3.1 HYETOGRAPH ESTIMATION METHODS...............................................................9

3.1.1. STORM EVENT ANALYSIS...............................................................................9
3.1.2. IDF RELATIONSHIP..........................................................................................12
3.1.3. STOCHASTIC APPROCH..................................................................................15

IV METHODOLOGY......................................................................................................17

4.1 CHARECTERISTICS OF STUDY AREA.................................................................17

4.2 COLLECTION OF DATA..........................................................................................18
4.3 FREQUENCY ANALYSIS.........................................................................................21
4.4 PRESENTATION OF RAINFALL DATA.................................................................22

iv
4.4.1 INTENSITY-DURATION-FREQUENCY (IDF) CURVE.................................23
Table of Contents (continued)

4.5 DEPTH-DURATION-FREQUENCY RELATIONSHIP............................................26

4.6 WIRASAT ULLAH KHAN METHOD OF HYETOGRAPH DEVELOPMENT......27
4.7 HUFF CURVES...........................................................................................................28
4.8 TRIANGULAR HYETOGRAPH................................................................................29
4.9 CATCHMENT AREA.................................................................................................31
4.10 DIGITAL ELEVATION MODEL (DEM)..................................................................33
4.11 FLOOD ESTIMATION METHOD.............................................................................34
4.12 HEC-HMS SOFTWARE.............................................................................................34
4.12.1 LOSS METHOD SELECTION...........................................................................35
4.12.2 TRANSFORM METHOD...................................................................................36
4.12.3 BASE FLOW.......................................................................................................40
4.12.4 REACH ROUTING METHOD...........................................................................42
4.12.5 METEOROLOGIC MODEL...............................................................................43
4.12.6 HYDROLOGIC SIMULATION.........................................................................44
4.13 MODEL CALIBRATION...........................................................................................44
4.14 DESIGN RAINFALL..................................................................................................47
4.15 FLOOD ESTIMATION...............................................................................................48

V RESULTS AND DISCUSSION..................................................................................50

5.1. DEVELOPMENT OF DEPTH DURATION FREQUENCY CURVES.....................50

5.1.1. DEPTH-FREQUENCY RELATIONSHIP ( Step I )...........................................50
5.1.2. DEPTH-DURATION RELATIONSHIP ( Step II ).............................................53
5.1.3. DEPTH-DURATION-FREQUENCY RELATIONSHIP ( Step III )..................54
5.2. DEVELOPMENT OF DESIGN HYETOGRAPHS....................................................59
5.2.1. WARASAT ULLAH KHAN METHOD.............................................................59
5.2.2. HUFF CURVES...................................................................................................61
5.2.3. TRIANGULAR HYETOGRAPH........................................................................63
5.3. COMPARISION BETWEEN HYETOGRAPHS........................................................64
5.4. COMPARISIONS OF HYETOGRAPHS WITH 2010 RAINFALL EVENT............67
5.5. HEC-HMS MODEL CALIBRATION........................................................................67
5.6. COMPUTATION OF PEAK FLOODS USING HEC-HMS......................................68
5.7. COMPARISION BETWEEN FLOOD HYDROGRAPHS.........................................69

v
5.7.1. FLOOD HYDROGRAPHS WITH MORDERATE FLOOD PEAKS................73
Table of Contents (continued)

5.7.2. FLOOD HYDROGRAPHS WITH LOW FLOOD PEAKS................................74

VI CONCLUSIONS AND RECOMMENDATIONS.......................................................75

6.1. CONCLUSIONS.......................................................................................................75
6.2. RECOMMENDATIONS.............................................................................................76

REFERENCES.........................................................................................................................77

APPENDIX I RAINFALL DATA...………………….…………………………………...82

vi
 LIST OF TABLES

Table No. Title Page #

2.1 Depth-Frequency relationship with 1 hour data after Khan (1986)............................4

4.1 Inventory of stream gauges and rain gauges............................................................19

4.2 Statistics of selected storms for huff curves development.......................................29

4.3 Curve no of AMC-I and III against AMC-II............................................................36

4.4 Catchment parameters of the study area...................................................................38

4.5 Parameters of reach routing......................................................................................43

4.6 Representative rain gauge for sub basins.................................................................45

4.7 Actual storm distribution for model calibration ‘28 to 29 June 2007’.....................46

4.8 1-day design rainfall for swat basin by Thessien polygon Method..........................47

4.9 Rainfall distribution (mm) used in HEC-HMS model in addition to SCS type
distribution................................................................................................................49

5.1 Depth-Frequency relationships at selected raingauge station...................................50

5.2 Depth duration relationships at Selected raingauge Station.....................................53

5.3 Depth duration frequency relationships for Kalam..................................................55

5.4 Depth duration frequency relationships for Daggar.................................................56

5.5 Depth duration frequency relationships for Mardan.................................................57

5.6 Depth duration frequency relationships for Besham Qila........................................58

5.7 Summary of results for Khan’s Method...................................................................59

5.8 Regional rainfall distribution for Swat region with Khan’s Method........................60

vii
5.9 Fractions for time distribution of Huff curves……………………………………62

5.10 Fractions for cumulative rainfall depths and intensity of resulted triangular
hyetograph………………………………………………………………………..64

5.11 Time distributions of Khan, Huff, Triangular,SCS Type-1,2,3,1A and 2010

storm………………………………………………………………..…………….66

5.12 Peak floods with various hyetographs at outlet (Cumecs)…….………………....69

viii
 LIST OF FIGURES

Figure No. Title Page

#

3.1 First, second, third,and fourth-quartile distributions of storm rainfall for Illinois...10

3.2 Triangular Hyetograph.............................................................................................11

3.3 Synthetic storm hyetograph by SCS (1986).............................................................12

3.4 Keifer and Chu Hyetograph with peak r=0.5...........................................................14

3.5 Representative Hyetograph with Alternative Block Method...................................15

4.1 Locations of rain gauges and stream gauging stations.............................................20

4.2 IDF curves for Kalam...............................................................................................24

4.3 IDF curves for Besham Qila.....................................................................................24

4.4 IDF curves for Daggar..............................................................................................25

4.5 IDF curves for Mardan.............................................................................................25

4.6 Altitude variation of study area................................................................................32

4.7 Map of the selected basin.........................................................................................33

4.8 Available elements in HEC-HMS software..............................................................35

4.9 Schematic of study area in HEC-HMS model..........................................................39

4.10 Flow hydrographs for the wettest month..................................................................41

4.11 Thessien polygons of study area...............................................................................48

4.12 Mass curves of rainfall distribution (mm) used in HEC-HMS model in addition to
SCS type distribution................................................................................................49

5.1 Frequency analysis of 1-day annual maximum rainfall at Kalam............................51

ix
5.2 Frequency analysis of 1-day annual maximum rainfall at Daggar...........................51

5.3 Frequency analysis of 1-day annual maximum rainfall at Mardan..........................52

5.4 Frequency analysis of 1-day annual maximum rainfall at Besham Qila..................52

5.5 Mass curves of depth duration relationship..............................................................54

5.6 Mass curves of depth duration frequency relationship for Kalam............................55

5.7 Mass curves of depth duration frequency relationship for Daggar..........................56

5.8 Mass curves of depth duration frequency relationship for Mardan..........................57

5.9 Mass curves of depth duration frequency relationship for Besham Qila.................58

5.10 Regional hyetograph for Swat region with Khan’s Method.....................................61

5.11 Mass curves for Huff curves.....................................................................................62

5.12 Representative Triangular hyetograph for Swat region............................................63

5.13 Mass curves of hyetographs developed from various methods and prevailing
curves........................................................................................................................65

5.14 Resulted hydrograph of model calibration...............................................................68

5.15 Flood Hydrographs with Khan’s hyetograph...........................................................69

5.16 Flood hydrographs with triangular hyetograph........................................................70

5.17 Flood hydrographs with Huff curves........................................................................70

5.18 Flood Hydrographs with SCS type-1........................................................................71

5.19 Flood hydrograph with SCS type-2..........................................................................71

5.20 Flood hydrograph with SCS type-3..........................................................................72

5.21 Flood hydrograph with SCS type-1A.......................................................................72

5.22 Flood hydrographs with SCS Type-1, Triangular and Khan’s Hyetograph.............74

x
 LIST OF ABBREVATIONS

ASTER Advanced Space Born Thermal Emission and

Reflection

Cumecs Cubic meter per second

Cº Degree centigrade

D Degree

DDF Depth-Duration-Frequency

EV I Extreme value Type I

GoP Government of Pakistan

іave Average Intensity

IDF Intensity-Duration-Frequency

KPK Khyber Pakhton Khawa

M Minutes

mm millimeters

N North

PMD Pakistan Meteorological Department

SRTM Shuttle Radar Topography Mission

SWHP Surface Water Hydrology Project

Td Total Storm Duration

T Return Period

Tc Time of Concentration

xi
 Chapter I
INTRODUCTION

1.1. GENRAL

The hydrograph of storm runoff from a drainage area is based partly on the time distribution

of rainfall during the storm. The variation in rainfall intensity that occurs from the beginning

of the storm through the storm peak and the end of the storm is represented in the time

distribution of rainfall that can be represented graphically as a hyetograph. Whereas,

hyetograph is a representation of the rainfall distribution over time. To design any important

hydraulic structure such as bridges, dams and barrages requires ‘design flood’ to be

determined at the location of this structure. At places where appropriate historic flow data is

available, design floods are computed by carrying out frequency analysis of the historic

floods. Whereas, rainfall runoff models are commonly develop for un-gauged basins. The

design rainfall is required to compute design flood from a rainfall runoff model. Design

rainfall is determined by carrying out frequency analysis of historic annual maximum rainfall

events of estimated design storm duration. The design rainfall is applied in rainfall runoff

models in the form of its temporal and spatial distributions. The most important aspect of this

application is the selection of design rainfall-hyetograph. In hydrologic design problems,

three parameters of the hyetograph (i.e. the peak rainfall duration, percentage of storm

magnitude in peak duration and its position in storm) are particularly important for the

development of design hyetograph. If regional studies are available, local storm distributions

are applied used in rainfall runoff analysis. Otherwise standard storm distributions can be

used to develop relationship between rainfall and the consequent runoff.

1.2. RESEARCH BACKGROUND

Information about the magnitude and frequencies of extreme precipitation, in many cases, is

not available in Pakistan. However, some regional rainfall distributions are available in

1
Baluchistan province. Intense rainfall and floods were witnessed in the Indus River

catchment in 2010 during monsoon season. Swat region is one of the areas which

experienced most significant downpour in between 25 July 2010 and 01 August 2010. This

rainfall resulted in severe floods within the Swat valley, the likes of which have not been

witnessed by inhabitants in history. The extraordinary rainfall event of 2010 provide a unique

opportunity to analyze the storm pattern that prevailed over Swat and determine design

rainfall hyetograph for the Swat region, which can later be used to estimate design floods for

important hydraulic structures within the region.

1.3. OBJECTIVE

The specific objectives of the study are as follows:

1. To establish depth-duration-frequency relationship at selected gauging stations in the

Swat region.

2. To develop a regional hyetograph by using long term record of annual daily maximum

rainfall and compare it with the prevailing curves.

3. To compute and compare peak flood in the selected region using regional hyetograph

and standard rainfall distributions in HEC-HMS model.

1.4. SCOPE OF RESEARCH

The study will help developing design hyetograph for the Swat region that can be utilized for

estimating design flood at any proposed hydraulic structure. The development of regional

hyetograph include in-depth investigation on type, pattern and durations of historic as well as

most recent extreme rainfall events experienced in the selected study area. The design storm

behavior in the form of intensity-duration-frequency (IDF) curves would increase the level of

2
confidence while estimating the design flood based on rainfall-runoff relationship. The results

of present research will be helpful for hydrologists, researchers and consulting firms.

3
 Chapter II
REVIEW OF LITERATURE

2.1. INTRODUCTION

Many scientists have carried out research on design storm distribution in the last six decades.

Some approaches of design storm distribution are used worldwide whereas others are based

on regional conditions using empirical methods. The approaches to establish depth-duration-

frequency relationship and hyetograph development are discussed in this chapter.

2.2. DEPTH DURATION FREQUENCY RELATIONSHIP

Khan (1986) analyzed the daily rainfall record for a large number of stations in Baluchistan,

Pakistan. Separate analysis was carried out on 1 hour rainfall records at Quetta, which was

the only station with at least ten years of record. Depth-duration-frequency (DDF)

relationships were developed in two steps. In the first step depth-frequency relationship was

established with daily recorded historic data by taking average of design rainfall for all

nearby rain gauges.

The threshold of rainfall magnitude was decided by making criteria that a duration having

rainfall magnitude less than the decided value was not included in analysis for all durations

(1hr, 2hr, 3hr… 24hr).

In the second step, peak over threshold (POT) analysis was carried out to develop depth

duration relationship on the short lengths of hourly data. The results of analyses are given

Table 2.1 showing the depth-frequency relationship developed with 1 hour rainfall data.

Table 2.1. Depth-Frequency relationship with 1 hour data after Khan (1986)

Return Period (years)

2 5 10 25 50 100

Rainfall (inches) 0.61 1.01 1.27 1.62 1.88 2.14

4
USGS (2005) conducted a regional frequency analyses to estimate the depth-duration

frequency of precipitation for 12 durations i.e. (15, 30, and 60 minutes; 1, 2, 3, 6, 12, and 24

hours; and 1, 3, and 7 days) in Oklahoma, USA. Three different rain-gage networks data

(15minute, 1-hour, and 1-day) were used. The 60-minute, 1-hour, 24-hour, and 1-day

durations were analyzed separately. The precipitation depths for selected frequencies were

contoured from weighted depth surfaces to produce maps. The precipitation depth-duration-

frequency curve for selected storm durations could be determined from these maps for any

site in Oklahoma.

Overeem et al. (2009) used radar data having higher temporal and spatial resolution to

develop depth-duration-frequency (DDF) relationship in the Netherland. The radar data of 11

years (1997 to 2008) for durations of 15 min to 24 hours was used for detailed analysis. The

radar data were adjusted using rain gauges by combining an hourly mean-field bias

adjustment with a daily spatial adjustment. Assuming a generalized extreme value (GEV)

distribution, the index flood method was used to describe the distribution of the annual radar

rainfall maxima. Regional variability in the GEV location parameter was studied. GEV

parameters based on radar and rain gauge data were compared and turn out to be in

reasonable agreement. Furthermore, radar rainfall depth-duration-frequency (DDF) curves

and their uncertainties were derived and compared with those based on rain gauge data.

Although, uncertainties became large for long durations, the radar data were suitable to

construct DDF curves.

Kent (1968) presented soil conservation services (SCS) technology for estimating runoff

volume and peak discharge from small catchments. In his presentation, a number of

interesting facts were revealed. The SCS Type-I and Type-II design rainfall distributions

were documented by Kent. They were based on analysis of Weather Bureau (now National

5
Weather Service) rainfall atlases. The country was divided into two geographic regions: one

west from the Sierra Nevada and Cascade mountain ranges and the second east of that

geographic divide. The distributions were based on the depth duration curves for a return

interval of 25 years. Although not documented, it appears that the depth-duration curves were

non-dimensionalized so that they could be compared for purposes of fitting a curve. The

distributions were constructed by placing the greatest 30-minute depth near the middle of the

24-hour storm period. The next greatest 30-minute depth was placed subsequent to the

greatest 30-minute depth. The second largest 30-minute depth was placed in the preceding the

greatest 30-minute depth. This pattern of placing ranked 30-minute rainfall depths in

alternating positions was continued to fill the 24-hour rainfall duration.

2.3. DEVELOPMENT OF HYETOGRAPH

Bonta and Rao (1989) compared dimensionless mass hyetographs of north Appalachian

experimental watershed Coshocton, Ohio (NAEW), Illinois, Chicago, and Texas. The work

of Huff (1967) served as a basis for comparison. According to their findings, significant

samples sizes are required to stabilize the curves, relatively few storms were available for

development of third and fourth quartiles for the Illinois and Texas events. The research

concluded that visual differences between the Ohio and Illinois Huff curves depend on the

quartile being investigated, but was not generally statistically significant. Furthermore,

examination of the 10, 50, and 90 percent curves for areas extending from Ohio, to Illinois, to

Texas were essentially the same. The results suggested that the Huff curves were insensitive

to variations in their development, indicating that they may be used over extensive areas of

the country.

Bonta and Rao (1988) compared dimensionless rainfall mass curves developed using four

different procedures: the original Huff curves, simple triangular shape, complex

6
trapezoidal/rectangular shape, and complex triangular shape for Illinois. The latter three

hyetographs represented rainfall as a single major burst of rainfall preceded and followed by

low intensity rainfall periods. The data set used for the study was the 40-year (1947-1987)

from north Appalachian experimental watershed Coshocton, Ohio (NAEW).

Arnaud and Lavabre (1999) developed a stochastic model for generating hourly hyetographs

by using data from observed precipitation records to simulate rainfall patterns. This model

made it possible to study maximum precipitation distributions for normal or exceptional

frequencies over long periods of time. The modeling provides observations (up to 10-year

frequency) of quintiles similar to those observed by directly fitting a law of statistical

distribution onto an observed distribution.

Roso (2006) investigated that the traditional design flood estimation procedure described in

Australian Rainfall & Runoff (AR&R) manual (IEAust, 2001) is burst based method that can

under-predict discharges at a catchment’s outlet, relative to those obtained from an embedded

design storm (EDS) approach. A series of catchments located in Wollongong, ranging in size

from 0.5 km2 to 105 km2, were used to explore these impacts. All catchments were relatively

complex and partly urbanized catchments, with a number of storages and diversions present

in each. This under-prediction was quite significant in some catchments (up to 40 percent),

causing concern as to the ability of the present AR&R burst based procedure to adequately

simulate design discharges from ‘storage sensitive’ catchments. Application of the EDS

procedure in recent studies further suggested that the AR&R burst based procedure could

result in even greater levels of under prediction at internal locations within a catchment. An

investigation into these internal flows was therefore undertaken to quantify and better

understand the differences between AR&R burst based discharges and EDS discharges within

a catchment.

7
Alfieri et al. (2006) was examined five design hyetographs in a set of simulations, based upon

the following steps:

(i) An ideal river basin was defined, characterized by a Beta distribution shaped unit

hydrograph (UH).

(ii) 1000 years of synthetic rainfall were artificially generated.

(iii) A discharge time-series was obtained from the convolution of the rainfall time-series

and the UH, and the reference T-years flood was computed from this series.

(iv) For the same return period T, the parameters of the intensity–duration–frequency (IDF)

curve were estimated from the 1000 years of synthetic rainfall.

(v) Five design hyetographs were determined from the IDF curves and were convolved with

the discrete UH to find the corresponding design hydrographs.

(vi) The hydrograph peaks were compared with the reference T-years flood and the

advantages and drawbacks of each of the five approaches were evaluated.

The research concluded that all design hyetographs produced flood peak estimates that were

consistently biased in most of the climatic and hydrologic conditions considered. In

particular, significant underestimation of the design flood results from the adoption of any

rectangular hyetograph used in the context of the rational formula.

8
 Chapter III
THEORATICAL BACKGROUND

3.1 HYETOGRAPH ESTIMATION METHODS

There are many approaches that can be used for the development of a hyetograph. Many

scientists have worked on the development of hyetograph and come up with their own

theories. In this selection of dissertation, three approaches widely used for hyetograph

development, are discussed:

3.1.1. STORM EVENT ANALYSIS

By analysis of observed storm events, the time sequence of precipitation in typical storms can

be determined. The most dominant work on storm event analysis for hyetograph development

were carried out by Huff (1967), Pilgram and Cordery (1975) and Yen and Chow (1980). A

brief description of research work is given as:

Huff (1967) analyzed that a major portion of the total storm rainfall occurs in a small part of

the total storm time, regardless of storm duration, areal mean rainfall, and the total number of

showers or bursts in the storm period. A criteria for the selection of storms in Illinois was

developed according to which all storms were selected having storm duration 3 to 48 hours

and also having network mean rainfall 0.5 inches or one rain gauge having rainfall more than

one inch. Total 261 storms qualified for the study as per criteria. The storms were classified

into four quartiles. The first and second quartile takes up to two-third of the events, only

fifteen is for the forth and the rest are for the third. The storms were presented in the form of

cumulative percentage of storm time and cumulative percent of storm rainfall as shown in

Figure 3.1. The figure shows that the storms that have heaviest rainfall in first 25% storm

duration called as first quartile storms, second quartile storms which have heaviest rainfall in

25 to 50% storm duration, third quartile represent the storms those have heaviest rainfall in

9
50 to 75% storm duration and fourth quartile storms having heaviest rainfall in last 25%

storm duration.

Fractions Of Storm Rainfall

First Second
Quartile Quartile Third Quar-
tile Fourth
Quartile

Fraction of Storm Duration

Figure 3.1 First-, second-, third-, and fourth-quartile distributions of storm rainfall for
Illinois

Yen and Chow (1980) stated that once the design precipitation depth P and duration T dare

known, the base length and the height of the triangle are determined. Consider a triangle

having base length is Td and the height h, so the total depth of precipitation in the hyetograph

1 2P
is given by P= from whichh= . . Storm advancement coefficient r was defined as
2 h Td Td

the ratio of the time before peak ta to the total duration used to know where the peak of storm

occurs. A suitable value of r was determined by computing the ratio of the peak intensity time

to the storm duration for a series of storms of various durations. The mean of these ratios,

weighted according to the duration of each event, was used for r. The data for locations:

Urbana, Illinois;Boston, Massachusetts;Elizabeth City, New Jersey; and San Luis Obispo;

California were used. The analysis indicated that the triangular hyetograph for most heavy

storms were nearly identical in shape, with only secondary effects from storm duration,

10
measurement inaccuracies, and geographic location. The triangular hyetograph is shown in

Figure 3.2.

i
Rainfall intensity

Figure 3.2 Triangular Hyetograph

Pilgram and Cordery (1975) developed a hyetograph method based on the ranking the time

intervals in a storm by the depth of precipitation occurring in each, and repeating this exercise

for many storms in the region. By averaging each interval and taking percentage of each rank

a typical hyetograph shape was appeared. They derived patterns for the Sydney from the first

10, 25 and 50 ranked storms showed a number of intersecting features. The intensities for the

first 10 storms were less variable than for the first 50 storms, with those for the first 25

storms intermediate but closer to those for the first 50 i.e. the heavier rain tended to be more

uniform. Another trend indicated was that the patterns of longer periods of intense rain were

much more variable than those of the shorter durations. While a constant number of periods

for all durations were preferable for investigation of these trends, the varying numbers and

lengths of periods were selected to suit the criteria outlined earlier, so that the derived

patterns would be the maximum usefulness in practical design.

11
The U.S department of agriculture, Soil Conservation Service (SCS) (1986) developed

synthetic storm hyetographs for use in the United States for the storms of 6 and 24 hours

duration. There are four 24-hour duration storms, called Types I, IA, II and III and they also

give the geographic location within the United States where they should be applied. Types I

and IA are for the pacific maritime climate with wet winters and dry summers. Type III is for

the Gulf of Mexico and Atlantic coastal areas, where tropical storms result in large 24-hour

rainfall amounts. Type-II is for the remainder of the nation. The representative dimensionless

curves for each type are shown in Figure 3.3. The trend from rise to end of each curve is

different. Each curve having heaviest rainfall magnitude at different storm locations and

heaviest rainfall magnitude is also different.

Fraction of 24 hour rainfall

IA

III

II

Time (Hours)

Figure 3.3 Synthetic storm hyetograph by SCS (1986)

3.1.2. IDF RELATIONSHIP

Design hyetographs can also be developed from intensity - duration - frequency relationship.

A simple way, used also as default in some computer programs are the alternating block

12
method, referred to also as a center-loaded storm, as the peak of the hyetograph is placed at

the middle of the storm duration. The Chicago method (Keifer and Chow 1957) and the

alternating block method (Chow et al. 1988) are the representatives of this technique.

Keifer and Chu (1957) developed a synthetic hyetograph of this type for use in sewer system

design for Chicago. The procedure utilizes the relationship of intensity-duration-frequency

(IDF) curve to convert the presentation in to a design hyetograph. The average rainfall

intensity iave was defined as in equation 3.1.

c
iave= (3.1)
T d ͤ +f

Where;

іave = Average Intensity

T d = Total storm duration

c, e and f = constants

The method provides equation 3.2 for calculating peak intensity, and then redistributes the

rainfall before and after the peak with appropriate equations.

c [ ( 1−e ) T d ͤ + f ]
i= (3.2)
( T d ͤ + f )²
Where;

T d is the total storm duration and c, e and f are the constants.

13
 Precipitation intensity(in/hr)

Time (minutes)

Figure 3.4 Keifer and Chu Hyetograph with peak r=0.5

Chow et al. (1988) was introduced a simple way of developing a design hyetograph from an

intensity-duration-frequency curve. The design hyetograph produced by this method specifies

the precipitation depth occurring in n successive time intervals of duration ‘∆t’ over a total

duration (T d = n∆t). After selecting the design return period, the intensity is read from IDF

curve for each of duration’s ∆t, 2∆t, 3∆t…, and the corresponding precipitation depth found

as the product of intensity and duration. By taking differences between successive

precipitation depth values, the amount of precipitation to be added for each additional unit of

time ∆t is found. These increments, or blocks, are reordered into a time sequence with the

maximum intensity occurring at the center of the required duration T d and remaining blocks

arranged in descending order alternately to the right and left of the central block to form the

design hyetograph shown in Figure 3.5.

14
ti

ll

it
o
n
o

y
a

e
s

s
f
r

I
i

Time (hours)

Figure 3.5 Representative Hyetograph with Alternative Block Method

3.1.3. STOCHASTIC APPROCH

A stochastic model is a tool for estimating probability distributions of potential outcomes by

allowing for random variation in one or more inputs over time. The random variation is

usually based on fluctuations observed in historical data for a selected period using standard

time-series techniques. Distributions of potential outcomes are derived from a large number

of simulations (stochastic projections) which reflect the random variation in the input(s). The

most leading work on stochastic approch for hyetograph development were carried out by

Koutsoyiannis and Foufoula (1993), Guzman and Oliver (1993).

Koutsoyiannis and Foufoula (1993) concluded that empirical evidence suggests that statistical

properties of storm rainfall at a location and within a homogeneous catchment have a well-

structured dependence on storm duration. To explain this dependence; a simple scaling model

for rainfall intensity within a storm was hypothesized. It was shown both analytically and

empirically that such a model can explain reasonably well the observed statistical structure in

15
the interior of storms providing thus an efficient parameterization of storms of varying

durations and total depths. The simple scaling model was also consistent with, and provides a

theoretical basis for the mass curves are extensively used in hydrologic design. In contrast,

popular stationary models of rainfall intensity were shown unable to capture the duration

dependent statistical structure of storm depths and were also inconsistent with the concept of

mass curves.

Guzman and Oliver (1993) proposed to describe the time distribution of rainfall within a

rainy event by modeled process, the dimensionless hyetograph H(tj) that represents the

fraction of rainfall accumulated over the time interval (0, tj), where tj is a fraction of the total

duration. In any rainy episode lasting n+1 units of time, the sequence H(t1), …, H(tn) was

described as an ordered sample from the beta distribution. Dimensionless rainfall from hourly

data at three stations in Southern Spain show little dependence on rainfall duration, amount of

precipitation, or period of the year. These results, though not essential to the model structure,

simplify its formulation, reducing the number of parameters. Other features of the historical

records, such as correlations, maximum dimensionless hourly precipitation, or frequency

curves, were reasonably well reproduced by the model.

16
 Chapter IV
METHODOLOGY

4.1 CHARECTERISTICS OF STUDY AREA

Swat valley situated in the Hundukush mountain ranges of Northern KPK is a combination of

three districts, namely Upper Malakand (Tehsil Swat Ranizai), Lower Dir (Tehsil Adenzai)

and Swat District (all six sub-Tehsil, namely Kalam, Bahrain, Kwazakhala, Matta, Mingora,

Kabal, and Barikot) that collectively form River Swat catchment. District Swat makes

80.81%, Lower Dir 11.98% and Malakand 7.21% of the catchment. The study area comprise

of the entire catchment located at 34° - 36° North Latitude and 71° - 73° East Longitude

surrounded by mountains with an elevation ranging from 574-5808 m above MSL (Mean Sea

Level) and extends over a total area of 6144 Km². The total population of Swat is 1.25

million (GoP, 1999). The drainage area of the catchment has a complex geological history of

organic disturbances and erosion and depositional cycle. Average annual rainfall ranges form

800-1270 mm. A minor proportion of water is also diverted from streams and the river for

irrigation at suitable locations. The major crops grown in the area include wheat, rice,

potatoes, fodder and maize. Major portion of rainfall is received during the monsoon season

(June to September). According to meteorological data (Weather station Saidu Sharif) the

minimum and maximum temperature varies in the range of –0.50Cº to 390Cº. The mean

relative humidity varies from a minimum of 40% in April to a maximum of 85% in the month

of July. The overall climate of the area can be classified as sub-tropical in the Southern part

and Mediterranean in the northern parts (Aslam, et al 1991). The soil is mainly sandy loam

type occupying the maximum area with a land slope varying from 0 to 8 %, maximum slope

of some hilly parts of the watershed is up to 22% and the soil depth ranges from 0 to 45 cm

(Rashid, et al 1999). It is observed that the area exhibit high erosion rate, especially in the

17
areas where vegetation covers is less. These areas include lower Swat situated in the area of

Lower Dir and Upper Malakand.

4.2 COLLECTION OF DATA

The Swat region lies in the template zone, the rainfall varies greatly in the region, although

number of rainfall gauging stations are very rare. The scope of present study needs analysis

of temporal distributions of historic rainfall events. Only two rainfall stations Kalam and

Saidu Sharif are available in the study area, out of which Kalam have a recording type

raingauge. Data from a single station is not enough for development of regional hyetograph

and analysis of variations of temporal distribution of rainfall. Therefore, raingauges having

operational and recording type in the adjacent areas were also included for analysis. Thus,

total four raingauge stations were selected in the present study namely; Kalam, Besham Qila,

Daggar, and Mardan. Daily data of Saidu Sharif and Dir was also collected. One day design

rainfall magnitudes of Saidu Sharif and Dir were required for flood estimation. Stream flow

data were required for model calibration and base flow estimation. The Swat River is gauged

at Kalam and Chakdara, mean daily flow data and annual maximum instantaneous flows were

collected. Historic rainfall and stream flow data was collected from surface water hydrology

project (SWHP) and Pakistan Metrological Department (PMD).

The detail description and location of rainfall and stream gauging stations is given in the

Table.4.1 and shown in Figure.4.1. The annual maximum rainfall data for the selected

raingauge station is annexed as Appendix I.

18
 Table 4.1 Inventory of stream gauges and rain gauges

Latitud
Longitude Period of
e Instantanou
S. No. Name of Station Daily Agency
s Peaks
D M D M Data

A. STREAM GAUGING STATION

1990-
1 Sawat River at Kalam 72 35 35 21 SWHP 1990-2009
2005
Sawat River at 1990-
2 72 22 34 45 SWHP 1990-2009
Chakdara 2005
Period Of
B. RAING GAUGES Hourly
Data
1962-
1 Kalam 72 35 35 32 SWHP 1990-2010
2010
1962-
2 Saidu Sharif 72 22 34 45 PMD Nil
2010
1969-
3 Besham Qila 72 53 34 56 SWHP 1990-2010
2010
1963-
4 Daggar 72 29 34 30 SWHP 1990-2010
2010
1963-
5 Mardan 72 3 34 12 SWHP 1990-2010
2010
1983-
6 Dir 71 52 35 12 PMD Nil
2010
PMD
Pakistan Meteorological Department
:
SWH
Surface Water Hydrology Project
:

19
Figure 4.1 Locations of rain gauges and stream gauging stations
20
4.3 FREQUENCY ANALYSIS

Extreme Value Type I (EVI) distribution by using frequency factor was selected for

frequency analysis. Extreme value distributions are widely used in hydrologic analysis. Storm

rainfalls are most commonly modeled by the EVI distribution (Chow. 1953; Tomlinson,

1980).

The study of extreme hydrologic events involves the selection of the largest observations

from sets of rainfall data. The study of storm uses one day annual maximum rainfall recorded

each year at a rain-gauge station out of thousands of recorded values to make up a set of

extreme values that is analyzed statistically.

The magnitude XT of a hydrologic event may be represented as the mean u plus the departure

∆Xᴛ of the variate from the mean.

Xᴛ =u+ ∆ Xᴛ (4.1)

The departure may be taken as equal to the product of the standard deviation σ and a

frequency factor KT; that is∆ Xᴛ =K ᴛ σ . The departure∆ Xᴛ and the frequency factor KT are

functions of the return period and the type of probability distribution to be used in the

analysis. Equation 4.1 may therefore be expressed as:

Xᴛ =u+ Kᴛσ(4.2)

This may be approximated by:

Xᴛ =x͞ + KᴛSy (4.3)

In the event that the variable analyzed is y = logx, then the same method is applied to

the statistics for the logarithms of the data, using relationship:

Yᴛ = y͞ + KᴛSy( 4.4)

21
Frequency analysis begins with the calculation of the statistical parameters required for a

proposed probability distribution by the method of moments from the given data. For a given

return period, the frequency factor can be determined from the K-T relationship for the

proposed distribution, and the magnitude Xᴛ computed by Equation 4.3 or Equation 4.4.

For the EVI distribution, Chow (1953) derived the expression:

Kᴛ =
−√ 6
π [ { ( )}]
0.5772+ ln ln
T
T −1
(4.5)

Where,

Kᴛ = Frequency Factor

T = Return Period

For the study area frequency analysis were carried out with following steps:

(i) One day annual maximum rainfall recorded each year at a raingauge station was

separated out of thousands of recorded values to make up a set of extreme values or

annual series.

(ii) Frequency factor was calculated for (2.33, 5, 10, 25, 50, 100, 500 and 1000 years) return

period ‘T’ by puting the value of return period in Equation 4.5.

(iii) The mean and standard deviation ‘Sy’ were calculated for each annual series.

(iv) The values of required return period were resulted by solving the Equation 4.2.

4.4 PRESENTATION OF RAINFALL DATA

22
 Varieties of approaches are available in the literature for the presentation of rainfall

data (e.g. isohytel maps, isoplluvial maps etc). In the present study intensity duration

frequency curves were developed for the presentation of rainfall data at different return

periods.

4.4.1 INTENSITY-DURATION-FREQUENCY (IDF) CURVE

One of the first steps in many hydrologic design projects, such as in urban drainage design, is

the determination of the rainfall events to be used. The most common approach is to use a

design storm or event that involves a relationship between rainfall intensity, duration, and the

frequency or return period appropriate for the facility and site location. In many cases,

hydrologists have standard intensity duration frequency (IDF) curves available for the site

and do not have to perform this analysis. However, it is worthwhile to understand the

procedure used to develop the relationships. Usually, the information is presented as a graph,

with duration plotted on the horizontal axis, intensity on the vertical axis, and a series of

curves are generated, one for each design return period, as illustrated for Kalam, Besham

Qila, Daggar and Mardan in Figure 4.2 to Figure 4.5. The intensity is the time rate of

precipitation, that is depth per unit time (mm/h or in/h). It can be either the instantaneous

intensity or the average intensity over the duration of the rainfall.

Following steps were followed for the development of IDF curves:

(i) The annual maximum rainfall series of different durations (i.e. 1hr, 2hr …up to 24hr)

were generated separately at all selected stations using the point rainfall data for each

year.

(ii) Frequency analysis were carried out for rainfall series of six selected durations using

extreme value Type-I distribution at different return periods.

23
(iii) Rainfall intensities were calculated by dividing the rainfall magnitude with the time

period of magnitude.

(iv) Curves of different return period were developed by plotting rainfall intensities against

selected durations.

45

40
T=100 Years
35 T= 50 Years
T= 25 Years
30
T= 10 Years
25 T= 5 Years
m
m
n

h
e

T=2.33 Years
(

r
)
t

i
t

20

15

10

0
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (Hours)

Figure 4.2 IDF curves for Kalam derived from recorded data

24
 50
T=100 Years
T= 50 Years
40
T= 25 Years
T= 10 Years
30
T= 5 Years
m
m
n

h
e
s
I

r
)
t

i
t

T=2.33 Years
20

10

0
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (Hours)

Figure 4.3 IDF curves for Besham Qila derived from recorded data

90

80
T=100 Years
70
T= 50 Years
60 T= 25 Years

50 T= 10 Years
m
m

T= 5 Years
n

h
e

s
I

r
)
t

i
t

40
T=2.33 Years
30

20

10

0
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (Hours)

Figure 4.4 IDF curves for Daggar derived from recorded data

25
 100

90

80 T=100 Years

70 T= 50 Years
T= 25 Years
60
T= 10 Years
m
m
n

h
e

s
I

r
)
t

i
t

50
T= 5 Years
40
T=2.33 Years
30

20

10

0
0 2 4 6 8 10 12 14 16 18 20 22 24
Time (Hours)

Figure 4.5 IDF curves for Mardan derived from recorded data

4.5 DEPTH-DURATION-FREQUENCY RELATIONSHIP

Depth duration frequency (DDF) relationship shows the variation of rainfall depth with

change in duration and return period. Knowledge of DDF relationship is a necessary step in

the design of any structures which have to pass or withstand flood events, whether the

structure is a small road culvert or a large dam. DDF relationship was developed in two steps.

In the first step, depth frequency relationship was determined and depth duration relationship

developed in second step. After simulation the results of both steps discussed above, depth

duration frequency relationship was resulted.

Depth-Frequency relationship for the 24 hour rainfalls at each station were established by

using available records. The detail of available records is given in Table.4.1. Following steps

were followed for the development of depth-frequency relationship:

26
(i) Frequency analysis of one day annual maximum rainfall data for Kalam, Mardan,

Besham Qila and Daggar were carried out.

(ii) The rainfall depths at the required return period were calculated for all gauges.

Depth-Duration relationship was developed by following steps:

(i) Rainfall data (1,2,3,6 and 24 hr) were prepared. This data was arranged and analyzed so

that there was one maximum value for all durations in a year.

(ii) These data sets were then analyzed for each of four rainfgauges Kalam, Mardan,

Besham Qial and Daggar.

(iii) The resulted average depth is the representative of particular duration for a raingauge.

Depth-Duration-Frequency relationship was established by following steps:

(i) Frequency analyses were carried out for the rainfall of different durations as calculated

in depth-duration relationship.

(ii) The rainfall depth for five durations (1hr, 2hr, 3hr, 6hr and 24hr) and at seven return

periods (2.33, 5, 25, 50, 100, 500, and 1000) was established.

4.6 WIRASAT ULLAH KHAN METHOD OF HYETOGRAPH DEVELOPMENT

Khan (1980) developed a technique for hyetograph development by using depth duration

frequency relationship of a region. The regional design hyetograph was developed by using

the available hourly data of four raingauges. Following steps were followed for the

development of hyetograph:

(i) Rainfall depths (1, 2,3,6 and 24 hr rainfall records) were prepared from the data. The

annual maximum series were prepared for selected durations.

27
(ii) Design rainfall for seven return periods (2.33, 5, 25, 50, 100, 500, 1000) were computed

with EVI distribution for selected raingauges. Design storm distributions were resulted

for a gauge at seven return periods.

(iii) The average rainfall depths as fractions of average of 24 hr rainfall depth were

computed.

(iv) Design storm distribution was selected at 2.33 years return period of all raingauges,

because the fractions at 2.33 ‘T’ were identical. Make a representative values for

durations of four raingauges by taking simple average of five durations.

(v) A power term ‘n’ was calculated using the relationship ¿ by hit and trial procedure. The

appropriate value of ‘n’ yields the average rainfall depth (fraction of 24 hr rainfall

depth) for the specific duration of rainfall in generalized way. The value of ‘n’ should

vary between 0 to 1.

(vi) The rainfall depth for the required duration was calculated using equation 4.6

t 0.32
Rainfall depth = PTp ( ) (4.6)
24

Where ‘t’ is the duration at which rainfall is required, Tp is return period, P is the storm

magnitude, 24 is storm duration and PTp is the 24 hour design rainfall depths at return period

T years.

4.7 HUFF CURVES

To develop huff curves, data for the 21 years period (1990 to 2010) was used. Huff (1967)

divided storms into four quartiles (1st, 2nd,3rd and 4th). The storms that have heaviest

rainfall in first 25% storm duration were denoted as first quartile storms whereas second

quartile storms were those having heaviest rainfall in 25 to 50% storm duration. Third

28
quartile represent the storms having heaviest rainfall in 50 to 75% storm duration and fourth

quartile storms having heaviest rainfall in last 25% storm duration.

Following steps were followed for the development of Huff curves:

(i) Hourly rainfall data for the four gauges was tabulated.

(ii) The events in which the network mean rainfall exceeded 0.5 inches and one or more

raingauges recorded over 1 inch were selected.

(iii) The storms having duration from 3 to 48 hours were selected for the study. Within the

data period, 389 network storms were qualified for the study.

(iv) Time distributions were expressed as cumulative percentages of storm rainfall and storm

duration to make valid comparisons between storms and to simplify analyses and

presentation of the data.

(v) The quartile was assigned to each storm.

(vi) A standard time distribution was selected for all storms and percentage of each storm

magnitude at selected duration was calculated by interpolation.

(vii) The average of storms distributions having same quartile was the representative storm of

that quartile.

The statistics of storms, the quartile group of storms, and number of storms in rainfall

durations are given in Table.4.2.

Statistics of storms given in Table 4.2 shows that second and third quartile storms are the

representative of 12-24 hours rainfall durations. The percentage of storm duration from 12 to

24 hours is greater in 3rd quartile than 2nd quartile, so third quartile storms are the

representative of the 12-24 hours rainfall events.

29
 Table 4.2 Statistics of selected storms for Huff curves development

Storm First Second Third Fourth

Total
Durations Quartile Quartile Quartile Quartile
Storms
(Hours) storms storms storms storms
3 to 6 21 32 17 13 83

6 to 12 37 45 26 32 140

12 to 24 19 48 40 26 133

24 to 48 4 5 15 9 33

Total 81 130 98 80 389

Storms

4.8 TRIANGULAR HYETOGRAPH

Triangular hyetograph were developed using Yen and Chow (1980) technique. They analyze

9869 storms of following cities: Urbana, Illinois; Boston, Massachusett Elizabeth City, New

Jersey and San Luis Obispo California. The analysis indicated that the triangular hyetographs

for most heavy storms were nearly identical in shape, with only secondary effects from storm

duration, measurement inaccuracies and geographic location.

The input data required for triangular hyetograph development is, storm advancement

coefficient, total storm depth and duration. Storm advancement coefficient (r) is the peak

intensity location within the storm; it can be determined by computing the fraction of the

peak intensity time to the storm duration for a series of storms of various durations. The mean

of these ratios, weighted according to the duration of each event was used for r. Following

steps were followed for the development of triangular hyetograph:

(i) Due to data constraints 389 storms were used in the form of their ratios.

(ii) Total storm time and storm magnitude was taken as 1, while 1 represents the fraction of

100%.

30
(iii) Storm advancement coefficient (r) was determined by computing the fraction of the

peak intensity time to the storm duration for all storms. The mean of these fractions,

weighted according to the duration of each event was used for r.

(iv) The peak intensity h was calculated as

2P
h=
Td

Where;

P = Total depth of precipitation

T d = Total storm duration

(v) The time ‘t b’ to the peak intensity was calculated as:

t ₐ=r T d

(vi) The recession time ‘t’ was calculated as:

t b=T d −t ₐ

4.9 CATCHMENT AREA

The catchment area of Swat River is characterized with steep vegetated slopes, forests and

some towns (Kalam, Mangora, Chakdara etc) are within the catchment area. The outlet point

of river is at the confluence with Panjkora River. The catchment area of Swat river is 6144

km² upto confluence with Panjkora River. In Swat; generally the slopes are cultivated in

terraces which give rise to higher infiltration losses as compared to areas where slopes are not

terraced. The length of Swat River up to confluence with Panjkora River is 206 Km. The

Swat catchment having higher altitude variation is shown in Figure 4.6. The flows of Swat

31
river are recorded at Chakdara and Kalam, the average, minimum and maximum flows at

Chakdara are, 180 cumecs, 4.1 cumecs on 6 Oct 1991 and 6100 cumecs on 29 July 2010.

32
Figure 4.6 Altitude variation of study area

33
4.10 DIGITAL ELEVATION MODEL (DEM)

SRTM 90m Digital Elevation Model (DEM) was utilized to generate the input data of HEC-

HMS model. Stream network of the study area was generated from DEM. Outlet point of the

Sawat River was selected at the confluence of the Panjkora River. The catchment area was

delineated by using DEM. The catchment was sub divided into twenty sub catchments. The

confluence of each catchment is at junction and each junction connected with other junction

with a reach. The utility of DEM further explained below where it required. The sub basins,

junctions and reaches are shown in the Figure 4.7.

Figure 4.7 Map of the selected basin

4.11 FLOOD ESTIMATION METHOD
34
Several procedures were found in the literature for flood estimation but United States soil

conversation services (USSCS) unit hydrograph method is most sophisticated method and

world widely used.

The unit hydrograph developed by the L.K.Sherman in (1937). The unit hydrograph

procedure assumes that discharge at time is proportional to the volume of runoff and that time

factors effecting shape and constant. For flood estimation unit hydrograph approach was

adopted.

4.12 HEC-HMS SOFTWARE

The Hydrologic Modeling System is designed to simulate the precipitation-runoff processes

of watershed systems. This includes large river basin water supply and flood hydrology, and

small urban natural watershed runoff. Hydrographs produced by the program are used

directly or in conjunction with other software for studies of water availability, urban

drainage, flow forecasting, future urbanization impact, reservoir spillway design, flood

damage reduction, flood plain regulation, and system operation.

The physical representation of a watershed is accomplished with a basin model. Hydrologic

elements are connected in a network to simulate runoff processes. Available elements are:

sub basin, reach, junction, reservoir, diversion, source and sink. Computation proceeds from

upstream elements in a downstream direction. The elements are shown in Figure 4.8. The

steps involved in the estimation of the flood are following:

35
 Figure 4.8 Available elements in HEC-HMS software

4.12.1 LOSS METHOD SELECTION

Loss method was required to estimate the excesses rainfall. In HEC-HMS model loss

methods, initial constant, SCS curve number, gridded SCS curve number, exponential, Green

Ampt, one-layer deficit constant method and smith parlange can be used to compute excess

rainfall. The SCS curve no method was selected as a loss method for computations of excess

rainfall.

The SCS procedure consists of selecting a storm and computing the direct runoff by the use

of curves founded on field studies of the amount of measured runoff from numerous soil

cover combinations. The curve number for AMC-II condition was estimated as 71. For major

rainfall events, the AMC-III condition is the critical soil moisture condition. The curve for

AMC-III was estimated as 85, which was used in the analysis. According to USSCS, if five

days antecedent rainfall is less than 36 mm during growing season, then it is Antecedent

Moisture Condition (AMC) - I condition, if it is greater than 53 mm, then it called AMC-III

36
condition, otherwise it is AMC-II condition. The detail of curve no against AMC conditions

is given Table 4.3.

Table 4.3 Curve no of AMC-I and III against AMC-II

CN for CN for AMC CN for CN for AMC

AMC II I III AMC II I III
0 0 0 60 40 78
5 2 13 62 42 79
10 4 22 64 44 81
15 6 30 66 46 82
20 9 37 68 48 84
25 12 43 70 51 85
30 15 50 72 53 86
32 16 52 74 55 88
34 18 54 76 58 89
36 19 56 78 60 90
38 21 58 80 63 91
40 22 60 82 66 92
42 24 62 84 68 93
44 25 64 86 72 94
46 27 66 88 75 95
48 29 68 90 78 96
50 31 70 92 81 97
52 32 71 94 85 98
54 34 73 96 89 99
56 36 75 98 94 99
58 38 76 100 100 100

4.12.2 TRANSFORM METHOD

Seven different methods are included for transforming excess precipitation into surface

runoff. Unit hydrograph method includes the Clark, Synder and SCS techniques. User

specified unit hydrograph or S-graph ordinates can also be used. The modified Clark method,

Mod clark, is a linear quasi-distributed unit hydrograph method that can be used with gridded

meteorological data. An implementation of Kinematic wave method with multiple planes and

channels is also included.

SCS unit hydrograph method was selected as transform method. The parameters required for

transformation are discussed under

37
(i) Lag Time

The time from center of mass of excess rainfall to the hydrograph peak is lag time. Lag time

is also referred to as basin lag. For unit hydrograph method HEC-HMS model required lag

time in minutes.

Lag time = 0.6 Tc minutes

Tc = Time of Concentration

(ii) Time of Concentration

The travel time from the hydraulically remotest point in a watershed to the outlet. This is also

defined as the time from the end of rainfall excess to the recession curve inflection point as

illustrated on the accompanying hydrograph.

Several methods have been found in the literature. But the Kirpich formula is mostly used for

computation of time of concentration which is as follows:

0. 77
L
T c=
7700xS0 .385

Where, Tc represents Time of concentration (hours), L is Length of the longest stream

(meters) and S is Vertical Fall in length L (meters).

In the present study Swat region was divided into twenty sub-catchments. The time of

concentration and lag time for each sub catchment was computed separately. The catchment

area, time of concentration and lag time for each sub catchment is shown in Table 4.4 and the

schematic layout of HEC-HMS model are shown in Figure 4.9. The data for computation of

time of concentration and delineation of sub catchment areas were prepared by using SRTM

90m DEM (longest stream length, Elevation difference after every 100 meter interval) shown

in Figure 4.9

38
Table 4.4 Catchment parameters of the study area

Catchment
Sub Basin Lag Time
Sr No Area
Name (minutes)
(Km²)
1 S-1 1226 210
2 S-2 789 188
3 S-3 206 88
4 S-4 184 45
5 S-5 273 98
6 S-6 259 65
7 S-7 218 54
8 S-8 196 71
9 S-9 440 143
10 S-10 258 120
11 S-11 93 43
12 S-12 275 102
13 S-13 360 155
14 S-14 198 131
15 S-15 274 104
16 S-16 227 131
17 S-17 277 118
18 S-18 123 223
19 S-19 99 118
20 S-20 169 135

39
 N

SB=Sub basin, R=Reach,

J=Junction

Figure 4.9 Schematic of study area in HEC-HMS model

40
4.12.3 BASE FLOW

Base flow can be defined as the flow which results from the snow melt or when channel bed

meets with ground water table or intrusion of ground water in channel. Five different

methods are available in HEC-HMS model for base flow modeling. The recession method

gives an exponentially decreasing base flow from a single event or multiple sequential events.

The constant monthly method can work well for continuous simulation. The linear reservoir

method conserves mass by routing infiltrated precipitation to the channel. The nonlinear

Boussinesq method provides a response similar to the recession method but the parameters

can be estimated from measurable qualities of the catchment.

Constant monthly method was adopted for base flow simulation and a constant value was

required for the month in which flood hydrograph was modeled. By analyzing the data,

maximum base flows were found in the month of June and July due to snow melt in sub

basins SB-1 and SB-2. Following procedure was adopted to calculate the base flow for the

month of July, as all historic flood creating events occurred during the month of July.

Daily flow data and annual maximum peaks of Kalam and Chakdara on the Swat River is

available from 1990 to 2005. Kalam station is at upstream whereas Chakdara gauging station

is located at downstream with in the catchment of the Swat River. The catchment area at

Kalam is 2020 km2 and at Chakdara the catchment area is 5776 Km2. The flow data of these

stream gauging stations was analyzed for the month of July 1997 in which total rainfall

during the month was only 10.5 mm at Kalam gauging station and consequently all the flow

recorded at the site was primarily the base flow.

The maximum daily flow recorded at Kalam from 1990 to 2009 was 505 m 3/s, while mean

average flow was 386 m3/s. At Chakdara, the maximum daily flow and average flow during

41
July 1997 were 545 m3/s and 424 m3/s respectively. Hydrographs of Kalam and Chakdara for

July 1997 are shown in Figure.4.10.

The pattern of the flows at the two sites shows resemblance. It may be noted that since there

has been negligible rainfall in the upstream catchment (10.5 mm rainfall during the whole

month), so it can be concluded that maximum portion of the flow at Kalam gauging station is

base flow. The catchment area above Kalam gets contribution from snow melt, while

downstream of Kalam, the contribution of snow melt is negligible. Since the increase in flow

between Kalam and Chakdara is negligible (26 m3/s), while the corresponding increase in

catchment area is substantial (3756 km2), it can be concluded that there is negligible

contribution of base flow downstream of Kalam gauging station. For calibrations of model,

base flow was adopted by analyzing daily flow data of selected storm.

600
Kalam
550
Chakdara
500
Flow (cumecs)

450

400

350

300

250

200
0 5 10 15 20 25 30
Time (days)

Figure 4.10 Flow hydrographs for the wettest month

4.

42
4.12.4 REACH ROUTING METHOD

The basin elements are connected with reaches to each other in HEC-HMS software. To

model the attenuation of design floods reach routing was required. Eleven reaches were

included in Swat River model.

Total of six hydrologic routing methods are included in HEC-HMS software. Routing with no

attenuation can be modeled with the lag method. The traditional Muskingum method is

included along with the straddle stagger method for simple approximation of attenuation. The

modified puls method can be used to model a reach as a series of cascading, level pools with

user-specified storage discharge relationship. Channels with trapezoidal, rectangular,

triangular or circular cross sections can be modeled with the kinematic wave or Muskingum-

cunge methods. Channels with overbank areas can be modeled with the Muskingum-cunge

method and an 8-point cross section. Additionally, channel losses can also be included in the

routing. The constant loss method can be added to any routing method while the percolation

method can be used only with the modified plus or Muskingum-cunge methods.

Muskingum-Cunge routing method was selected for reach routing in the present study . The

data required for Muskingum-Cunge routing was the reach length, slope, manning’s ‘n’ and

cross section for each reach. The geometric data was generated from DEM and manning’s ‘n’

was selected to be 0.033 for main channel and 0.04 for banks. The manning’s ‘n’ was

selected from reviewing the pictures of the river bed. The length, slope and invert level for

each reach’s representative cross section is given in Table 4.5.

43
 Table 4.5 Parameters of reach routing

Cross
Reach Length Slope
Sr.No Section
Name (m) (m/m)
Invert (m)
1 R-1 18900 0.016 1668
2 R-2 16200 0.014 1493
3 R-3 8600 0.01 1336
4 R-4 10100 0.009 1254
5 R-5 27400 0.008 1126
6 R-6 20500 0.0065 965
7 R-7 6100 0.006 831
8 R-8 13400 0.005 804
9 R-9 20400 0.004 749
10 R-10 19100 0.0033 601
11 R-11 10200 0.0005 566

4.12.5 METEOROLOGIC MODEL

Four different methods for analyzing historical precipitation are included in HEC-HMS. The

user specified hyetograph method is for precipitation data analyzed outside the program. The

gage weight method uses an unlimited number of recording and non-recording gages. The

thiessen technique is on possibility for determining the weights. The inverse distance method

addresses dynamic data problems. An unlimited number of recording and non-recording

gages can be used to automatically proceed when missing data is encounter. The gridded

precipitation method uses radar rainfall data.

Four different methods for producing synthetic precipitation are included. The frequency

storm method uses statistical data to produce balanced storms with a specified exceedance

probability. The standard project storm method implements the regulations for precipitation

when estimating the standard project flood. The SCS hypothetical storm method implements

the primary precipitation distributions for design analysis using SCS criteria. The user-

44
specified hyetograph method can be used with a synthetic hyetograph resulting from analysis

outside the program.

The selection of appropriate metrological model by keeping in view the required metrological

event is essential from methods given in the HEC-HMS. So by studying the manual, an

understanding was developed for design storms, frequency storm method used and for

calibrations, specified hyetograph method used. The SCS storm was also used for flood

computation and compared with other.

4.12.6 HYDROLOGIC SIMULATION

The time span of a simulation is controlled by control specifications. Control specifications

include a starting date and time, ending date and time, and a time interval. A simulation run

was created by computing a basin model, meteorological model, and control specifications.

Run options include a precipitation or flow ratio, capability to save all basin state information

at a point in time, and ability to begin a simulation run from previously saved state

information.

Simulation results can be viewed from basin map. Global and element summary tables

include information on peak flow and total volume, a time-series table and graph are

available for elements. Results from multiple elements and multiple simulation runs can also

be viewed. All graphs and tables can be printed

4.13 MODEL CALIBRATION

HEC-HMS model was calibrated with Chakdara stream gauging station. The peak discharge

on 29 June 2007 was 874.86 cumecs at Chakdara. The curve no 51 was taken at AMC I

condition, adaptation of curve no against AMC condition discussed in the previous section

4.11.1. The specified hyetograph method of metrological model was used. In the model

45
component time series data, create all rain gauges used in study. Incorporate hourly rainfall

data of 28 and 29 June 2007 in all rain gauges given in Table 4.7. Now in the metrological

model component assign rainfall to each sub basin, by making criteria, which rainfall station

nearest to the sub basin. The detail about the rainfall assigned to sub basins is given in Table

4.6.

Peak discharge at Kalam stream gauging station was 319.14 cumecs and only 28 mm rainfall

recorded at Kalam rainfall station in two day. The time distribution of 28mm rainfall at

Kalam gauging stations was not able to produce runoff (model results). So 319.14 cumecs

discharge at Kalam was taking as constant base flow. Saidu Sharif is non-recording rain

gauge but within the catchment. The rainfall magnitudes of Saidu Sharif are more reliable

than other stations. Hourly distribution of recorded rainfall at Saidu Sharif on 29 June 2007

was generated by analyzing the storm pattern at Daggar, Besham Qila and Mardan.

Table 4.6 Representative rain gauge for sub basins

Sr.No Sub Basin Rainfall Sr.No Sub Basin Rainfall

Name Station Name Station
1 S-1 Kalam 11 S-11 Saidu Sharif
2 S-2 Kalam 12 S-12 Saidu Sharif
3 S-3 Kalam 13 S-13 Saidu Sharif
4 S-4 Kalam 14 S-14 Saidu Sharif
5 S-5 Kalam 15 S-15 Saidu Sharif
6 S-6 Kalam 16 S-16 Saidu Sharif
7 S-7 Besham Qila 17 S-17 Saidu Sharif
8 S-8 Besham Qila 18 S-18 Mardan
9 S-9 Besham Qila 19 S-19 Mardan
10 S-10 Besham Qila 20 S-20 Mardan

46
 Table 4.7 Actual storm distribution for model calibration ‘28 to 29 June 2007’

Saidu Besham
Mardan Kalam Daggar
Date and Time Sharif Qila
(mm) (mm) (mm) (mm) (mm)
28Jun2007, 01:00 8.0 0.0 1.3 5.1 2.5
28Jun2007, 02:00 7.0 0.0 0.0 1.3 12.7
28Jun2007, 03:00 5.0 0.0 0.0 8.9 6.4
28Jun2007, 04:00 3.0 0.0 0.0 14 3.8
28Jun2007, 05:00 2.0 25.4 0.0 10.2 6.4
28Jun2007, 06:00 5.0 38.1 0.0 10.2 6.4
28Jun2007, 07:00 6.0 12.7 0.0 8.9 0.0
28Jun2007, 08:00 8.0 1.3 0.0 2.5 0.0
28Jun2007, 09:00 0.0 0.3 0.0 0.5 0.0
28Jun2007, 10:00 0.0 0.0 0.0 4.6 3.8
28Jun2007, 11:00 0.0 0.0 0.0 0.0 0.0
28Jun2007, 12:00 0.0 0.0 0.0 0.0 0.0
28Jun2007, 13:00 0.0 0.0 0.0 0.0 0.0
28Jun2007, 14:00 0.0 0.0 0.0 1.3 0.0
28Jun2007, 15:00 0.0 0.0 1.3 3.8 1.3
28Jun2007, 16:00 0.0 0.0 0.0 6.4 2.5
28Jun2007, 17:00 3.0 0.0 0.0 3.8 1.3
28Jun2007, 18:00 5.0 0.0 0.0 8.9 5.1
28Jun2007, 19:00 6.0 0.0 0.0 3.8 2.5
28Jun2007, 20:00 8.9 0.0 0.0 3.8 2.5
28Jun2007, 21:00 4 2.5 0.8 6.4 1.3
28Jun2007, 22:00 1.3 0.0 0.0 3.8 6.4
28Jun2007, 23:00 3.8 0.0 0.5 1.3 3.8
29Jun2007, 00:00 6.4 0.0 0.0 3 3.8
29Jun2007, 01:00 3.8 0.0 0.0 0.0 1.3
29Jun2007, 02:00 8 0.0 0.0 0.0 6.4
29Jun2007, 03:00 3.8 0.0 0.0 0.0 0.0
29Jun2007, 04:00 3.8 0.0 0.0 0.0 1.3
29Jun2007, 05:00 6.4 0.0 0.0 1.3 0.0
29Jun2007, 06:00 3.8 1.3 0.0 0.0 1.3
29Jun2007, 07:00 1.3 0.0 0.0 0.5 0.0
29Jun2007, 08:00 0.0 0.0 0.0 4.6 5.1
29Jun2007, 09:00 0.0 1.3 0.0 3 0.0
29Jun2007, 10:00 0.0 0.0 0.0 2.5 0.0
29Jun2007, 11:00 0.0 0.0 0.0 0.0 0.0
29Jun2007, 12:00 0.0 0.0 0.0 0.0 0.0
29Jun2007, 13:00 0.0 2.8 0.0 0.0 0.0

47
4.14 DESIGN RAINFALL

The rain gauge density was not as per specifications of WMO (World Metrological

Organization) within the study area, so thessien polygon method was used to estimate the

area of influence of each rain gauge station and to calculate the average design rainfall for the

catchment. ARC-Hydro is an extension of Arc map, was used to draw the thessien Polygons.

Following steps were followed for the development of thessien polygons:

(i) For delineation of thessien polygones in Arc map, spatial referenced points of

raingauges locations were required. The points were marked by preparing the shape file

of points.

(ii) Open toolbox, search for thessien polygons, a popup window appears by clicking the

thessien polygon option.

(iii) Enter the point file and click on ok, the arc map software draw polygones or area of

influence of raingauges.

(iv) The area of influence of each raingauge within the catchment was calculated.

Thessien polygons are shown in Figure 4.11. Design rainfall was calculated at different return

periods and given in Table 4.8.

Table 4.8 1-day design rainfall for swat basin by Thessien polygon Method

Sr Gauge Are 1-Day Rainfall for different Return Periods (mm)

a
No Name 2.33 5 10 25 50 100 500 100 1000
Km² 0 0
1 Besham 177 41 66 86 112 131 150 194 213 275
Qila
2 Mardan 237 91 115 136 161 180 199 242 261 323
3 Kalam 315 59 79 96 116 132 147 183 198 248
5
4 Dir 70 86 112 134 161 181 201 247 268 334
5 Saidu Sharif 250 70 90 106 127 142 157 191 206 256
8
Total 614 64 85 102 123 139 154 189 205 256
7

48
 Figure 4.11 Thessien polygons of study area

4.15 FLOOD ESTIMATION

Floods were estimated by using HEC-HMS model. The 100 year frequency storm was used

in metrological model. The temporal distribution was generated by using hyetographs

developed during the study and distribution built in the HEC-HMS model. The floods were

computed with three developed (Huff, triangular,Khan) rainfall temporal distributions and

also with four distributions built in the software (Type-1, Type-2, Type-3 and Type 1-A).

Only magnitude of rainfall is required, when flood is estimated with SCS distributions in

HEC-HMS. The observed average base flow 424 cumecs in the driest month July 1997 was

49
used as a constant base flow. The distribution of rainfall for three developed hyetographs

against durations is given in Table.4.9 ans shown in Figure 4.12.

Table 4.9 Rainfall distribution (mm) used in HEC-HMS model in addition to SCS type
distribution

Time Khan Huff Triangular

(Hours)
1 56 12 20
2 69 23 40
3 79 34 54
6 99 63 91
12 123 108 135
24 154 154 154

160

140

120

100
Rainfall(mm)

80

60 Khan

40 Huff

20 Triangular

0
0 4 8 12 16 20 24

Time (hours)

Figure 4.12 Mass curves of rainfall distribution (mm) used in HEC-HMS model in
addition to SCS type distribution

50
 Chapter V
RESULTS AND DISCUSSION

5.1. DEVELOPMENT OF DEPTH DURATION FREQUENCY CURVES

DDF relationship for the study area has been developed in three steps.

5.1.1. DEPTH-FREQUENCY RELATIONSHIP ( Step I )

Frequency analysis of annual maximum daily rainfall at four selected stations is carried out

using Extream Value Type-I distribution. The rainfall depth at different return periods is

calculated separately at all selected stations, and tabulated in Table 5.1. The average rainfall

depth is then calculated for the selected return periods (i.e. one rainfall depth is representative

of study area for each return period). The results of frequency analysis at the selected stations

are graphically presented in Figures 5.1 to 5.4. The depth-frequency relationship is given in

Table.5.1.

Table 5.1 Depth-Frequency relationships at selected raingauge station

Return Period 2.33 5 10 25 50 100 500 1000

Sr no.
Station Rainfall (mm)

1 Kalam 59 79 96 117 132 147 183 198

2 Daggar 91 115 135 161 180 199 243 261

3 Besham Qila 41 66 86 112 131 150 194 213

4 Mardan 91 115 136 161 180 199 242 261

Average 73 96 116 140 158 176 217 234

The values of Kalam and Besham Qila showed great variation from Mardan and Daggar in

Table 5.1. By reviewing the annual series, it was observed that rainfall magnitudes of

Besham Qila and Kalam are lesser than Daggar and Mardan, the values are given in

Appendix I.

51
 300

250
Fitted Line
Observations
R² 0.94
200

150
Rainfall(mm)

100

10000-Year
1000-Year
500-Year
100-Year
10-Year

25-Year

50-Year
5-Year

50

0
-2 -1 0 1 2 3 4 5 6 7
Gumbel's Constant 'K'

Figure 5.1 Frequency analysis of 1-day annual maximum rainfall at Kalam

Fitted Line
300
Observations
R² 0.98
250

200
Rainfall ( mm)

150

100
10000-Year
1000-Year
500-Year
100-Year
10-Year

25-Year

50-Year
5-Year

50

0
-2 -1 0 1 2 3 4 5 6 7
Gumbel's Constant 'K'

Figure 5.2 Frequency analysis of 1-day annual maximum rainfall at Daggar

52
 350
Fitted Line
300 Observations
R² 0.98

250

200
Rainfall ( mm)

150

100

10000-Year
1000-Year
500-Year
100-Year
10-Year

25-Year

50-Year
5-Year

50

0
-2 -1 0 1 2 3 4 5 6 7
Gumbel's Constant 'K'

Figure 5.3 Frequency analysis of 1-day annual maximum rainfall at Mardan

300

Fitted Line
Observations
250
R² 0.93

200

150
Rainfall (mm)

100
10000-Year
1000-Year
500-Year
100-Year
10-Year

25-Year

50-Year
5-Year

50

0
-2 -1 0 1 2 3 4 5 6 7
Gumbel's Constant 'K'

Figure 5.4 Frequency analysis of 1-day annual maximum rainfall at Besham Qila
53
5.1.2. DEPTH-DURATION RELATIONSHIP ( Step II )

The relationship between depth of rainfall and rainfall duration is established as:

(i) Rainfall data for 21 years (1990 to 2010) at the selected stations is arranged in

chronological order.

(ii) The depth of rainfall at each station is averaged for the duration of 1,2,3,6 and 24 hrs

duration respectively.

(iii) The maximum rainfall depth for all the selected duration at different station is extracted

and presented in Table 5.2.

Table 5.2 Depth duration relationships at selected raingauge Station

Time
1 2 3 6 24
Sr (hours)
no.
Station Rainfall (mm)

1 Kalam 11 16 20 30 67

2 Daggar 26 39 50 67 99
Besham
3 15 20 25 36 74
Qila
4 Mardan 35 50 58 67 87

Figure 5.5 shows the mass curves (depth-duration relationship) of the selected stations. The

representative rainfall depths of Kalam and Besham Qila are showing great difference from

Mardan and Daggar in Table 5.2. The rainfall magnitudes at 6 hour duration for Mardan and

Daggar were same but at 1, 2, 3, and 24 hours, difference of 9,11,8 and 12mm respectively

were found.

54
 100

80

60
Depth (mm)

Daggar
40
Mardan
Besham
20
Kalam

0
0 4 8 12 16 20 24

Figure 5.5 Mass curves of depth duration relationship

5.1.3. DEPTH-DURATION-FREQUENCY RELATIONSHIP ( Step III )

The depth duration frequency (DDF) relationship is developed by summing the depth

frequency relationship and depth duration relationship as discussed earlier in sections 5.1.1

and 5.1.2. Frequency analyses were carried out for the rainfall of durations extracted in depth

duration relationship. The rainfall depth is presented at selected seven return periods (i.e.

2.33, 5, 10, 25, 50, 100 and 1000 years) and for five durations (i.e. 1,2,3,6 and 24 hours) at

the selected stations respectively. The results of DDF relationship for Kalam, Daggar,

Mardan and Besham Qila are presented in Tables 5.3 to 5.6, respectively. Figures 5.6 to 5.9,

shows the DDF curves developed at selected stations.

55
 Table 5.3 Depth duration frequency relationships for Kalam

Rainfall
Duration (hours) 1 2 3 6 24
Return Period
(years) Rainfall (mm)

2.33 12 17 22 31 59
5 18 25 30 42 80
10 24 32 37 51 96
25 30 40 46 62 117
50 35 46 52 70 132
100 40 52 58 78 148
1000 56 72 80 104 198

200

160 T=1000 years

T=100 years
T=50 yeras
Depth (mm)

120
T=25 years
T= 10 years
80
T=5 years
T=2.33 years
40

0
1 4 7 10 13 16 19 22
Time (hours)

Figure 5.6 Mass curves of depth duration frequency relationship for Kalam

The representative rainfall values of different durations at Kalam (Table 5.6) show that 28%

to 36% of total storm magnitude occurred in first two hours. The resulted seven curves are

showing uniform slope with minor fluctuations. DDF curves (Figure 5.6) show the maximum

rise between 1 to 2 hours duration. The intermittent distance of curves is almost same except

1000 years curve.

56
 Table 5.4 Depth duration frequency relationships for Daggar

Rainfall
Duration (hours) 1 2 3 6 24
Return Period
(years) Rainfall Duration (hours)

2.33 27 40 51 68 91
5 40 59 72 92 116
10 51 74 89 111 135
25 65 93 111 136 160
50 75 107 127 154 179
100 86 121 143 172 197
1000 119 167 196 232 257

300

250
T=1000 years
200 T=100 years
T=50 years
Depth (mm)

150 T=25 years

T=10 years
100 T=5 years
T= 2.33 years
50

0
1 4 7 10 13 16 19 22
Time (hours)

Figure 5.7 Mass curves of depth duration frequency relationship for Daggar

Table 5.7 shows that 44% to 65% of total storm magnitude at Daggar occurred in first two

hours and 29% to 46% in first hour. The results in table show that with increase in return

period, the percentage of storm magnitude uniformly increases in first and second hour. The

rise of curves in first 3 hours is very sharp and after 3 hours slope of curves became mild.

57
 Table 5.5 Depth duration frequency relationships for Mardan

Rainfall
Duration (hours) 1 2 3 6 24
Return Period
(years) Rainfall Duration (hours)

2.33 36 51 59 69 90
5 50 71 80 94 115
10 61 86 98 115 136
25 76 106 120 142 161
50 87 121 137 161 180
100 98 135 153 181 199
1000 133 183 207 245 261

300

250
T=1000 years
T=100 years
200
T=50 years
T=25 years
Depth (mm)

150
T=10 years
100 T=5 years
T= 2.33 years
50

0
1 4 7 10 13 16 19 22
Time (hours)

Figure 5.8 Mass curves of depth duration frequency relationship for Mardan

Figure 5.8 illustrates that rainfall intensity is greater in the beginning of the storms for all

return periods. Almost 50% of total 24 hour rainfall storm occurred in first two hours. Table

5.5 shows that with increase in return period, the percentage of storm magnitude increases

uniformly in first and second hour. The rise of curves in first 3 hours is very sharp and after 3

hours slope of curves became mild.

58
 Table 5.6 Depth duration frequency relationships for Besham Qila

Rainfall
Duration (hours) 1 2 3 6 24
Return Period
(years) Rainfall Duration (hours)

2.33 15 21 26 36 66
5 24 31 36 46 80
10 31 39 45 53 91
25 40 49 56 63 104
50 47 56 64 75 115
100 53 64 72 82 124
1000 75 88 98 115 158

180

160

140 T=1000 years

T=100 years
120
T=50 years
100
Depth (mm)

T=25 years
80
T=10 years
60
T=5 years
40
T= 2.33 years
20

0
1 4 7 10 13 16 19 22

Figure 5.9 Mass curves of depth duration frequency relationship for Besham Qila

The result of Besham Qila shows that 31% to 56% of total storm magnitude occurred in first

two hours. The resulted seven DDF curves showed uniform slop with minor fluctuations.

DDF curves (Figure 5.9) show the maximum rise between 1 to 2 hours duration.

59
5.2. DEVELOPMENT OF DESIGN HYETOGRAPHS

The hyetograph can be developed by using three approaches; intensity duration frequency

(IDF) curves, storm event analysis and stochastic approach, as describe in detail in Chapter

III of this dissertation. However, in the present study, only one approach is discussed in detail

that is the analysis of observed storms. This approach is adopted because of earlier research

work (Khan, 1980) on the development of design hyetograph in Balochistan province of

Pakistan.

5.2.1. WARASAT ULLAH KHAN METHOD

The hyetograph at the selected raingauge stations of Kalam, Daggar, Besham Qila and

Mardan were developed using the methodology derived by Khan (1980). The results are

given in Table 5.7 in term of fractions of 24 hour rainfall depths. The value of ‘n’ was

approximated as 0.32.

Table 5.7 Summary of results for Khan’s Method

Rainfall depths as Fractions of 24 hours Depth
Station
Rainfall duration (hours)
1 2 3 6 24
Kalam 0.29 0.36 0.40 0.54 1.00
Daggar 0.46 0.60 0.71 0.86 1.00
Besham Qila 0.38 0.43 0.49 0.55 1.00
Mardan 0.53 0.68 0.68 0.90 1.00
Average 0.42 0.52 0.57 0.71 1.00
Excluding Daggar 0.34 0.39 0.44 0.54 1.00
& Mardan
Power term 0.36 0.45 0.51 0.64 ¿( Time1.00 ) ᴺ
‘n’=0.32 Total Time
The table shows that rainfall fractions at Daggar and Mardan are higher than Kalam and

Besham Qila. Due to over estimation of rainfall fractions at Daggar and Mardan their values

were neglected for calculation of power term ‘n’. The value of ‘n’ for the study area was

60
calculated as 0.32. The rainfall depth was calculated for different durations using Equation

4.4.

Alternative block method approach was used for presentation of regional hyetograph. The

rainfall intensity for each time increment was computed by dividing cumulative rainfall

fractions with respective time interval. These increments, or blocks, are recorded into a time

sequence with the maximum intensity occurring at the center of the required duration and

remaining blocks arranged in descending order alternatively to the right and left of the center

block to form the design hyetograph. The regional hyetograph is shown in Figure 5.10 and

ratios of intensity at each time step is shown in Table 5.8.

Table 5.8 Regional rainfall distribution for Swat region with Khan’s Method

Time Fractions of Time Fractions of

(fractions of Rainfall (fractions of Rainfall
24 hours) intensity 24 hours) intensity
0.04 0.04 0.54 0.23
0.08 0.05 0.58 0.14
0.13 0.05 0.63 0.11
0.17 0.05 0.67 0.09
0.21 0.06 0.71 0.08
0.25 0.06 0.75 0.07
0.29 0.07 0.79 0.06
0.33 0.08 0.83 0.05
0.38 0.10 0.88 0.05
0.42 0.12 0.92 0.05
0.46 0.17 0.96 0.04
0.50 0.36 1.00 0.04

61
it
ll
ti
n

n
a

y
r
c

e
s

s
f

t
i

Fractions of time

Figure 5.10 Regional hyetograph for Swat region with Khan’s Method

5.2.2. HUFF CURVES

The curves were plotted by taking fraction of cumulative percentage of rainfall duration on

horizontal axis and fraction of cumulative percentage of rainfall magnitudes on vertical axis.

The percentage fractions were calculated by dividing the storm percentage in any duration

with 100. The four quartiles are shown in Figure 5.11 and expressed as fraction of

percentages of total storm duration and rainfall in Table 5.9.

The curves in Figure 5.9 showing that in first quartile during 0 to 25% storm duration;

rainfall magnitude is 41% of total magnitude. Second quartile is showing 38% of storm

magnitude in 25% to 50% of storm duration. Third quartile is showing 35% of storm

magnitude in 50% to 75% of storm duration, fourth quartile is showing 40% of storm

magnitude in 75% to 100% of storm duration.

62
 1.00

0.80

0.60
F

o
n
o

n
a
c

a
s
r

f
r

f
t
i

l
l
0.40
First Quartile

Second Quartile
0.20 Third Quartile

Fourth Quartile

0.00
0.00 0.20 0.40 0.60 0.80 1.00
Fractions of Time

Figure 5.11 Mass curves for Huff curve

Table 5.9 Fractions for time distribution of Huff curves

Cumulative Cumulative fractions of storm magnitude

fractions of First Second Third Fourth
time
0 Quartile
0 Quartile
0 Quartile
0 Quartile
0
0.05 0.08 0.03 0.04 0.03
0.10 0.16 0.07 0.08 0.07
0.15 0.25 0.11 0.12 0.10
0.20 0.34 0.15 0.16 0.14
0.25 0.41 0.22 0.20 0.18
0.30 0.47 0.29 0.25 0.22
0.35 0.52 0.37 0.29 0.25
0.40 0.56 0.45 0.34 0.29
0.45 0.60 0.53 0.39 0.33
0.50 0.64 0.60 0.45 0.37
0.55 0.69 0.66 0.52 0.41
0.60 0.73 0.71 0.59 0.45
0.65 0.77 0.75 0.66 0.50
0.70 0.80 0.79 0.74 0.55
0.75 0.84 0.83 0.80 0.60
0.80 0.87 0.87 0.86 0.67
0.85 0.91 0.90 0.90 0.75
0.90 0.94 0.94 0.94 0.82
0.95 0.97 0.97 0.97 0.91
1.00 1.00 1.00 1.00 1.00

63
5.2.3. TRIANGULAR HYETOGRAPH

Storm advancement coefficient ‘r’ was determined from a series of storms of various

durations by computing mean of the ratio of peak intensity time to the storm duration. The

resulted value of r is 0.49. The resulting triangular hyetograph for Swat region was plotted in

Figure 5.12 by taking center of storm at 49% storm time. Values of precipitation intensity at

regular intervals was calculated and converted to fraction of precipitation depth given in

Table 5.10. For comparisons the intensity fractions were converted to the depth by

multiplying h=2 with durations and arrange them in ascending order.

2.50

ta=0.51 tb=0.49
2.00

1.50
an

hr
Ri

ns
In
fa

m
m
te
it
ll

y
(

)
/

1.00 h=2

0.50

0.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Time in fractions

Figure 5.12 Representative Triangular hyetograph for Swat region

64
 Table 5.10 Fractions for cumulative rainfall depths and intensity of resulted triangular
hyetograph
Time Intensity Depth Time Intensity Depth
fractions (mm/hr) fractions fractions (mm/hr) fractions
0.05 0.20 0.01 0.55 1.8 0.65
0.10 0.4 0.01 0.60 1.6 0.74
0.15 0.6 0.09 0.65 1.4 0.82
0.20 0.8 0.16 0.70 1.2 0.83
0.25 1.0 0.19 0.75 1.0 0.89
0.30 1.2 0.26 0.80 0.8 0.94
0.35 1.4 0.35 0.85 0.6 0.97
0.40 1.6 0.37 0.90 0.4 0.98
0.45 1.8 0.50 0.95 0.2 0.99
0.50 2.0 0.63 1.00 0.0 1.00

5.3. COMPARISION BETWEEN HYETOGRAPHS

This study was aimed to find out that which of selected methods can best represent the time

distribution of storms in Swat region. A comparison between the selected

methods/approaches of hyetograph development i.e. DDF relationship, curves generated by

the quartile system of Huff (1967) and the triangular hyetograph of Yen and Chow (1980)

and prevailing curves in the HEC-HMS software, was made. For this comparison cumulative

percentages of storm rainfall and cumulative percentages of storm durations are used after

arranging them in the same sequence by interpolation as given in Table 5.11. The results

presented in Table 5.11 and graphically presented in Figure 5.13.

65
 1.00

2010 storm
0.80
Khan

Triangular
0.60
Rainfall in fractions

Huff

SCS Type-1
0.40
SCS Type-2

SCS Type-3
0.20
SCS Type-1A

0.00
0.00 0.20 0.40 0.60 0.80 1.00
Time in fractions

Figure 5.13 Mass curves of hyetographs developed from various methods and prevailing
curves

The plot of hyetographs (Figure 5.13) shows a significant difference between the time

distributions of rainfall. In design storms, the location of the heaviest rainfall is not important,

it could be at any position in storm but the magnitude of heaviest rainfall is important. The

magnitude of heaviest rainfall could be adjusted in design storms after initial abstractions for

computation of design flood. SCS mass curves having minor fluctuations, when arrange them

in the same sequence. SCS Type-1, SCS Type-1A and triangular hyetograph have similar rise

in first five percent of time but they show significant difference after first five percent. SCS

Type-2 and type-3 distributions show respectively 46% and 38% rainfall in first 5% of time.

The time distribution of Triangular hyetograph and 2010 storm event are identical;

particularly the magnitude of rainfall in peak rainfall duration is same. The triangular

hyetograph is seems to be best approach for Swat region.

66
 Table 5.12 Time distributions of Khan, Huff, Triangular,SCS Type-1,2,3,1A and 2010
storm

Depth in Fractions
Time
Fractions 2010
Type-1 Type-1A Type-2 Type-3 Khan Huff Triangular
Storm

0.00 0.000 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.05 0.176 0.16 0.46 0.38 0.13 0.34 0.08 0.13

0.10 0.204 0.19 0.48 0.42 0.27 0.47 0.15 0.25

0.15 0.233 0.23 0.50 0.45 0.38 0.54 0.22 0.33

0.20 0.259 0.26 0.52 0.47 0.47 0.60 0.29 0.41

0.25 0.281 0.29 0.55 0.50 0.56 0.64 0.35 0.49

0.30 0.304 0.32 0.59 0.55 0.65 0.68 0.41 0.55

0.35 0.342 0.34 0.60 0.61 0.72 0.71 0.47 0.62

0.40 0.417 0.39 0.62 0.63 0.78 0.75 0.52 0.68

0.45 0.460 0.44 0.63 0.64 0.83 0.77 0.57 0.73

0.50 0.537 0.48 0.72 0.69 0.87 0.80 0.62 0.78

0.55 0.581 0.52 0.74 0.76 0.89 0.83 0.66 0.83

0.60 0.660 0.56 0.77 0.78 0.91 0.85 0.70 0.87

0.65 0.684 0.61 0.81 0.82 0.93 0.87 0.74 0.91

0.70 0.723 0.70 0.83 0.86 0.94 0.89 0.78 0.93

0.75 0.748 0.73 0.86 0.88 0.96 0.91 0.82 0.95

0.80 0.779 0.80 0.87 0.89 0.97 0.93 0.86 0.97

0.85 0.828 0.86 0.90 0.90 0.98 0.95 0.90 0.98

0.90 0.892 0.90 0.93 0.93 0.99 0.97 0.94 0.99

0.95 0.944 0.95 0.96 0.96 0.99 0.98 0.97 1.00

1.00 1.000 1.00 1.00 1.00 1.00 1.00 1.00 1.00

67
5.4. COMPARISIONS OF HYETOGRAPHS WITH 2010 RAINFALL EVENT

A comparison was carried out of developed hyetographs and 2010 storm. The hourly data of

severest storm event of 2010 is only available for Daggar and Mardan gauging stations out of

selected raingauges. Hourly rainfall data of Daggar was used for comparison. The heaviest 24

hour rainfall magnitude was extracted from 27 to 31 July 2010 storm. For comparison the

incremental fractions of rainfall magnitudes were arranged in descending order and

cumulative fractions were calculated. The mass curves are shown in Figure 5.13 and

cumulative storm distributions are given in Table 5.11.

SCS Type-1, SCS Type-1A, triangular hyetograph and 2010 storm are 18%, 16%, 13% and

13% rainfall respectively in first 5% time, which are quite similar. Huff curves only having

storm percentage on lower side in first 5% time. The Khan, SCS type-2 and SCS type-3 are

showing higher percentage more than 15% than 2010 storm in first 5% storm time. In first

5% time the storm magnitude of triangular hyetograph and 2010 storm is same. The

triangular hyetograph developed during the study could be used for design flood estimation.

5.5. HEC-HMS MODEL CALIBRATION

HEC-HMS model was calibrated with 28 to 29 June 2007 Storm event with Chakdara

gauging station. The resulted peak flood is 885.9 cumecs on 29 June 2007, which is very

close to observed 874.86 cumecs. The HEC-HMS model calibration results are is shown in

Figure 5.14.

In Figure 5.14 hydrograph of 28 to 29 June rainfall storm event is shown. The first and last

discharge ordinate of the hydrograph is 319.14 cumecs which represents the constant base

flow. The dotted line shows the inflow hydrograph at junction which is less than outflows

because the sub basins 16 and 17 are contributing flood directly to junction-10. The

hydrographs shown below are the out flow hydrographs of subbain16 and 17 which shows

68
that the peak of these basin having major contribution in attaining flood peak of the

calibration point.

J-10 Outflow
-------- J-10 Inflow
…….. S-17 outflow
- - - - S-16 Inflow

Figure 5.14 Resulted hydrograph of model calibration

5.6. COMPUTATION OF PEAK FLOODS USING HEC-HMS

The flood hydrographs are computed using HEC-HMS model. The rainfall of 100 year return

period was used and seven time distributions of storm (i.e. hyetographs) were incorporated as

input to HEC-HMS model. The input hyetographs are presented in Table 5.11. The flood

peaks calculated at the basin outlet with different hyetographs are shown in Table 5.12. The

results show that the highest flood peak of 16650 cumecs was resulted with SCS type 3

distributions and the lowest peak of 8650 cumecs is obtained with Huff approach. Khan’s

approach of hyetograph development yielded flood peak of 11500 cumecs, which is

comparable with the resulted peak flood computed in case triangular hyetograph distribution.

The flood hydrographs computed by using selected hyetographs are shown in Figures 5.15 to

5.21

69
 Table 5.12 Peak floods with various hyetographs at outlet (Cumecs)

Hyetographs Peak Flood (Cumecs)

SCS Type 1 12960
SCS Type 1-A 10300
SCS Type-2 16300
SCS Type-3 16650
Triangular Hyetograph 12020
Khan 11500
Huff 8650

5.7. COMPARISION BETWEEN FLOOD HYDROGRAPHS

The floods were computed separately by applying seven storm distributions (Huff,

Triangular, Khan, SCS Type-1, 1A, 2, and 3). The flood peaks generated by storm

distributions is given in table 5.12. The hydrograph developed from these storm distributions

are shown in Figures 5.15 to 5.21.

Outflow
- - - - - Inflow
…….. S-20 outflow

Figure 5.15 Flood Hydrographs with Khan’s hyetograph

70
 Outflow
- - - - - Inflow
…….. S-20 outflow

Figure 5.16 Flood hydrographs with triangular hyetograph

Outflow
- - - - - Inflow
…….. S-20 outflow

Figure 5.17 Flood hydrographs with Huff curves

71
 Outflow
- - - - Inflow
…… S-20 outflow

Figure 5.18 Flood Hydrographs with SCS type-1

Outflow
- - - - - Inflow
…….. S-20 outflow

Figure 5.19 Flood hydrograph with SCS type-2

72
 Outflow
- - - - - Inflow
…….. S-20 outflow

Figure 5.20 Flood hydrograph with SCS type-3

Outflow
- - - - - Inflow
…….. S-20 outflow

Figure 5.21 Flood hydrograph with SCS type-1A

The flood peaks and hydrograph shapes of SCS Type-2 and SCS Type-3 are same with late

rise and early recession. The resulted flood peaks of SCS type-2 and type-3 are having 4500

and 4200 cumecs difference from Khan, Triangular and SCS Type-1 on higher side.

73
Huff are having flood peaks lower than all other hydrographs, the flood peaks were about

3000 cumecs lesser than the Triangular, Khan’s and SCS-type-1 hyetographs. The resulted

peak flood divided the storm distributions into three categories, hydrographs having high

peaks with SCS type-2 and type-3, hydrographs having moderate flood peaks with Khan,

triangular and SCS type-1 and hydrographs having low flood peaks with Huff and SCS type-

1A.

5.7.1. FLOOD HYDROGRAPHS WITH MORDERATE FLOOD PEAKS

The flood peaks generated with Khan, Triangular and SCS Type-1 hyetographs are quite

similar. The flood hydrographs generated from these three hyetographs are shown in Figures

5.15, 5.16 and 5.18 respectively. The hyetographs are plotted in combine form in Figure 5.22.

The flood hydrograph resulted from Khan’s hyetograph represent sharp flood peak. The sharp

flood peak is due to time distribution of design storm. The recession limb is start form sharp

slope and then moves down with uniform slope. The first and last ordinates of all

hydrographs are 424 cumecs, which is the constant base flow.

The flood hydrograph resulted from Triangular hyetograph is quite similar with Khan’s

except the slope of recession limb. The slope of rising limb and recession limb is same in

triangular. The flood hydrograph peak resulted from SCS Type-1 is almost same with early

peak.

The resulted out flow flood hydrographs from SCS Type-1, Khan and Triangular hyetographs

were plotted separately in Figure 5.22. Three of the two storm distributions applied were

developed by using the data of study area and resulted flood peak of SCS Type-1 was near to

Khan and Triangular. The moderate category was selected for future use as design storm

distribution or design hyetograph on the basis of same 100 year design.

74
 14000

SCS Type-1
12000
Triangular Hyetograph

10000 Wirasat Ullah Khan

Discharge (cumecs)

8000

6000

4000

2000

0
0 10 20 30 40 50 60
Time (hrs)

Figure 5.22 Flood hydrographs with SCS Type-1, Triangular and Khan’s Hyetograph

5.7.2. FLOOD HYDROGRAPHS WITH LOW FLOOD PEAKS

The flood peaks with Huff and SCS Type-1A hyetographs are similar. The flood hydrographs

generated from these two hyetographs are shown in Figure 5.17 and Figure 5.21 respectively.

The flood computed from these two hyetographs shows resemblance although, flood

computed from Huff distribution is less than Khan and Triangular method. These two

hyetographs with low peaks were not suitable for adaptation as the Khan and triangular lies in

moderate category.

75
 `

Chapter VI
CONCLUSIONS AND RECOMMENDATIONS

6.1. CONCLUSIONS

i. The magnitude in peak intensity duration of 2010 storm and triangular hyetograph is

13% of total storm magnitude. The time distribution of triangular hyetograph and 2010

storm event are almost same.

ii. The method of regional hyetograph development could be different for other regions,

different regional design hyetograph development methods gives different results.

iii. The SCS rainfall distributions are developed for the different areas of United States.

These curves, when used for design flood computations, gave either underestimated

results (e.g. for SCS type 1A curves) or overestimated design flood (e.g. for SCS type 2

and type 3 curves). Therefore, regional methods, if available should be considered in

combination of conventional rainfall distribution curves like SCS, Huff and triangular.

76
`

6.2. RECOMMENDATIONS

Finally it is recommended that;

i. The triangular hyetograph and Khan methods are proposed to develop regional design

hyetographs and hyetographs developed with this method during the study will be used

in hydrologic design problems in Sawat region.

ii. SCS Type-1 distribution is also proposed for Sawat region and will be used for design

flood estimation in Sawat region

77
`

REFERENCES

 Alfieri L, Laio F, and Claps P (2007) ‘A simulation experiment for optimal design

hyetograph Selection’ Hydrological Processes DOI: 10.1002/hyp.6646.

 Arnaud P and Lavabre J (1999) ‘Using a stochastic model for generating hourly

hyetographs to study extreme rainfalls’ Hydrological Sciences-Journal-des Sciences

Hydrologiques, 44(3).

 Aidan Millerick (2005) ‘Validation of FSR AND FEH Depth/Duration/Frequency’

The 3rd CIWEM National Conference Bretton Hall, Wakefield.

 Aslam, M.K., Anwar M and Baig S (1991) ‘Mountain environmental management in

Swat District, Pakistan.’ Mountain and Natural Resource Conservation, Published by

International Center for Integrated Mountain Development (ICIMOD), Nepal.

MEM5:1-5.

 Bonta, J.V and Rao, A.R (1989) ‘Regionalization of storm hyetographs’

WaterResources Bulletin, 25(1), pp 211-217.

 Bonta, J.V and Rao A.R (1988) ‘Comparison of four design-storm

hyetographs’Transactions ASAE, 31(1), pp 102-106.

 Bonta, J.V and Rao A.R (1988) ‘Fitting equations to families of dimensionless

cumulative hyetographs’ Transactions ASAE, 31(3), pp 756-760.

 Bonta J.V and Shahalam A (2003) ‘Cumulative storm rainfall distributions’

comparison of Huff Curves Journal of Hydrology (NZ) 42 (1) 64-74.

78
`

 Chow V.T (1953) ‘Frequency analysis of hydrologic data, with special application to

rainfall intensities’ Urbana, Ill. : University of Illinois bulletin, v 50, No81.

 Chow V.T, Maidment David R., and Mays L.W (1988) ‘Applied Hydrology’

McGraw-Hill New York.

 GoP (1999) ‘Population Census Report of Pakistan’ Bauru of Statistics, Government

of Pakistan, Islamabad.

 Gwo-Fong Lin, Lu-Hsien Chen and Shih-Chieh Kao (2005) ‘Development of regional

design hyetographs’ Hydrological processes Hydrol. Process. 19, 937–946

 Guzman G.A and Oliver A.E (1993) ‘A stochastic model of dimensionless

hyetograph’ Water Resources Research 29(7): 2363–2370.

 Hershfield D. M (1961) ‘Rainfall Frequency Atlas of the United States. U.S.

Department of Commerce’ Weather Bureau Technical Paper 40, Washington, DC,

115 p.

 Huff F. A (1967) ‘Time distribution of rainfall in heavy storms’ Water Resources

Research, 3(4), 1007-1019.

 Raghunath H.M (1995) ‘Applied hydrology hand book’.

 IE Aust (2001) ‘Australian Rainfall and Runoff’ A guide to flood estimation volume 1

Australia.

 Keifer C.J and Chu H. H (1957) ‘Synthetic storm pattern for drainage design’ Journal

of the Hydraulics Division, 83(HY4), ASCE, paper 1332, pp 125.

79
`

 Koutsoyiannis D and Foufoula-Georgiou E (1993) ‘A scaling model of a storm

hyetograph’ Water Resources Research, 29(7): 2345-2361.

 Kent K. M (1968) ‘A method for estimating volume and rate of runoff in small

watersheds’ TP-149, Soil Conservation Service, U. S. Department of Agriculture.

 Khan W.U (1980), Consultant Hydrometeorologist, Hydrometeorology of

Baluochistan, Volume I, II & III, UNDP-PAK/73/032.

 NRCS (2007) ‘National Engineering Hand Book of Hydrology 630’ Published by

Natural Recourses Soil Conservation Services of USA.

 NERC (1975) Flood Studies Report. Natural Environment Research Council, London,

vols 1-5.

 Overeem A, Buishand T.A, Holleman I (2009) ‘Extreme rainfall analysis and

estimation of depth-duration-frequency curves using weather radar’ Water Resources

Research, VOL. 45, W10424, 15 PP.

 PMD Publications, Meteorological Data by ‘Pakistan Meteorological Department’

Lahore.

 Pilgrim D.H. and Cordery I (19750 ‘Rainfall temporal patterns for design floods’

Journal of the Hydraulics Division, 101(HY1), ASCE, pp 81-95.

 Rigby E and Bannigan D (1996) ‘The embedded design storm concept – A critical

review’ In Proceedings: 23rd Hydrology and Water Resources Symposium. Hobart,

Australia.

80
`

 Roso S and Rigby T (2006) ‘The Impact of Embedded Design Storms on Flows

within a Catchment’ In Proceedings: 30th Hydrology and Water Resources

Symposium Launceston, TAS.

 Rashid M.G, Archer G. and Marjan G (1999) ‘Resource management plan for Swat

forest range of Swat Forest Division’. Forest Management Center KPK, Forest Deptt.

With Inter Cooperation Govt of Switzerland, pp. 6.

 Shaw L.Y, Charles E, Kent P.A and Hamilton (1984) ‘Temporal Distribution Of

Rainfall In Virgina’ A Cooperative Organization Sponsored Jointly by the (Virginia

Department of Highways & Transportation and the University of Virginia) In

Cooperation with the U. S. Department of Transportation Federal Highway

Administration Charlottesville, Virginia.

 SWH Publications (1961-2010), Annual Reports of River & Climatological Data of

Pakistan, Surface Water Hydrology(SWH) by WAPDA, Volume I & II.

 Tomilnson A.I (1980) ‘The frequency of high intensity rainfalls in Newzeland’ Part I.

MWD Technical publication No.19 Willington, 36p.

 U.S. Soil Conservation Service (1973) ‘A method for estimating volume and rate of

runoff in small watersheds’ Report No. SC-TP-149, Dept. of Agriculture,

Washington, D.C.

 U. S. Department of The Interior, U.S. Geological Survey (2005) ‘Depth-Duration

Frequency Of Precipitation For Oklahama’ Water-Resources Investigations Report

99–4232. Prepared in Cooperation with the Oklahama Department of Transportation.

81
`

 U.S. Soil Conservation Service (1986) ‘Urban hydrology for small watersheds’ Tech.

Rep. 55, Engineering Division, Soil Conservation Service, US Department of

Agriculture, Washington, DC.

 Yen B. C and Chow V.T (1980) ‘Design hyetographs for small drainage structures’

ASCE Journal of the hydraulics division, 106 (HY6), 1055-1076.

82
`

Table A-1: 1-Day Annual Maximum Rainfall at Kalam

Year Rainfall (mm) Year Rainfall (mm)

1962 40 1987 49

1963 47 1988 52

1964 63 1990 19

1965 85 1991 80

1966 41 1992 94

1967 52 1993 38

1968 67 1994 58

1969 45 1995 108

1970 42 1996 52

1971 40 1997 84

1972 43 1998 83

1973 58 1999 99

1974 34 2000 65

1975 138 2001 33

1978 60 2002 60

1979 63 2003 69

1980 40 2004 80

1982 57 2005 63

1983 85 2006 70

1984 49 2007 70

1985 33 2008 99

1986 81 2009 57

83
`

Table A-2: 1-Day Annual Maximum Rainfall at Saidu Sharif

Year Rainfall (mm) Year Rainfall (mm)

1962 38 1987 92

1963 61 1988 88

1964 43 1990 78

1965 46 1991 81

1966 46 1992 96

1967 89 1993 89

1968 49 1994 75

1969 54 1995 67

1970 42 1996 46

1971 60 1997 54

1972 53 1998 78

1973 57 1999 74

1974 34 2000 51

1975 52 2001 81

1978 94 2002 54

1979 58 2003 99

1980 102 2004 60

1982 83 2005 63

1983 63 2006 64

1984 28 2007 100

1985 59 2008 65

1986 98 2009 78

2010 187

84
`

Table A-3: 1-Day Annual Maximum Rainfall at Daggar

Year Rainfall (mm) Year Rainfall (mm)

1963 85 1988 120

1964 197 1990 91

1965 102 1991 54

1966 51 1992 134

1967 102 1993 69

1968 56 1994 94

1969 74 1995 92

1970 96 1996 70

1971 105 1997 65

1972 51 1998 128

1973 113 1999 76

1974 39 2000 124

1975 53 2001 112

1978 99 2002 99

1979 72 2003 147

1980 60 2004 49

1982 81 2005 69

1983 108 2006 105

1984 62 2007 112

1985 73 2008 85

1986 102 2009 55

1987 68 2010 141

85
`

Table A-4: 1-Day Annual Maximum Rainfall at Mardan

Year Rainfall (mm) Year Rainfall (mm)

1963 48 1988 142

1964 62 1990 100

1965 95 1991 93

1966 44 1992 55

1967 99 1993 102

1968 82 1994 60

1969 65 1995 78

1970 74 1996 82

1971 120 1997 89

1972 41 1998 80

1973 71 1999 41

1974 29 2000 90

1975 72 2001 91

1978 107 2002 126

1979 58 2003 81

1980 99 2004 57

1982 110 2005 23

1983 91 2006 49

1984 83 2007 67

1985 67. 2008 66

1986 57 2009 79

1987 42. 2010 145

86
`

Table A-5: 1-Day Annual Maximum Rainfall at Besham Qila

Year Rainfall (mm) Year Rainfall (mm)

1969 59 1992 76

1970 41 1993 59

1971 53 1994 64

1972 57 1995 82

1973 66 1996 56

1974 44 1997 88

1975 74 1998 79

1978 51 1999 73

1979 51 2000 34

1980 57 2001 76

1982 62 2002 102

1983 62 2003 58

1984 46 2004 52

1985 77 2005 82

1986 60 2006 58

1987 68 2007 72

1988 58 2008 63

1990 64 2009 70

1991 78 2010 129

87
`

Table A-6: 1-Day Annual Maximum Rainfall at Dir

Year Rainfall (mm) Year Rainfall (mm)

1983 82 1997 62

1984 92 1998 65

1985 68 1999 67

1986 102 2000 57

1987 90 2001 72

1988 78 2002 66

1989 147 2003 98

1990 68 2004 166

1991 98 2005 87

1992 98 2006 80

1993 125 2007 103

1994 95 2008 76

1995 64 2009 71

1996 57 2010 149

88