Flood2007 Report

8/13/2019 Flood2007 Report

1/78

HYDROLOGICAL ASPECTS OF FLOOD 2007

FINAL REPORT

R 03/ 2008

May 2008

A.K.M. SAIFUL ISLAMANISUL HAQUE

SUJIT KUMAR BALA

INSTITUTE OF WATER AND FLOOD MANAGEMENT

BANGLADESH UNIVERSITY OF ENGINEERING AND TECHNOLOGYDhaka-1000, Bangladesh


2/78

ii

Table of Contents

Table of Contents............................................................................................................................ ii

List of Tables .................................................................................................................................. v

List of Figures ................................................................................................................................ vi

List of Photographs ...................................................................................................................... viiiList of Abbreviations ..................................................................................................................... ix

Acknowledgement .......................................................................................................................... x

Executive Summary ....................................................................................................................... xi

Chapter 1: Introduction ................................................................................................................... 1

1.1 Background ........................................................................................................................... 1

1.2 Objective ............................................................................................................................... 2

1.3 Data and Methodology.......................................................................................................... 2

Chapter 2: Source and causes of Flood 2007.................................................................................. 4

2.1 Chronology of Flood 2007.................................................................................................... 4

2.2 Causes of Flood..................................................................................................................... 5

Chapter 3: Hydrologic characteristics of Flood 2007..................................................................... 9

3.1 Introduction........................................................................................................................... 9

3.2 Comparison of Floods in Major Rivers ................................................................................ 9

3.2.2 Brahmaputra at Bahdurabad station............................................................................. 12

3.2.3 Ganges at Hardinge Bridge station .............................................................................. 12

3.2.4 Meghna River at Bhairab Bazar Staion ....................................................................... 13

3.3 Conclusions......................................................................................................................... 16

3.3.1 Flood Magnitude / Peak ............................................................................................... 16

3.3.2 Flood Entrance ............................................................................................................. 16

3.3.3 Flood Duration............................................................................................................. 16

Chapter 4: Inundation maps using satellite images....................................................................... 18

4.1 Introduction......................................................................................................................... 184.2 Study Area .......................................................................................................................... 20

4.3 Data and Satellite Images.................................................................................................... 22

4.3.1 MODIS/TERRA Satellite images ................................................................................ 22

4.3.2 Flood inundation map based on RADARSAT images ................................................ 22

4.3.3 Water level data in major rivers of Bangladesh........................................................... 22


3/78

iii

4.4 Methods............................................................................................................................... 23

4.4.1 Detecting water related surface using MODIS data .................................................... 23

4.4.2 Modified algorithm of flood inundation maps............................................................. 24

4.5 Validation of proposed techniques...................................................................................... 29

4.6 Results and discussions....................................................................................................... 304.6.1 Spatial extents of floods in Bangladesh ....................................................................... 30

4.6.2 Temporal characteristics of floods............................................................................... 35

4.7 Conclusions......................................................................................................................... 37

Chapter 5: Floods around Dhaka city ........................................................................................... 38

5.1 Hydrologic condition of Dhaka city ................................................................................... 38

5.2 Hydrologic Characteristics of Major Floods in Dhaka City............................................... 40

5.2.1 Buriganga River at Dhaka (Mill barrack) Station........................................................ 40

5.2.2 Balu River at Demra Station ........................................................................................ 40

5.2.3 Turag River at Mirpur Station...................................................................................... 43

5.2.4 Tongi Khal at Tongi Station ........................................................................................ 43

5.3 Conclusions......................................................................................................................... 45

5.3.1 Flood Peak ................................................................................................................... 45

5.3.2 Flood Duration............................................................................................................. 45

5.3.3 Flood Entrance ............................................................................................................. 45

Chapter 6: Performance of flood control works around Dhaka city............................................. 466.1 Comparison of the performance of Flood control structures .............................................. 46

6.2 Performance of Goranchat Bari and Kallayanpur pump station......................................... 49

6.3 Performance of Rampura and Dholai Khal pump station................................................... 49

6.4 Unprotected East part of the Dhaka city ............................................................................. 50

6.5 Filed visit to flood control structures of Dhaka City on August 16, 2007.......................... 52

6.5.1 Dholai Khal Pump House ............................................................................................ 52

6.5.2 Rampura temporary pump station................................................................................ 52

6.5.3 Segunbagicha temporary pump station........................................................................ 52

6.5.4 Kallayanpur pump station............................................................................................ 53

6.5.5 Goranchatbari Pump station......................................................................................... 53

6.5.6 Temporary Pumps along the Buriganga River............................................................. 53

6.5.7 Hatir Jheel Lake and Begun Bari Khal ........................................................................ 54

6.5.8 Begun Bari Khal .......................................................................................................... 54


4/78

iv

Chapter 7: Conclusions and recommendations............................................................................. 57

7.1 Conclusions......................................................................................................................... 57

7.2 Recommendation ................................................................................................................ 60

References..................................................................................................................................... 62


5/78

v

List of Tables

Table No. Title Page No .

Table 3.1: Comparison of characteristics of Flood 2007, 2004, 1998 and 1988 in major rivers. ............................................................................................................................... 11

Table 4.1: MODIS derived indices used to detect spatial and temporal distribution of flood...

............................................................................................................................... 24

Table 5.1: Comparison of characteristics of Flood 2007, 2004, 1998 and 1988 in the

surrounding rivers of Dhaka city. ......................................................................................... 42


6/78

vi

List of Figures

Figure No. Title Page No.

Figure 2.1: Date of crossing danger level in various upstream gauge stations adjacent the boarder

of the country. ......................................................................................................................... 5Figure 2.2: Ganges-Brahmaputra-Meghna (GBM) Basin and River networks enter into

Bangladesh.............................................................................................................................. 7

Figure 2.3: TRMM 3B42 3-hourly rainfall over the globe............................................................. 7

Figure 2.4: Monthly rainfall over Ganges Basin during July using TRMM 3-hourly data. ........... 8

Figure 2.5: Monthly rainfall over Brahmaputra Basin during July using TRMM 3-hourly data. .. 8

Figure 2.6: Monthly rainfall over Meghna Basin during July using TRMM 3-hourly data........... 8

Figure 3.1: Water Level Hydrographs of major rivers for four major flooding years: (a) 1988, (b)

1998, (c) 2004 and (d) 2007.................................................................................................. 10

Figure 3.2: Water level hydrograph of (a) the Brahmaputra at Bahadurabad, (b) the Ganges at

Hardinge Birdge and (c) the Meghna river at Bhairab Bazaar stations of floods in 2007,

2004, 1998 and 1988............................................................................................................. 14

Figure 4.1: Figure Location map of the study area....................................................................... 21

Figure 4.1: ..................................................................................................................................... 22

Figure 4.2: Water level hydrographs of major rivers in Bangladesh for (a) year 2007 and (b) year

2004. Flood starts earlier in 2004 than in 2007. ................................................................... 26

Figure 4.3: Flood chart for developing Flood inundation map using MODIS data...................... 27

Figure 4.4: MODIS derived indices: (a) EVI, (b) LSWI and (c) DVEL for the Seven Land use /

Land cover areas shown in Figure 4.1 .................................................................................. 28

Figure 4.5: Spatial comparison of (a) MODIS derived inundation map of 29 July 2007, (b)

MODIS 8-day false colour composite map (RGB=Band 6,Band2,Band1) of 29 July 2007,

(c) Long term water bodies in 2007 using MODIS data, (d) Inundation map using

RADARSAT image on August 03........................................................................................ 31Figure 4.6: Overlay of MODIS derived inundation map on DOY 209 with the nearest available

RADARSAT derived inundation map on DOY 215. Inundation area derived from MODIS

includes Flood pixels and pixels of Long term water bodies................................................ 32

Figure 4.7: Correlation of inundation area determining from MODIS with that of RADARSAT.

Correlation coefficient R 2 is found very high of about 0.96................................................. 32


7/78

vii

Figure 4.8: Temporal Changes of Flood and Mixture pixels for four types of paddy fields (Single

cropped irrigate rice, Single cropped rain fed rice, Double cropped irrigated rice, and Triple

cropped irrigated rice), Forest and Settlement areas............................................................. 33

Figure 4.9: Spatial distribution of flood inundated area during July and August months of 2007

and 2004. Flood pixels are in blue colour, Mixture pixels are in green colour, Long termwater bodies are in white colours and other areas are in white colour. ................................ 34

Figure 4.10: Flood inundation map of Bangladesh using MODIS images for (a) Year 2004 and

(b) Year 2007. Flood pixels are in blue colour, Mixture pixels are in green colour, Long

term water bodies are in white colours and other areas are in white colour......................... 35

Figure 4.11: Spatial distribution of (a) start date, (b) end date and (c) duration of flood for 2004

and 2007................................................................................................................................ 36

Figure 5.1: Greater Dhaka city area and surrounding river systems............................................. 39

Figure 5.2: Water Level Hydrographs of surrounding rivers for four major flooding years: (a)

1988, (b) 1998, (c) 2004 and (d) 2007 .................................................................................. 41

Figure 5.3: Water level hydrograph of (a) the Buriganga at Dhaka, (b) the Balu at Demar, (c) the

Turag river at Mirpur and (d) the Tongi Khal at Tongi stations of floods in 2007, 2004,

1998 and 1988....................................................................................................................... 44

Figure 6.1: Map of flood control works of Dhaka city during Flood 2007. ................................. 47

Figure 6.2: Difference of riverside and protected side water level during floods in (a) 1998, (b)

2004 and (c) 2007 of the surrounding major flood control works........................................ 48Figure 6.3: Preformence of Goranchat Bari and Kallyanpur pump station during floods in (a)

1998, (b) 2004 and (c) 2007................................................................................................. 51

Figure 6.4: Performance of Rampura and Dholai Khal pump station during floods in (a) 1998, (b)

2004 and (c) 2007 ................................................................................................................ 51


8/78

viii

List of Photographs

Photograph No. Title Page No.

Photo 6.1: Inlet of Dholai Khal pump house [Floating mat of waste carries by Dholai khal. Solid

waste management of the city needs to be improve ] ........................................................... 55

Photo 6.2: Rampura temporary pump station [Negligible impact on drainage congestion.

Permanent pumps needs to be installed.].............................................................................. 55

Photo 6.3: Data collection from Kallayanpur pump station [Digital data recording system can

help researchers and other users to get information quickly] ............................................... 55

Photo 6.4: Retention pond of Goranchat Bari [Only pumping stations where water is not much

polluted and air is fresh.] ...................................................................................................... 55Photo 6.5: Dredging in Turag river to fill wetlands of Dhaka west. [Serious thread to retention

ponds of Goranchat Bari pumping stations] ......................................................................... 55

Photo 6.6: Temporary pumps to drain water logged near Kellar More [Drainage congestion

inside embankment. Some says that this polluted water is major cause of water born

diseases ] .............................................................................................................................. 55

Photo 6.7: Temporary pumps near Hazaribag [Untreated waste water causes serious pollution of

Buriganga river. Water treatment system should be used.] .................................................. 56

Photo 6.8: Retention ponds of Kallayanpur pump station [Encroachments gradually increases and

reduces retention pond area. Authority should act immediately. ] ....................................... 56

Photo 6.9: Buriganga River near Kellar mor [Water hyacinth grows due to polluted water comes

from the pumping of untreated drainage water inside embankment] ................................... 56

Photo 6.10: Inside WL of Segun bagicha temporary pump station [Similar story of Rampura

pump station. Permanent system should be installed.] ......................................................... 56

Photo 6.11: Illegal encroachments tried to grasp Hatir Jheel Lake, one of the major retention area

of the heart of the city. .......................................................................................................... 56

Photo 6.12: Polluted water is discharge in the Begun Bari Khal. The water becomes polluted

while it was carried by a Panthopath Box culvert................................................................. 56


9/78


10/78

x

Acknowledgement

The authors wish to express their sincere thanks to the Committee for Coordination and

Development of the Institute of Water and Flood Management (IWFM) of Bangladesh

University of Engineering and Technology (BUET) for taking initiatives to conduct this presentstudy. We also pay our sincere gratitude to Prof. Rezaur Rahman, the then Director of the

institute for his continuous support in completing the research study successfully.

We also like to express our sincere gratitude to the Employees of BWDB and Dhaka WASA for

helping the BUET study team to provide necessary in situ data during the flood. We like to give

special thanks to Mr. Shariful Islam, Executive Engineer of Goranchat Bari Pump Station of

BWDB for this cooperation and support during our field visit. We are very thankful to

employees of Rampura regulator and Segunbagicha regulator of Dhaka WASA for providing

valuable information about flood control structures around the Dhaka city.

We offer our special thanks to Mr. Ahmedul Hasan of CEGIS for providing Inundation maps

produced by the RADASAT satellite. We also thank Mr. Motaleb Hossain Sarkar of CEGIS for

his valuable suggestion regarding flood inundation maps. This study has been funded by

Bangladesh University of Engineering and Technology (BUET).


11/78

xi

Executive Summary

In the middle of 2007, Bangladesh was hit by a severe flood, a natural disaster which takes

peoples life, kills livestock, destroys infrastructures and communication system, damages crops,

and fish ponds. Despite many adverse impacts, flood is an acceptable phenomenon forBangladesh because of its immense benefits on agriculture and food production, fisheries and

livestocks.

Normal flooding brings many benefits such as increase of soil fertility, recharge of aquifer,

improvement of ecosystem, and increase of fish production, etc. It is an established fact that after

each major flood, the agricultural sector of the country gets boom up and the food production

increases dramatically.

Flood 2007 is the 5 th major flood for the period of last twenty years, when more than 35% of the

country area was inundated. The people of the country suffered greatly and the production of

Aman crop was seriously damaged during the 2007 flood. Like other major floods in the past,

Flood 2007 also has its own significance and characteristics, which need to be studied to adopt

better flood management strategies. Institute of Water and Flood Management (IWFM) usually

carries studies on water and flood management of Bangladesh. In no exceptions, a study team

was formed by the institute right after the Flood 2007 to study the hydrologic aspects of it. The

findings of the investigation have been presented in this report.

This study has given focus on the hydrologic characteristics of floods by analyzing hydrologic

data of three major rivers: the Ganges, the Brahmaputra and the Meghna (GBM). The main

source of monsoon river flood in Bangladesh is rainfall in the GBM basin which is carried by the

tributaries and distributaries of these three major rivers. The first crossing of danger level inside

Bangladesh was observed on 19th

July 2007 at Durgapur station of the Someswari river and atSunamganj station of the Surma river. On next day, flood water crosses danger level at

Kanaighat station of the same Surma river. However, crossing of danger level at Amalshid

station in the Kushiyara river of Meghna basin was much later on 28 th July 2007. Flood in

Brahmaputra basin was observed on the last week of July. On 26 th July 2007, flood water crossed

danger level at Dalia of the Teesta river. On the next day, water level of the Dharala river at


12/78

xii

Kurigram station was above the danger level. The Brahmaputra river at Noonkhawa and at

Chilmari stations overtopped danger levels on 28 th July 2007. During 2007 Flood, the water level

of the Ganges river did not cross the danger level. The far most gauge station inside the country

on the Ganges is located at Pankha, where the water level was always below the danger level. No

peak synchronization of flood like 1998 and 2004 was occurred in the country during 2007 flood,as the Ganges experienced no crossing of danger level of water over the flood period.

The magnitude of water level above the danger level in the Brahmaputra river during Flood 2007

was higher than that of during Flood 2004, similar to Flood 1998 and much lower in Flood 1998.

In the Ganges river, there was no flood during 2007 and 2004 floods. On the other hand, the

most severe flood was observed in the Ganges basin of Bangladesh during 1998 and 1988 floods.

The magnitude of water level above danger level in the Meghna river during 2007 was almost

half of that was during 2004 flood. The most severe flood in terms of magnitude occurred in this

basin during 2004 and 1988 floods. The magnitude of the peak flood during 1998 flood in this

basin was also much higher than that of 2007 flood. The duration of water level above the danger

level in the Brahmaputra river in 2007 flood was longer than that of 2004 and 1988 floods but

much shorter than 1998 flood (about one third). In 1998 and 1988, flooding was occurred in the

Ganges basin and the most prolonged flooding was observed during 1998 flood. Duration of

floods in the Meghna river basin of Bangladesh in 2007 flood is similar to that of 2004 flood. It

is to be mentioned that long lasting flood was observed in this basin during 1988 flood and wastwice the duration of flood during 2007 flood. During 1998 flood, the duration of flood in the

Meghna river basin of Bangladesh was shorter than that of 1988 flood but was much longer than

that during the 2007 and 2004 floods.

It is a fact that the main source of river floods in Bangladesh is the huge rainfall in upper

catchment of the GBM basin. So, rainfall data from TRMM satellite was analyzed to reveal

characteristics and amount of rainfall over the GBM basin during 2007 flood. Mean monthly

rainfall plot of TRMM satellite data shows that for both the Meghna and Brahmaputra basins,

rainfall was higher in July 2007 than that of last two years. This excess rainfall was accumulated

and carried by the Brahmaputra and the Meghna rivers to Bangladesh and caused flooding.

For detecting spatio-temporal extent of inundation of floods during 2007 and 2004 floods, time-

series MODIS surface reflectance data were used for the study. Flood inundation maps were


13/78

xiii

developed from vegetation and land water surface indices derived from surface reflectance. The

inundation map developed from MODIS data has been compared with a consequent

RADARSAT image. The estimates show a strong correlation with the inundated area derived

from RADARSAT products [R 2: 0.96]. The products derived from MODIS 500m imagery show

the ability to study flood dynamics and perform similar to RADARSAT based flood assessments.As MODIS products have a great advantage in the high-frequent observation, it may be

concluded that this developed method is a useful one to clarify the entire extent of the temporal

floods in Bangladesh.

A special emphasis has been given to reveal hydrologic characteristics of floods in and around

the Dhaka city. Dhaka is not only the nations capital but also will become a mega city by 2010.

Dhaka city always draws special attention during flood period due to its strategic importance.

Date of crossing of danger level at the rising stage of flood was first week of August during 2007

flood. The danger level was first crossed at Tongi Khal on 1 st August 2007. On next day, the

danger level was crossed on Balu rive at Demra and after two days the danger level was crossed

at Mirpur of the Turag river. After one week of crossing danger level of the Tongi Khal, water

level of the Buriganga river crossed its danger level. In 1998 and 1988, floods observed first in

the Turag river, later in the Tongi Khal and finally in the Buriganga river. But, Flood 2004 was

observed first in the Turag river, then in the Buriganga river and finally in the Tongi Khal.

Magnitude of the peak flood above danger level of all the rivers around the Dhaka city in 2007

was below the level of flood in 2004, 1998 and 1988. The heights of flood peak were found for

1988 for all the rivers around Dhaka. In 1998 and 2004, magnitude of the highest flood level was

similar and lower than that of in 1988. Duration of floods in the Buriganga river was only a

single day in 2007. Flood duration of other adjacent rivers was more than twenty days in 2007.

The longest flood was found in 1998 and it was almost 3 times longer than floods in 2007.

Durations of water level above danger level in 1998 and in 2004 were similar to the duration of

flood 2007.

The performance of three permanent pump stations at Dholai Khal, Goranchat Bari and

Kallayanpur was found satisfactory. The water levels of protected side were far below than that

of river side. On the other hand, the difference of the water level between country and protected


14/78

xiv

side at Rampura regulator, where temporary pumps were installed, was close to zero during all

the major floods. Performance of the temporary pumps was insignificant and had very little

effect on the drainage congestion in flood season. It was found during field visits that urban

encroachments are gradually increasing and the retention pond areas of the Kallaynpur and

Goranchat Bari pump stations are reducing. This phenomenon is supposed to enhance the waterlogging from a small amount of intense precipitation.

The eastern part of Dhaka city has suffered from flooding during all the major floods. This part

of the city should be protected by road cum embankment. But, before the construction of such

an embankment, the drainage system of the Dhaka East should be well planned. The city

dwellers have bad experience of water logging and drainage congestion in the western part due

to the embankment. Lessons should be learned from the drainage situation of western part of

the city. During floods in 2007, the number of patients of Diarrhoea and other water borne

diseases was all time high. The main source of illness is drinking water supplied by Dhaka

Water Supply and Sewerage Authority (WASA). Entrance of waste water in the water supply

pipe through leakage and faulty joints are reported as one of the main causes of these diseases.

Except a few, the most of the industries discharged untreated waste water in the river. This

untreated water is recognized as the main source of water pollution during flood.


15/78

1

1.1

Chapter 1: Introduction

Background

Flood is natural and recurring phenomenon for Bangladesh due to its unique geographic location.

The country is located at a floodplain delta of three major river basins: the Ganges, the

Brahmaputra and the Meghna (GBM). Every year, one fourth to one third of the country is

inundated during monsoon season by overflowing rivers. These normal flooding brings many

benefits such as increase of soil fertility, recharge of aquifer, improve of ecosystem, and increase

of fish production, etc. This regular and natural flood is acceptable for the nation because of its

immense benefits on agriculture and food production, fisheries and livelihood. However, the

degree of this inundation sometimes become severe and causes damage to infrastructures, crops,communication system, and human being. Flood in 2007 was one such sever flood in recent

years in terms of magnitude and duration.

Four types of flood are encountered in Bangladesh: Flash Floods, River Floods, Tidal Floods and

Storm Surge (K. M. N. Islam, 2006). Flash floods are observed in northern and north-east hilly

region of Bangladesh in the pre-monsoon months of April and May. Flash floods are known for

its sudden high discharges and velocities, and abrupt rise and recession. The steeper slopes cause

rapid high velocities of runoff in response to the rainfall in the hills. The most common forms of

floods in the country are known as river floods caused from huge rainfall during monsoon season

from June to September. Rainfalls in the GBM basin during monsoon season are carried by the

tributaries and distributaries of the respective major rivers system. Finally, huge amount of water

carried by these three major rivers often overflows banks of the country and stays long time

during the monsoon season. Another type is tidal flood which is observed in the south west and

south central parts of the country. Tidal rivers and estuaries experience this type of flood due to

astronomical tide from the Bay of Bengal. Sometimes this type of flood in the estuary can prolong the duration of the river flood by slowing down of the flood recession. Apart from the

above three types of normal flood, coastal area is occasionally hit by storm surge flood when a

heavy cyclone passes over the land. Storm surge often causes huge fatalities and deaths of

livestock due to its magnitude and the fastest travel time.


16/78

2

1.2

1.3

In 2007, Bangladesh has been suffered from heavy monsoon river floods during the month of

July and August. Flood water came mostly through the two major rivers of the Brahmaputra and

the Meghna. Flood 2007 is the 5 th major flood of last twenty years where more than 35% of the

country area is inundated. People suffered greatly and the production of Aman crop wasseriously damaged during this flood. Unlike other major floods in the past, Flood 2007 also has

its own significance and characteristics which need to be studied to adopt better flood

management strategies.

Institute of Water and Flood Management (IWFM) usually carries studies on water management

and floods of Bangladesh. Islam et al. (2002) have conducted a study on the Hydrologic

characteristics of floods in 1998. Chowdhury et al. (1998) have studied the impact of 1998 flood

on Dhaka city and performance of flood control works. Rahman et al. (2005) have investigated

the hydrologic aspects of Flood-2004 in major rivers of Bangladesh and special emphasis has

given on the floods of Dhaka city. In no exceptions, a study team has been formed by the

institute right after the Flood 2007 to study the hydrologic aspects of it. The findings of that

investigation have been presented in this report.

Objective

The main objectives of this study are as follows:

to investigate hydrologic characteristics of flood in 2007,

to compare Flood 2007 with the major floods in recent history,

to investigate the causes of flood 2007,

to develop flood inundation map using satellite images, and

to evaluate the performance of flood control works around Dhaka city.

Data and Methodology

This study has been carried out based on secondary data and field visits. Water level data has

been collected from Bangladesh Water Development Board (BWDB). Daily rainfall data were

provided from various meteorological stations of Bangladesh Meteorological Department


17/78

3

(BMD). Several field visits were conducted in and around Dhaka city during the September and

October 2007. The main objectives of this field visits was to investigate the flood situation and

the performance of the flood control works around the city. During the field visits, the study

team interviewed local people and officials of the responsible authorities. Data analysis was

conducted based on the performance of flood control works.

Satellite images are taken from MODIS TERRA 8-day reflectance (Surface Reflectance 8-Day

L3 Global 500m) during the 2004 and 2007. A total of MODIS 92 images were used to

determine flood inundation area. Rule based classification techniques were used to derive

inundation maps from MODIS images. One RADARSAT ScanSAR Wide Beam image of 3 rd

August 2007 was used to compare inundation area derived from MODIS image.

3-hourly rainfall data with 0.250.25 degree resolution measured by Tropical Rainfall Measuring

Mission (TRMM) satellite with the 3B42 (Version 6) processing algorithm has been used to

quantify the precipitation over Ganges-Brahmaputra-Meghna (GBM) basin. Mean monthly

rainfall has been calculated for Monsoon season starting from satellite launch of 1998 to 2007.


18/78

4

2.1

Chapter 2: Source and causes of Flood 2007

Chronology of Flood 2007

Flood 2007 was observed, at first, in the rivers of the Meghna basin, later in the rivers of the

Brahmaputra basin. The date of crossing danger level is shown in the Figure 2.1 along the names

of the various gauge stations. The first crossing of danger level inside Bangladesh was observed

on 19 July 2007 at Durgapur station of the Someswari river and at Sunamganj station of the

Surma river. On the next day, flood water crossed danger level at Kanaighat station of the same

Surma river. However, crossing of danger level at Amalshid station in the Kushiyara river of the

Meghna basin was much later on 28 July 2007.

Flood in the Brahmaputra basin was observed on the last week of July. On 26 July 2007,

floodwater crossed danger level at Dalia of the Teesta river. On the next day, water level of

Dharala river at Kurigram station showed above the danger level. The Brahmaputra river at

Noonkhawa and at Chilmari stations was overtopped above their danger level on 28 July 2007.

During 2007 Flood, the water level of the Ganges river was not crossed above the danger level.

The far most gauge station inside the country on the Ganges is located at Pankha where water

level was always below the danger level. Since there was no sever flood in the Ganges river, so

unlike floods in 1998 and 2004, there occurred no peak synchronization in 2007 flood.


19/78

Figure 2.1: Date of crossing danger level in various upstream gauge stations adjacent the boarderof the country.

2.2 Causes of Flood

Figure 2.2 shows the GBM basin and its major river networks. Three major rivers - the Ganges,

the Brahmaputra and the Meghna carry huge amount of water during monsoon season and cause

monsoon flood of the country. Most of the rainfall occurs in the GBM basin during the monsoon

season, namely June to August. This heavy rainfall creates overland flows and runoff which

eventually flow into the above three major rivers. If rainfall amount of any year is much higherthan that of a normal year, river overflows its bank and cause flood. When these rivers enter into

Bangladesh, they carry huge amount of water and cause flooding in Bangladesh. It takes nearly a

month for rainwater in the GBM basin to reach Bangladesh carried by these major rivers.

5


20/78

6

To investigate the source of the floods 2007, the total monthly rainfall occurred in the GBM

basin during the monsoon season has been estimated and analyzed. Monthly rainfall over GBM

basin has been calculated using 3-hourly TRMM satellite data. Figure 2.3 shows an example plot

of 3-hourly rainfall over the globe using TRMM images of 0.250.25 degree resolution. Monthly

rainfall has been estimated by adding all the 3-hourly data over the monthly period. Finally,using ILWIS Geoinformatics software, the total monthly rainfall for each of the GBM basins has

been calculated. Figure 2.4 shows TRMM derived total monthly rainfall of July over the Ganges

basin for 1998-2007. In the Ganges basin, rainfall in 2007 was not much higher than that of other

years. This represents the fact that during 2007, there was no significant flood occurred in the

Ganges river basin. Figure 2.5 shows TRMM derive accumulated rainfall during July of each

year from 1998 to 2007 for the Brahmaputra basin. Rainfall in July of 2007 was much higher

than the average rainfall in the Brahmaputra basin. Due to this heavy rainfall in this basin during

July, the water level of the Brahmaputra river crossed danger level and flooding occurred at end

of July.

Similar scenario of the rainfall in the Brahmaputra basin can be found for the Meghna basin for

2007. Figure 2.6 shows rainfall occurred in the Meghna basin in July for 1998-2007 using

TRMM 3-hourly data. In this basin, rainfall is higher in July than last two years and causes

flooding in the downstream of this basin. Such heavy rainfall increase the amount of water in the

river and danger level of the Meghna river was crossed in the middle of July of 2007.

The amount of rainfall during 2007 in the Meghna basin was less than that of 2004. Therefore, in

this basin, duration and magnitude of flood in 2007 should be smaller than that of flood in 2004.

In the following sections, the magnitude and duration of the recent floods will be compared

where we will see the appropriateness of this conclusion. The amount of rainfall in the

Brahmaputra basin in 2007 was slightly less than that in 2004. But the magnitude of peak flood

in 2007 was slightly higher han that of flood in 2004. It can be inferred that the amount of

rainfall in the Brahmaputra basin is not the only factor that afects flood peak and duration.


21/78

Figure 2.2: Ganges-Brahmaputra-Meghna (GBM) Basin and River networks enter intoBangladesh

Figure 2.3: TRMM 3B42 3-hourly rainfall over the globe

7


22/78

0

2000040000

60000

80000

100000

120000

140000

1 9 9 8

1 9 9 9

2 0 0 0

2 0 0 1

2 0 0 2

2 0 0 3

2 0 0 4

2 0 0 5

2 0 0 6

2 0 0 7

Year

M o n t

h l y R a i n

f a l l ( m m

)

Figure 2.4: Monthly rainfall over Ganges Basin during July using TRMM 3-hourly data.

0

10000

20000

30000

40000

50000

60000

70000

1 9 9 8

1 9 9 9

2 0 0 0

2 0 0 1

2 0 0 2

2 0 0 3

2 0 0 4

2 0 0 5

2 0 0 6

2 0 0 7

Year

M o n

t h l y R a i n

f a l l ( m m )

Figure 2.5: Monthly rainfall over Brahmaputra Basin during July using TRMM 3-hourly data.

0

2000

4000

6000

8000

10000

12000

1 9 9 8

1 9 9 9

2 0 0 0

2 0 0 1

2 0 0 2

2 0 0 3

2 0 0 4

2 0 0 5

2 0 0 6

2 0 0 7

Year

M o n

t h l y R a i n

f a l l ( m m

)

Figure 2.6: Monthly rainfall over Meghna Basin during July using TRMM 3-hourly data.

8


23/78

9

3.1

3.2

Chapter 3: Hydrologic characteristics of Flood 2007

Introduction

The main sources of river floods in Bangladesh are the bank overflow from the major rivers and

their tributaries or distributaries. Three major river systems of Bangladesh - the Brahmaputra, the

Ganges and the Meghna carries huge amount of water from the precipitation over large area of

the GBM basin. About 80% of the annual rainfall of Bangladesh occurs during monsoon season

between June to September. The major cause of the monsoon flood relies of the intensity,

duration and magnitude of the rainfall in the GBM basin. The study of the hydrological behavior

of the three above mentioned major rivers would pave to the understanding of the hydrologiccharacteristics of flood 2007. Monsoon flood in major rivers for 2007 has been compared with

major floods in recent past. The elements of the comparison are mainly focused on the

magnitude, flood peak, duration of flood.

Comparison of Floods in Major Rivers

A comparison of floods in major rivers in 2007 with other major floods of recent past in 2004,

1998 and 1988 was conducted. Water level data from one gauge station in each major river is

used for this comparison. Bahadurabad in the Brahmaputra river, Hardinge Bridge in the Ganges

river and Bhariab Bazar in the Meghna river were the three well known gauge stations where

records of historic flood are available and used in this study. Figure 3.1 shows water level

hydrographs of three major rives for floods in 2007, 2004, 1998 and 1988. It can be found for

these hydrographs that floods in major river systems have distinct characteristics and patterns.

Comparison was made in terms of danger level, the date of crossing of danger level of falling

curve, the date of crossing of danger level of recession curve, the height of peak flood level

above local datum (PWD) and the duration of flood above danger level. A summary of this flood

statistics are presented in Table 3.1. In the following sub sections, the characteristics of floods in

each major rivers of Bangladesh will be discussed in detail.


24/78

(a) Flood 1988

-

4

8

12

16

20

24

1 - M a y

1 6 - M a y

3 1 - M a y

1 5 - J u n

3 0 - J u n

1 5 - J u l

3 0 - J u l

1 4 - A u g

2 9 - A u g

1 3 - S e p

2 8 - S e p

W a t e r

l e v e

l ( m , P

W D )

Ganges(at Hardinge Bridge)

RHWL

RHWL

RHWL

DL

DL

DL

Brahmaputra

(at Bahadurabad)

Meghna(at Bhairab Bazar)

(b) Flood 1998

-

4

8

12

16

20

24

1 - M a y

1 6 - M a y

3 1 - M a y

1 5 - J u n

3 0 - J u n

1 5 - J u l

3 0 - J u l

1 4 - A u g

2 9 - A u g

1 3 - S e p

2 8 - S e p

W a t e r

l e v e

l ( m , P

W D )


RHWL

RHWL

RHWL

DL

DL

DL

Brahmaputra

(at Bahadurabad)


(c) Flood 2004

-

4

8

12

16

20

24

1 - M a y

1 6 - M a y

3 1 - M a y

1 5 - J u n

3 0 - J u n

1 5 - J u l

3 0 - J u l

1 4 - A u g

2 9 - A u g

1 3 - S e p

2 8 - S e p

W a t e r

l e v e

l ( m , P

W D )


RHWL

RHWL

RHWL

DL

DL

DL

Brahmaputra(at Bahadurabad)


(d) Flood 2007

-

4

8

12

16

20

24

1 - M a y

1 6 - M a y

3 1 - M a y

1 5 - J u n

3 0 - J u n

1 5 - J u l

3 0 - J u l

1 4 - A u g

2 9 - A u g

1 3 - S e p

2 8 - S e p

W a t e r

l e v e

l ( m , P

W D )


RHWL

RHWL

RHWL

DL

DL

DL

Brahmaputra(at Bahadurabad)


Figure 3.1: Water Level Hydrographs of major rivers for four major flooding years: (a) 1988, (b)1998, (c) 2004 and (d) 2007

10


25/78

11

Table 3.1: Comparison of characteristics of Flood 2007, 2004, 1998 and 1988 in majorrivers.

Parameters River Gauge Stn. 2007 2004 1998 1988

Brahmaputra Bahadurabad 19.5 19.5 19.5 19.5

Ganges HardingeBridge

14.25 14.25 14.25 14.25

Danger Level inmeters abovePWD datum

Meghna Bhairab Bazar 6.25 6.25 6.25 6.25

Brahmaputra Bahadurabad 27.07.07 &08.09.07

11.07.04 07.07.98 09.07.88 &24.08.88


- - 20.08.98 16.08.88

Date of crossingDanger Level atrising stage

Meghna Bhairab Bazar 30.07.07 &

12.09.07

11.07.04 20.07.98 06.07.88 &

14.08.88Brahmaputra Bahadurabad 06.08.07&

17.09.0726.07.04 12.09.98 12.07.88 &

04.09.88


- - 15.09.98 07.09.88

Date of crossingDanger Level atfalling stage

Meghna Bhairab Bazar 21.08.07 &25.09.07

18.08.04 25.09.98 05.08.88 &27.09.88

Brahmaputra Bahadurabad 0.88 0.68 0.87 1.12

Ganges Hardinge

Bridge

- - 0.94 0.62

Height of peakflood level in meteraboveDanger Level

Meghna Bhairab Bazar 0.69 1.53 1.08 1.41

Brahmaputra Bahadurabad 21 15 67 16


0 0 26 23

Duration of floodin days aboveDanger Level

Meghna Bhairab Bazar 37 38 67 75


26/78

12

3.2.2 Brahmaputra at Bahdurabad station

The magnitudes of water level of peak flows at Bahdurabad in the Brahmaputra in 2007, 2004,

1998, 1988 were found 0.88m, 0.68m, 0.87m and 1.12m PWD above danger level respectively

(Table 3.1). In this river basin, magnitude of Flood 2007 was much smaller than the other twomajor floods in 1998 and 1988. In terms of magnitude of peak flow, Flood 2007 in this basin is

higher than Flood 2004. Water level hydrographs of the Brahmaputra river at Bahdurabad was

plotted for floods in 2007, 2004, 1998 and 1988 and are shown in Figure 3.2 (a).

The date of crossing danger level in the Brahmaputra river at Bahdurabad during 2007, 2004,

1998 and 1988 was on 27 th, 11 th, 7 thand 9 th of July respectively. It has been found that Flood

2007 was delayed than the past three major floods. The recession also took place quickly and the

duration of flood was not long as compared to the floods in 1998. The durations of flood at

Bahdurabad in the Brahmaputra during 2007, 2004, 1998 and 1988 were found 21, 15, 67 and 75

days respectively. The duration of flood in the Brahmaputra river basin in Bangladesh in 2007

was higher than floods in 2004 and 1988 but much lower than floods in 1998.

3.2.3 Ganges at Hardinge Bridge station

At Hardinge bridge of the Ganges river, the magnitudes of water level during floods in 2007, and

2004 was always below the danger level. During 1998 and 1988, magnitudes of the peak flow

were found 0.94m and 0.62m PWD above danger level respectively. In terms of magnitude,

severe flooding occurred in the Ganges river basin of Bangladesh during 1998. Water level

hydrographs of the Ganges river at Hardinge Bridge was plotted for floods in 2007, 2004, 1998

and 1988 and are shown in Figure 3.2 (a). The dates of crossing of danger level in the Ganges

river for flood 1998 and 1988 were on 20 th and 16 th of August respectively. The durations of

flood at Hardinge Bridge in the Ganges in 1998 and 1988 were found 26 and 23 daysrespectively.


27/78

13

3.2.4 Meghna River at Bhairab Bazar Staion

The magnitudes of peak flows at Bhairab Bazar in the Meghna in 2007, 2004, 1998, 1988 were

found 0.69m, 1.53m, 1.08m and 1.41m PWD above danger level respectively (Table 3.1).Magnitude of Flood 2007 was much lower than the other three major floods of recent past. Water

level hydrographs of the Meghna river at Bhairab Bazar was plotted for floods in 2007, 2004,

1998 and 1988 and are shown in Figure 3.2 (c). The start of crossing danger level in the Meghna

river for flood 2007 was on the 30 th July which was delayed than other 3 major floods. The

recession also takes place quickly and the duration of flood was not long as compared to the

floods in 1998 and 1999. The durations of flood at Bhairab bazaar in the Meghna in 2007, 2004,

1998 and 1988 were found 37, 38, 67 and 75 days respectively. It can be inferred that in terms of

magnitude and duration, Flood 2007 in the Meghna river were less significant than those of other

three floods of recent past.


28/78

(a) Brahmaputra at Bahdurabad

17

18

19

20

21

3 0 - J u n

1 0 - J u

l

2 0 - J u

l

3 0 - J u

l

9 - A u g

1 9 - A u g

2 9 - A u g

8 - S e p

1 8 - S e p

2 8 - S e p

W a t e r

L e v e l

( m , P

W D )

1988

1998

2004

2007

DL

RHWL

RHWL

DL

(b) Ganges at Hardinge B ridge

8

9

10

11

12

13

14

15

16

3 0 - J u n

1 0 - J u

l

2 0 - J u

l

3 0 - J u

l

9 - A u g

1 9 - A u g

2 9 - A u g

8 - S e p

1 8 - S e p

2 8 - S e p

W a t e r

L e v e l

( m , P

W D )

1988

1998

2004

2007

DL

RHWL

RHWL

DL

(c) Meghna at Bhairab Bazar

4

5

6

7

8

3 0 - J u n

1 0 - J u

l

2 0 - J u

l

3 0 - J u

l

9 - A u g

1 9 - A u g

2 9 - A u g

8 - S e p

1 8 - S e p

2 8 - S e p

W a t e r

L e v e l

( m , P

W D )

1988

1998

2004

2007

DL

RHWL

RHWL

DL

Figure 3.2: Water level hydrograph of (a) the Brahmaputra at Bahadurabad, (b) theGanges at Hardinge Birdge and (c) the Meghna river at Bhairab Bazaar stations of floods

14


29/78

15

in 2007, 2004, 1998 and 1988.


30/78

16

3.3 Conclusions

Conclusion on the investigations on hydrological characteristics of floods in major rivers can be

focused on the following points:

3.3.1 Flood Magnitude / Peak

The magnitude of water level above danger level in the Brahmaputra river during Flood 2007

was higher than Flood 2004, similar to Flood 1998 and much lower than Flood 1998. In the

Ganges river, there was no flood occurred in Bangladesh during 2007 and 2004. On the other

hand, the most severe flood was observed in the Ganges basin of Bangladesh during 1998 and

1988. The magnitude of water level above danger level in the Meghna river during 2007 was

almost half of it during 2004. The most severe flood in terms of magnitude occurred in this basin

during 2004 and 1988. Also, during 1998 the magnitude of the peak flood in this basin was much

higher than that of 2007.

3.3.2 Flood Entrance

Date of crossing of danger level at Bhadurabad in the Brahmaputra river was at the end of July

2007. On the other hand, the first flood wave in the Brahmaputra river was observed in early

weeks of 2004, 1998 and 1988. There was no flooding in the Ganges river basin of Bangladesh

during 2007 and 2004. The entrance of flood wave in the Ganges basin was observed in the

middle of August in 1998 and 1988. In the Meghna basin, the first flood wave entered into

Bangladesh in the last week of July in 2007. In 2004 and 1988, flood was observed in the

Meghna river basin of Bangladesh much earlier (at the beginning of July) than in 2007, while in

1998, the first flood wave was observed in the middle of July.

3.3.3 Flood Duration

The duration of water level above danger level in the Brahmaputra river in 2007 was higher than

that of 2004 and 1988 but much lower than 1998 (about one third). Therefore, in terms of

duration, Flood 2007 in the Brahmaputra river basin was moderate than recent known floods.


31/78

17

There was no flood observed in the Ganges river basin of Bangladesh in 2007 and 2004. In 1998

and 1988 flooding was occurred in the Ganges basin and most prolonged flooding was observed

in 1998. Duration of floods in the Meghna river basin of Bangladesh in 2007 is similar to that of

2004. Although the long lasting flood was observed in this basin in 1988 and it was twice the

duration of flood in 2007. In 1998, the duration of flood in the Meghna river basin of Bangladeshwas less than 1988 but has much higher than in 2007 and 2004.


32/78

18

4.1

Chapter 4: Inundation maps using satellite images

Introduction

Flood is very common phenomenon in Bangladesh due to its geographic location. Natural river

floods occurred almost every year during monsoon season between June to September. Every

year one fourth to one third of the country is inundated during monsoon season by overflowing

rivers. Flood disrupts peoples live, damage infrastructures and road networks in urban areas. In

rural areas it damages crops, causes death to livestock and people become isolated due to the

unavailability of communication mode. On the other hand flood has many positive impacts to the

environment. Flood supplies nutrients to the soil, recharges ground water, enhances diversity of

aquatic species, and naturally washes out solid wastes. It is very common that crop production inthe country gets almost doubled right after any major flood. Therefore, understanding the current

status of flood inundation in time and space is important in evaluating the relationships between

variations in the water regime, local agricultural activity, and ecosystem behavior from a global

viewpoint.

Remote sensing images can be effective and efficient tools to determine flood inundation areas.

In the past, many studies have been conducted using remote sensing data to detect spatial and

temporal changes of flood inundation areas, delineate wetlands and study its changes, flood

damage assessments in urban areas, dynamics and behaviors of floods. Those studies mainly

detect surface water resources using a range of sensors and satellites. To select suitable sensor

which is both cost effective and efficient to develop flood inundation is a major challenge. One

of the major problems of optical sensors is its inability to penetrate clouds. On the other hands, in

the monsoon period when flooding occurs, the sky is covered most of the days by cloud. In this

context, Synthetic Aperture Radar (SAR) has been considered as the most effective sensor in

detecting flood inundated area. In the past, images from various satellites such as RADARSAT,JERS-1, ERS-1/2, and ENVISAT have previously been used to detect inundated areas in

numerous of possible ways (Henry, Chastanet, Fellah, & Desons, 2003; Heremans et al., 2005;

Hirose, Maruyama, Quy, Tsukada, & Shiokawa, 2001; Ishitsuka, Saito, Murakami, Ogawa, &

Okamoto, 2003; Laugier, Fellah, Tholey, Meyer, & De Fraipont, 1997; Liew et al., 1998;

Nguyen & Bui, 2001; Wang, 2002, 2004).


33/78

19

Although RADARSAT and other synthetic aperture radars are a capable of monitoring land

surface, it is not feasible to use them for monitoring a huge areas for a long time due to its high

data acquisition cost. Keeping in mind the cost of production, inundation maps were developed

using data from various low resolutions optical sensors. Data from NOAA/AVHRR of 1.1km,SPOT of 1km (Harris & Mason, 1989; Liu, Huang, Li, & Wan, 2002; Xiao et al., 2002a), SSM/I

of 13km, and MOS/SMR of 23km spatial resolution (Jin, 1999; Tanaka, Sugimura, & Tanaka,

2000; Tanaka, Sugimura, Tanaka, & Tamai, 2003) has been considered as alternative means of

mapping water surface. Most of these data are freely available from internet for daily basis which

makes it possible to detect changes of inundation areas for a large area.

The spatial resolution of SSM/I and MOS/SMR is very high as compare to NOAA/AVHRR and

SPOT data. This makes NOAA/AVHRR and SPOT an effective tool to detect inundation and

temporal changes in the extent of flooded areas. Various techniques such as discrimination of

low cloud coverage by Bryant and Rainey (2002), discrimination between water and land

surfaces using band rations by Sheng and Gong (2001) are also developed to reduce and detect

the effect of cloud cover. Deriving indices such as Normalized Difference Water Index (NDWI)

by Xiao et al. (2002b) is also efficient in detecting the inundation and transplantation of rice.

On the other hand, right after its launching in December 1999, MODIS satellite with itsmoderate-resolution optical sensor of 250500 m becomes useful tools for scientific studies and

research. Many studies were conducted to determine surface water content such as estimating the

extent of paddy fields by Xiao et al. (2006; 2005), detecting inundation areas through vegetation

cover conversion by Zhan et al. (2002) etc. The Dartmouth Flood Observatory (2006) monitors

flood disasters all over the world using MODIS data. Up to 2007, the observatory published an

annual inundation map of Bangladesh via the Internet (Anderson, Brakenridge, & Caquard,

2005).

Sakamoto et al. (2007) was developed a methodology to detect the spatio-temporal flood

distribution in the Cambodia and Vietnam using MODIS data. The main advantages of this

methodology are: (1) time series data is available during flood period, (2) data is available for the

globe, (3) data can be downloaded free of cost through internet and (4) the accuracy of flood

inundation map lies within the acceptable range [R 2 lies between 0.77 and 0.97]. Application of


34/78

20

4.2

their methodology for flood inundation mapping of Bangladesh is also a difficult task due to

dissimilarity of the hydro-geological conditions among Cambodia, Vietnam and Bangladesh.

Some modifications are essential to adopt their methodology that is proposed in this study. Using

this modified methodology, spatio-temporal changes in the extent of flood inundation of

Bangladesh are studied. Temporal changes in the extent of the inundated region in theBangladesh are assessed at a resolution of 500m for recent major floods in 2004 and 2007.

Study Area

Bangladesh is taken as study area which is located between Latitude 2027 N and Longitude

8893 E (Figure 4.1). The country is located a floodplain delta of three major river basins: the

Ganges, the Brahmaputra and the Meghna (GBM). About 80% of the annual rainfall of

Bangladesh occurred during monsoon season between June to September. The major cause of the

monsoon flood relies of the intensity, duration and magnitude of the rainfall in the GBM basin.

Every year one fourth to one third of the country is inundated during monsoon season by

overflowing rivers. However, the degree of this inundation sometimes become severe and cause

damage to infrastructures, crops, communication system, and human being. Floods in 2007 and

20004 were one such sever floods in recent years in terms of magnitude and duration.

The spectral signature has been changed for different land use types and therefore, study of thosesignatures is very important to classify the inundated areas from satellite image. A total of nine

categories of land use /land type have been selected to analyze inundation areas (Figure 4.1).

These are 1) single-crop rainfed rice in the Haor area of north-eastern region , 2) single-crop

irrigate rice in the south-west region, 3) double-cropped irrigated rice in north region, 4) triple-

cropped irrigated rice in the Brand area of north-west region, 5) forest area in the Sundarbans, 6)

settlement area of Dhaka city , 7) Kaptai lake, 8) Bay of Bengal ocean, and 9) the Padma river.

In the following sub section, the temporal variations of various indices based on spectral

reflections sensed by satellite on those nine land use categories will be discussed.


35/78

Figure 4.1: Figure Location map of the study area

21


36/78

22

4.3

Data and Satellite Images

4.3.1 MODIS/TERRA Satellite images

This study uses MODIS satellite images acquired by TERRA instrument which can be freely

downloaded through the Earth Observing System Data Gateway (EOS, 2006). The present study

involves an analysis of 8-day composite data of MODIS during 2007 and 2004. The label of this

product is MODIS/TERRA SURFACE REFLECTANCE 8-DAY L3 GLOBAL 500 M SIN

GRID V005. The spatial resolution of this product is approximately 500 m, and atmospheric

correction has already been carried out (Vermote & Vermeulen, 1999). This 8-days average data

is delivered as a composite product called MOD09 which took the best surface spectral-

reflectance within this period with the least effect of aerosols and other atmospheric ingredients.

4.3.2 Flood inundation map based on RADARSAT images

Inundation map produced by Center for Environmental Geographic Information Services

(CEGIS) are used as a reference to evaluate the estimates derived from MODIS data. This map

area produced based on Digital Elevation Model (DEM) data, hydrological data, and

RADARSAT images acquired on 3rd August (DOY 215) using the ScanSAR Narrow B Mode .

Bearing in mind that C-band microwaves can penetrate cloud cover and easily discriminate openwater on the basis of backscatter coefficient data at a high resolution (50 m), it is assumed that

inundation map based on the RADARSAT images reveal the details of the flood distribution at a

satisfactory spatial resolution, even under cloud coverage. The inundated areas in this map were

aggregated within each grid at 500 m resolution to enable comparisons with results derived from

the MODIS data.

4.3.3 Water level data in major rivers of Bangladesh

Three major rivers of the country - the Brahmaputra, the Ganges and the Meghna carry huge

amount of water from the precipitation over their basins. Figure 4.2 shows a time series of daily

averaged water level hydrographs of major rivers in Bangladesh for 2007 and 2004. Water level

data from one gauge station in each major river is used for this comparison. Hydrographs are


37/78

23

4.4

plotted at Bahadurabad sations of the Brahmaputra river, Hardinge Bridge station of the Ganges

river and Bhariab Bazar station of the Meghna river. These stations are very well known gauge

stations, where records of historic floods are available and used in this study. Magnitude of peak

floods in Brahmaputra river was found 0.88m and 0.68m above danger level during floods in

2007 and 2004 respectively. Flood 2007 was more severe in Brahmaputra river basin than Flood2004. The highest water level in Meghna river was found 0.69m and 1.53m above danger level

during floods in 2007 and 2004 respectively. In terms of magnitude, the floodwater level in

2007 was well below in comparison to that of 2004 in the Meghna river basin.

Duration of days above danger level in the Brahmaputra river was found 21day and 15days for

Flood 2007 and Flood 2004 respectively. Flood 2007 was longer than Flood 2004. In terms of

both duration and magnitude Flood 2007 exceeds Flood 2004 in the Meghna river basin. In the

Meghna river, duration of floods above danger level was 37days and 38days during floods in

2007 and 2004. Flood 2007 stays similar days to floods in 2004 in the Meghna river basin.

Methods

4.4.1 Detecting water related surface using MODIS data

In the past, Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water

Index (NDWI) were used to identify water related surface (Rogers & Kearney, 2004). The main

reason of using NDWI is that short-wave infrared (SWIR) is highly sensitive to moisture content

in the soil and the vegetation canopy. A number of studies have been conducted in use of the

spectroscopic characterization of SWIR to detect water content (Gao, 1996; Jackson et al., 2004;

McFeeters, 1996; Rogers & Kearney, 2004; Tong et al., 2004). Xiao et al. (2002b) showed that

NDWI in paddy fields exceeds NDVI derived from SPOT data for the same period of flooding

and rice-planting in eastern Jiangsu Province, China. In recent years, Xiao et al. (2006; 2005)

used anomalies between the Land Surface Water Index (LSWI) and Vegetation Indexes (NDVIor EVI) in an algorithm to estimate the distribution of paddy fields in South China and South and

Southeast Asia. In table 4.1, detail description of the indices derived from MODI S data along

with the band number and solar spectrum has been presented.


38/78

Methodology used in the present study is originally developed by Sakamoto et al. (2007) to

detect the spatio-temporal flood distribution in the Cambodia and Vietnam from the smoothed

indexes of these differences. Sakamoto et al. (2006; 2005) derived a methodology to classify

cropping systems (e.g., double-cropping system in the rainy/dry season, triple-cropping system)

and noted regions where the number of crops per year was increased from two to three over theinterval 2002 to 2003. With some modification, the same approach is applied to the EVI and

LSWI time-series to develop flood inundation maps. Moreover, analysis has been carried out to

determine spatial extents and temporal changes of flood inundation within Bangladesh during

floods in 2007 and 2004.

Table 4.1: MODIS derived indices used to detect spatial and temporal distribution of flood.

Indices Equation

Normalized Difference Vegetation Index (NDVI) RED NIR

RED NIR NDVI

+

=

Normalized Difference Water Index (NDWI)SWIR RED

SWIR RED NDWI

+

=

Enhanced Vegetation Index (EVI)15.76

5.2++

=

BLUE RED NIR

RED NIR EVI

Land Surface Water Index (LSWI)SWIR NIR

SWIR NIR LSWI

+

=

Where, NIR is the reflectance of near infrared (841875 nm, MODIS Band 2), RED is the

reflectance of red (621670 nm, MODIS Band 1), BLUE is the reflectance of blue (459479 nm,

MODIS Band 3) and NIR is reflectance of short-wave infrared (16281652 nm, MODIS Band 6)

of the solar spectrum.

4.4.2 Modified algorithm of flood inundation maps

The algorithm used by the Sakamoto et al. (2007) was modified in this study. A flow-chart of the

method used in this study has been shown in Figure 4.3. The previous algorithms are examined

and some components are excluded from it. In previous algorithm, wavelet based filter is used to

smooth data by removing noise component and interpolate of missing information. This

algorithm creates artificial data and therefore, that algorithm of filtering was not used in this

24


39/78

25

study. The first step is to detect cloud cover pixel from the image. If blue reflectance (Band 3 of

MODIS) is equal to or greater than 0.2 (Thenkabail, Schull, & Turral, 2005; Xiao et al., 2006), it

is considered as cloudy pixel. Using this formula data over cloudy pixel was removed from the

image. Next step is to estimate EVI, LSWI and their difference DVEL for each of the land class

cover types. In this study, discrimination of Water-related pixel and Non-Flood pixel wasconducted in accordance with the pioneering method developed by Xiao et al. (2006; 2005).

EVI, LSWI and DVEL are exclusively used to discriminate Flood, Mixed, Non-Flood and

Water-related pixels. Changes of EVI, LSI and DVEL for different land use types during 2007

are shown in Figure 4.4. If EVI is greater than 0.3, it can be classified as Non-Flood related

pixel. The EVI curve of Forest (the Sundarbans) land use type exhibits a value more than 0.3

during the year except flood season. EVI of permanent water bodies such as River and Sea

land use type are less than 0.05 or even negative value throughout the year. DVEL of River

and Sea land use type have a DVEL value less than 0.05. It can be infrared that water related

pixel should have DVEL less than 0.05. But for Lake land use type, DVEL value is not always

less than 0.05. To overcome this problem, another criterion is set to identify water related pixel.

In such cases, if EVI is less than or equal to 0.05 and LSWI is less than or equal to 0, the pixel

will be identified as Water-related pixel.

After identifying Water-related pixel it is essential to classify whether it is Flood pixel or Long

term water bodies or a Mixed type pixel. Due to moderate resolution (500m) sensor ofMODIS/TERRA, a pixel can be composed of various mixture types of land surfaces. It is

difficult to identify vegetation mixed with water and vegetation completely flooded by water. It

is found from Figure 4.4 that EVI of Sea, Lake or River is below 0.1 and this criterion can

be used for further classification of Water-related pixel. If Water-related pixel has EVI less than

0.1, it will be considered as Flood pixel. If EVI is greater than 0.1 but less than 0.3, Water-

related pixel will be identified as Mixed pixel. Finally, the areas which are inundated through out

the year should be separated from Flood and Mixed pixels. There have been many water bodies

in Bangladesh such as Beels and Haors where water can be found more than 6 months.

Therefore, Water-related pixel which has inundation period more than 120 days will be classified

as Long term water bodies.

Using this proposed methodology, changes of spatial extent with time are analyzed and flood

inundation maps are developed for 2004 and 2007.


40/78

Figure 4.2: Water level hydrographs of major rivers in Bangladesh for (a) year 2007 and (b) year2004. Flood starts earlier in 2004 than in 2007.

26


41/78

Figure 4.3: Flood chart for developing Flood inundation map using MODIS data

27


42/78

Figure 4.4: MODIS derived indices: (a) EVI, (b) LSWI and (c) DVEL for the Seven Land use /Land cover areas shown in Figure 4.1

.

28


43/78

29

4.5

Validation of proposed techniques

The proposed technique used to identify the water surface from MODIS time series data and

validated it with standard product. Figure 4.5(a) shows the distribution of inundated regions onthe 28th July 2007 (DOY 209) determined using the MODIS. In this image, Flooded, Mixed,

Long term water bodies are represented by Blue, Green and white colour. False-colour image of

MOD09 8-day composite data and daily MOD09 data for the same date are shown in Figure

4.5(b). Presence of cloud in the False colour composite image can be visualized as white

coloured pixel. Long term water bodies derived from MODIS data are also shown in the same

Figure 4.5(c). Distribution of inundation area on the 3 rd August of 2007 (DOY 215) using

RADARSAT is shown in Figure 4.5(d). Inundation of RADARSAT images were developed by

used ruled based method. A threshold value more than 0.6 was considered as inundation area for

RADARSAT image. Inundated region is clearly identified in detail without any cloud-cover

effects using RADARSAT images. The influence of cloud cover is considerably reduced in

MOD09 8-day composite data as shown in Figure 4.5(b). Although, MOD 09 production is not

cloud free and need correction for cloud removal.

A comparison of MODIS derived inundation map on 28 th July (DOY 209) with the subsequent

available RADARSAT derived inundation map on 3 rd August (DOY 215) is shown in Figure 4.6

using image crossing. Most of the area in both images shows quite a good match between images

from these two types of satellite and sensors. In south west part, MODIS shows more inundation

areas than RADARSAT. On the other hand, in the north-east region RADARSAT presents more

area as inundated than MODIS image. Figure 4.7 shows scattered plot of the crossing of both the

inundation maps. Value of R 2 was found 0.96 between the inundation map derived from MODIS

and RADARSAT. A very good agreement between these two products shows the high

capabilities of estimating inundating area using MODIS with this algorithm.

Although inundation areas in both images shown high correlation, it is difficult to detect

inundated regions under high-vegetation coverage using moderate resolution optical sensors such

as MODIS. Therefore, it would be difficult to accurately identify the inundated area of flooded

forests or marsh in the MODIS products. Figure 4.8 shows temporal changes of flood and

mixture pixels for four types of paddy fields (single cropped irrigate rice, single cropped rain fed


44/78

30

4.6

rice, double cropped irrigated rice, and triple cropped irrigated rice), forest and settlement areas.

During flood season (May to September), more mixed pixels were found in the triple and double

cropped irrigate areas than other land use types. Mixed pixels may lead to errors of

underestimation of Flood areas in MODIS products than that of RADARSAT products. Hence,

north east region where mostly double and triple irrigated rice grows shows less Flood pixelsthan other regions. It is possible to enhance the estimation accuracy using DEM data or higher-

resolution data (Brivio, Colombo, Maggi, & Tomasoni, 2002; Wang, Colby, & Mulcahy, 2002)

which need to be investigated in future. Although there is a fundamental overestimation problem

resulting from mixed pixel effects, it can be assumed that temporal MODIS products provide

useful criteria for determining flood inundation.

Results and discussions

4.6.1 Spatial extents of floods in Bangladesh

The spatial extents of flood inundation were changed with the progress of flood and can be

visualized through displaying successive maps. Changes of the extent of the flood inundation in

Bangladesh during floods in 2007 and 2004 were studied. Due to limitation of space, 3 images of

each major flood were used for analysis. Figure 4.9 shows the estimated flood area for the 20 th

July [DOY 201] to 5 th August [DOY 217] during floods in 2007 and 2004 respectively. In 2004,

flood started following a rapid increase in water level from 20 th July (Figure 4.2) and reached its

largest extent during 28 th July [DOY 209]. Flood 2007 started two weeks later than floods in

2004, therefore the increase of the extent of water level from 28 th July [DOY 209] and reaches its

largest extend on 5 th August [DOY 217]. Upon visualization, it is clear that the spatio-temporal

distribution of the inundated area varies from year to year. The scale of the inundation varies

with the amount of water carried by the rivers.

It is also possible to detect maximum extent of the flooded area by overlaying a series of images

during the flood (Sheng & Gong, 2001). This inundation map can be useful to create flood

vulnerability maps and flood risk zones. Figure 4.10 shows maximum area of food inundation

during floods in 2004 and 2007 respectively. Blue, green, and white colours represent areas of

Flood, Mixture, and Long-term water bodies, respectively. It is also evident from this map that

the extent of flooding varies from year to year. The areas which are common for both major

floods should be classified as the most vulnerable areas.


45/78

Figure 4.5: Spatial comparison of (a) MODIS derived inundation map of 29 July 2007, (b)

MODIS 8-day false colour composite map (RGB=Band 6,Band2,Band1) of 29 July 2007, (c)Long term water bodies in 2007 using MODIS data, (d) Inundation map using RADARSATimage on August 03

31


46/78

Figure 4.6: Overlay of MODIS derived inundation map on DOY 209 with the nearest availableRADARSAT derived inundation map on DOY 215. Inundation area derived from MODIS

includes Flood pixels and pixels of Long term water bodies.

Figure 4.7: Correlation of inundation area determining from MODIS with that of RADARSAT.Correlation coefficient R 2 is found very high of about 0.96.

32


47/78

Figure 4.8: Temporal Changes of Flood and Mixture pixels for four types of paddy fields (Singlecropped irrigate rice, Single cropped rain fed rice, Double cropped irrigated rice, and Triple

cropped irrigated rice), Forest and Settlement areas

33


48/78

Figure 4.9: Spatial distribution of flood inundated area during July and August months of 2007and 2004. Flood pixels are in blue colour, Mixture pixels are in green colour, Long term water

bodies are in white colours and other areas are in white colour.

34


49/78

Figure 4.10: Flood inundation map of Bangladesh using MODIS images for (a) Year 2004 and(b) Year 2007. Flood pixels are in blue colour, Mixture pixels are in green colour, Long term

water bodies are in white colours and other areas are in white colour.

4.6.2 Temporal characteristics of floods

The start date, end data and duration of flood varies for each flood. Time series MODIS data can

be effectively used to determine these dates and produce spatially distributed maps of thesedates. Figure 4.11 shows estimates of the start dates, end dates, and duration of inundation

during floods in 2007 and 2004. As these dates are based on images of 8-days average value, it

is not possible to represent exact start and end dates rather it represent start and end weeks of

flood.

The start dates for floods in 2004 are earlier floods in 2007. Date of crossing danger level at

Bahdurabad of the Brahmaputra river was found 27 th of July and 11 th of July during floods in

2007 and 2004 respectively (Figure 4.2). This similar pattern has been found in the spatial

distribution of flood inundation maps of 2007 and 2004 (Figure 4.11). The duration of floods at

the Bahadurabad in the Brahmaputra river was found 21 days and 15 days during floods in 2007

and 2004 respectively. The pixel information of the flood duration maps (Figure 4.11) shows

more red colour for floods in 2004 than floods in 2007.

35


50/78

Figure 4.11: Spatial distribution of (a) start date, (b) end date and (c) duration of flood for 2004and 2007

36


51/78

37

4.7

At Bhiarab bazaar station in the Meghna river, date of crossing danger level was 30 th of July and

11 th of July during floods in 2007 and 2004 respectively (Figure 4.2). When focusing on the

Flood pixels in Figure 4.11, the timing of the appearance of Flood pixels in 2004 and 2007 is

obviously shows the similar characteristics. Duration of floods above danger level at the Bhairab bazaar station in the Meghna river was 37 days and 38 days during floods in 2007 and 2004

respectively (Figure 4.2). In the flood duration map, similar features were found for this river

basin. The colour of inundation pixels is more reddish for floods in 2004 than that of 2007.

Conclusions

This study modifies a methodology which was developed by Sakamoto et al. in order to detect

spatial extents and temporal changes of flood inundation of Bangladesh during monsoon season.

Using this modified methodology, MODIS satellite images were used develop flood inundation

maps for floods in 2007 and 2004. This low resolution (500m) MODIS based maps area

compared with subsequent flood inundation maps based on high resolution (50m) RADARSAT

satellite images. MODIS estimates show strong correlation with the inundation areas derived

from RADARSAT with R 2 values of 0.96. Considering this fact, the flood maps derived from

MODIS images shows ability to flood characteristics and behavior. Such inundation maps will

be useful for integrating water resources management and the maintenance of ecosystems of

wetlands of Bangladesh.


52/78

38

5.1

Chapter 5: Floods around Dhaka city

Hydrologic condition of Dhaka city

Dhaka city is surrounded by four major river systems as shown in Figure 5.1. The south of

Dhaka city is surrounded by the Buriganga river. The western part of Dhaka is bounded by the

Truag river which is connected by a small Tongi Khal on the north. The eastern part of Dhaka is

bounded by the Balu river which is also hydrologically connected with Tongi Khal. In terms of

flood protection works, Dhaka city can be divided into two parts: Dhaka west and Dhaka east.

The area of Dhaka west is 243 km 2 and is surrounded by embankment and embankment cum

road, where as the area of Dhaka east is 119 km2 and consists of unprotected lowlands within the

floodplain of the Balu river (JICA, 1987). The most of the areas of Greater Dhaka city are urbanareas including residential areas, large commercial complex, offices, schools, hospitals and small

garments & other industries. The population of Greater Dhaka city was 4.47 million in 1990,

which is projected to increase to 8.59 million in 2010 (JICA, 1991). Dhaka is not only the

nations capital but also has potential to become a mega city by 2010. Condition of Dhaka city

during flood period always draws special attention due to its strategic importance.

Hence, a special emphasis was given to reveal the hydrologic characteristics of the Dhaka.

Analysis has been carried out on the flood situation of Dhaka city during the flood in 2007.

Hydrologic aspects of Flood 2007 of the surrounding rivers of Dhaka city are also compared

with the major floods in recent past.


53/78

Figure 5.1: Greater Dhaka city area and surrounding river systems

39


54/78

40

5.2 Hydrologic Characteristics of Major Floods in Dhaka City

Water level dat

Flood2007 Report

Documents

Transcript of Flood2007 Report