You are on page 1of 26

Using Multi-index Model to Assess the Risk From Flash

Flood: a Geospatial Approach on Sunamganj Haor


(Wetland) Region, Bangladesh
Gourab Saha 
(

gourab32@student.sust.edu
)
Shahjalal University of Science and Technology
Md. Najmul Kabir 
Shahjalal University of Science and Technology
Afrin Khandaker 
Shahjalal University of Science and Technology

Research Article

Keywords: Haor, Flash Flood, Multi-Index Model, Risk Assessment

Posted Date: September 27th, 2022

DOI: https://doi.org/10.21203/rs.3.rs-2093330/v1

License:


This work is licensed under a Creative Commons Attribution 4.0 International
License.
 
Read Full
License

Page 1/26
Abstract
Heavy downpours in Meghalaya and Assam regions are causing flash floods in Bangladesh's northeastern Haor
region at regular intervals. As high volume of water enters into the Surma river through the downstream river
network, the Haor region faces heavy damage almost every year. The major focus of the study was to assess the
risk of flash floods in the greater Sylhet region. Literature review of past flood scenarios have shown that
Sunamganj District directly faces the wrath of flash floods each year and witnesses tremendous damage. A
geospatial multi-index model was developed which considered four major indices (hazard, exposure, sensitivity, and
resilience) for this risk assessment. The model systematically accounted for various flood risk indicators related to
the economic, social, and physical environment of the Sunamganj District. Geostatistical methods were used to
calculate the risk from the indicators and Principal Component Analysis (PCA) was used for composite analysis.
The result showed that the high- and very high-risk zones occupied almost 40% (~ 1452.51 Km2), while the very low-
and low-risk zones covered about 45% (~ 1554.66 Km2) of the Sunamganj District. Based on the risk rank,
Dharampasha upazila had high percentage (~ 60%) which followed by Sulla, Derai, Jamalganj, Daksin Sunamganj,
and Tahirpur upazilas respectively. The proposed methods will help the researchers to identify the causes of flash
floods and the results will be helpful for policymakers and developers to take steps for flash flood risks reduction.

1. Introduction
More than 90% of disasters such as floods, droughts, hurricanes, and heat waves were caused by extreme weather
between 1995 and 2015 in the world (UNISDR, 2015). Floods have become the most frequent and most devastating
natural disaster. Among all these disaster, flash floods are the most severe natural disaster resulting in more than
5,000 deaths annually from a global perspective, along with social and economic losses (Bisht et al., 2018).

A flash flood is a hydrological event that occurs within a few hours of heavy rain, rapid snowmelt, a sudden glacial
lake outburst, embankment failure, or the rapid break up of an ice block due to rapid temperature rise (Hossain et al.,
2008). Flash flood usually happens in small areas. It is more common in the northeastern and southeastern regions
of Bangladesh during the pre-monsoon months of April and May because of the surrounding hilly Assam and
Meghalaya regions of India and the presence of the Haor basin (Mondal et al., 2021). The northeastern area
becomes vulnerable since the upstream basin topography is relatively steep and the concentration time of the basin
is relatively short (BWDB 2014; WMO 2003). The main cause of flash floods is heavy rains in the upstream area and
the Haor region, which faces the highest rainfall with an average annual rainfall ranging from 2200 mm in the
northwest to 5800 mm in the northeast and about 12,000 mm in India's adjacent state of Meghalaya (Kamal et al.,
2018). The flash flooding scenario causes the following percentile of damage in the month of March (75%), April
(70 to 90%), and May (15–40%) (Mondal et al., 2021).

During the pre-monsoon season (January to May), the Haor lands’ 80% coverage is occupied by Boro rice
production which contributes to 18% of the total rice production of the country (Mondal et al., 2021). In Haor areas,
the flash flood has a greater impact on agricultural production which resulted greater damage in 2000, 2002, 2007,
2010, 2017, 2019, and 2020 because of the maximum duration of the flood (Abedin & Khatun, 2020). Early flash
flood in 2017 caused the most damage in the Sunamganj district (Kumar Biswas et al., 2020). Heavy downpours in
the Meghalaya and Assam regions result in the unexpected rise of the water level and water flow speed in streams
and rivers that bring a large number of debris, boulders, uprooted trees, obliteration of infrastructures and
constructed buildings simultaneously and cause life loss, property damage, and infrastructure destruction in the
majority of cases (Kumar Biswas et al., 2020). Flash flooding accelerates water quality degradation due to

Page 2/26
excessive oil leakage from number of boats. Flash flood victims face food scarcity during the disaster and post-
disaster period which is a major impediment to long-term economic development (Kamruzzaman & Shaw, 2018).

Risk assessment approaches are designed to evaluate the loss and flood impact (Gigović et al., 2017) and develop
risk maps for land use and infrastructure development (Komolafe et al., 2019; Zeleňáková, M.,2019). The index
system techniques should be combined with GIS-based methodologies for successful monitoring and evaluation
(Y.K. Qiao & Peng, 2017) because the system considers all components to measure the flash flood risk. In addition,
GIS-based techniques are used in collaboration to produce spatial data (Abdelkarim & Gaber, 2019), whereas a
multi-index approach is used to assess flash flood risk (Zhao, J.W.,2017), which requires various indicators and a
large amount of data (Zhang et al., 2020). For this reason, the study used an index system method and a GIS-based
approach to build a spatial multi-index model for flash flood risk assessment of the hoar region.

The study aims to assess the risk of flash floods and identify the risk zone in the Haor region of the Sunamganj
district using a multi-index approach. Specifically, the study identifies the flash flood risk determination and various
risk zones by developing a map.

2. Study Area
Sunamganj, Habiganj, Netrokona, Kishoreganj, Sylhet, Moulavibazar and Brahmanbaria are the seven districts that
make up the core northeast Haor area, which spans 1.99 million ha area and 19.37 million people (CEGIS, 2012).
These districts contain about 373 Haors, covering an area of about 85,900 ha (859 km2), or about 43% of
Bangladesh's total Haor zone (CEGIS, 2012) and 23 transboundary rivers have been identified that flow from India
(Hossain et al., 2008). The maximum topography of Haor region is largely flat with maximum land surface
elevation less than 10 meters. Due to the physiography, floodwater is moving into Haor region within 3-to-6-hour
precipitation. The district of Sunamganj has been chosen as study area due to continuous flash flood occurrence
over years. Previous studies showed that Sunamganj District faced most agricultural land damages due to 2017,
2019, 2020 flash flood. The district is covered by major Haor systems of about 95 Haors in the north-eastern region
of Bangladesh (Mondal et al., 2021) and the physiography is accessible to the heavy rainfall and flood water from
the upstream of Meghalaya hills which contributes to the flash flood occurrence nearly every year during pre-
monsoon season (March-May).

Page 3/26
Table 1
Major Flood History of Sunamganj District
Year Flood Water entered the district Inundation occurred till

1999 03 May 06 May

2000 28 April 30 April

2002 14 April 18 April

2004 13 April 15 April

2007 25 April 28 April

2010 17 April 19 April

2017 28 March 05 April

2019 04 July 10 July

2020 26 June 28 June

(Source: Abedin & Khatun, 2020)

3. Methodology
To assess the flash flood risk in the Sunamganj District, a GIS based multi-index model was developed by this
study. The multi-index system can be explained into the object layer, the index layer, and the indicator layer. The
Sunamganj District flood risk assessment was the object layer; the index layer included the hazard index (H),
Sensitivity index (V), exposure index (E) and Resilience index (R); and the indicator layer included 16 flood risk
indicators. Data representing the 16 indicators were collected and pre-processed in the GIS environment.

Then, the data for each flood risk indicator were incorporated into the GIS. Finally, a risk distribution map of the
Sunamganj District was generated by the combination of four indexes (hazard, exposure, sensitivity and resilience)
and PCA (Principal Component Analysis) method.

3.1 Data Type and Source


The study is based on the secondary data and satellite images, water level data, rainfall data, demographic data
were required for the outcome.

Page 4/26
Table 2
Data Source and Type
Index Layer Parameters Data Types Temporal Source Year
Resolution

Hazar Index Flash Flood Water Level Observed 3 BWDB 2002, 2007, 2017,
hourly 2019, 2020

Exposure Precipitation Rainfall Daily BWDB 2002, 2007, 2017,


Index 2019, 2020

Elevation ALOS Word 3D – 30m   JAXA  

Drainage
Pattern

Soil Moisture

Slope

Sensitivity Demography Population and   BBS 2011


Index Housing Census

Geomorphology Sentinel 2A 10 Days ESA 2020

Wetland

Vegetation

Build Up Area

Agricultural
Land

Resilience Literacy Population and   BBS 2011


Housing Census
House
Condition

Economy Economic Census     2020

3.3 Hazard Index Analysis


The intensity of flash flood in a given area is proportional to the hazard level. Water level data was used for flood
depth extraction of Sunamganj district. BWDB 3-hourly water level data of 10 stations of the study area of five
years (2002, 2007, 2017, 2019, 2020) during flash floods were taken into consideration for the flood depth analysis.
The data were interpolated by Kriging method using “Kriging” tool in ArcGIS software for weights of these data.
Then Decay Coefficient Simulation Model was applied by using ALOS 30m DEM data for finding out the flood
depth.

3.4 Exposure Index Analysis


This analysis is used to identify life and property which are exposed to the flood hazard. Five parameters were
included to analyze the exposure index of this study.

3.4.1 Elevation

Page 5/26
The topographic profile was prepared based on ALOS Word 3D DEM of 30-meter spatial resolution and the
interpolate line of “3D analyst” tool of ArcMap was used for preparing the topographic profile where 10 cross
sections were taken. Then the “profile graph” tool was applied to create topo profile graph that revealed the
elevation and distance of cross section line.
3.4.2 Slope
In this study slope map was created directly from the DEM in ArcGIS environment by using newer implementation
“Slope” tool.

3.4.3 Drainage Density (DD)


This describes the ratio of the total length of rivers in each area to the area's size which means higher drainage
density indicates more surface runoff that arises flood risk (Elkhrachy, 2015).

ALOS DEM were downloaded from the JAXA website for the density calculation. The study area consisted of two
mosaic elevation models, “N024E091” and “N024E090” which were converted into a single raster and bounded by
the Area of Interest (AOI), the boundary of the study area. The “Hydrology” tool and Line Density method were used
for the drainage density in ArcGIS 10.8.

3.4.4 Rainfall Distribution


Rainfall data is considered as significant components for flood risk assessments by academics worldwide
(Mahmoud & Gan, 2018). BWDB daily rainfall data from 10 stations of Sunamganj district of five years (2002, 2007,
2017, 2019, 2020) during flash floods were taken into consideration.

10 days rainfall data during the flash flood of 2002, 2007, 2017, 2019 and 2020 were analyzed to understand the
rainfall pattern of Sunamganj district. A standard deviation value was calculated for frequency and uncertainty of
the rainfall pattern condition of this specific time. The standard deviation equation was:

2
∑ (x − μ)

σ =
n− 1

Where, σ = Standard Deviation; x = Each value; µ = Variables Average Value; n = Number of Values in the Sample

The standard deviation data were imported in ArcMap for interpolation by using “Kriging” tool. Finally, the rainfall
pattern map of Sunamganj District was produced.

3.4.5 Topographic Wetness Index (TWI)


The essential factor TWI for flash flood research quantifies topographic control on hydrological processes. It is
related to the soil moisture that influences rapid runoff and flash floods.

ALOS World 3D DEM data were projected for flow direction and flow accumulation. The slope of the area was
calculated. As the border pixels have zero flow accumulation value, value 1 was added with flow accumulation
since. If the value 1 wasn’t added, the watershed showed division which had no flow accumulation would have a
zero value and (0) will be undefined.

F lowacuumulation
TWI =
tanslope

Page 6/26
3.5 Sensitivity Index Analysis
Sensitivity is related to the degree of threat to a particular population or the capacity of a system, which can suffer
and respond harmfully during the occurrence of any kind of hazardous event. It is useful for risk reduction, hazard
and disaster management and climate change areas. Six indicators were included to assess the sensitivity index.

3.5.1 Geomorphology
Blue, Green, Red and Infrared these four bands of sentinel 2A images and ALOS world 3D DEM data were used for
geomorphology analysis. The Blue and Green bands were composite for DOP and Water Attenuation to calculate
bathymetry. NDVI, Maximum Likelihood supervised classification images, ISO Cluster unsupervised classification
were done by DEM, Bathymetry were conducted for geomorphological map of Sunamganj District.

3.5.2 Water Ratio Index (WRI)


Water Ratio Index was calculated using the following formula (Sajjad et al., 2021):

GREEN + RED
WRI =
NIR + SWIR

Where, GREEN= Green Band, RED= Red Band, NIR= Near Infrared Band, SWIR= Short-wave Infrared Band.

For WRI analysis Sentinel 2A images were used for WRI equation and the specific bands for Sentinel 2A for the
equation was:

Band3 + Band4
WRI =
Band8 + Band12

3.5.3 Normalize Difference Vegetation Index (NDVI)


According to this formula, the density of vegetation at a certain point of the image is equal to the difference in the
intensities of the reflected light in the red and infrared range divided by the sum of these intensities. This Index
defines values from − 1.0 to + 1.0 which is basically representing the greens, where negative values are mainly
formed from clouds, water, and snow. For NDVI calculation, Sentinel 2A images were used and the general equation
was:

NIR − RED
NDVI =
NIR + RED

Where, RED= Red Band, NIR= Near Infrared Band and specifically for Sentinel 2A, the bands were:

Band8 − Band4
NDVI =
Band8 + Band4

3.5.4 Normalized Difference Built-up Index (NDBI)


For NDBI, the value ranges from − 0.1 to 0.1 for barren rocks, sand and snow; from 0.2 to 0.5 for shrubs, grasslands
and senescing crops, from 0.6 to 0.9 for dense vegetation and tropical rainforest; -1 for deep water. The following
equation was used for NDBI (Khokhar et al., 2021):

Page 7/26
SWIR − NIR
NDBI =
SWIR + NIR

Where, NIR= Near Infrared Band, SWIR= Short-wave Infrared Band. For Sentinel 2A, the specific bands were:

Band11 − Band4
NDBI =
Band11 + Band4

3.5.5 Agricultural Land


Maximum Likelihood Supervised Classification method was applied for land use land cover (LULC) map of the
study area to extract agricultural land. Sentinel-2A, Level 1C (ID = L1C_T46RCN_A028444_20201202T043140 and
ID = L1C_T45RZH_A028444_20201222T0432140) acquired within the time of December 2020, were downloaded
from the United States Geological Survey (EarthExplorer, n.d.). Two images were used for mosaics and 10 bands
(2–8, 8A,11 and 12) of these two images were composited and the Sunamganj shapefile was clipped from the map.
1000 random points were taken and points were classified into five classes (vegetation, water body, barren land,
agricultural land, and built-up area) for creating proper signature file. Among all classes, agricultural land was
extracted and compared with DEM data for understanding the elevation of Agricultural land.

3.6 Resilience Index


Literacy rate, housing structure, poverty rate – these 3 parameters were taken under consideration for resilience
index. These data were collected from BBS 2011 “Community Book” of Sunamganj district and the “Poverty Maps
of Bangladesh 2016” by the collaboration of the Bangladesh Bureau of Statistics (BBS) and the World Food
Program (WFP). Then these data were input into the upazila shapefile of Sunamganj district using ArcGIS. After
that “conversion” tool of ArcGIS 10.8 was used and raster images were produced for each parameter for further PCA
analysis.

3.7 Principal Component Analysis (PCA)


The data in the input bands are transformed from the input multivariate attribute space to a new multivariate
attribute space whose axes are rotated with respect to the original space using the Principal Components tool after
the indices are analyzed. The axes (attributes) in the new space are uncorrelated. The main reason to transform the
data in a principal component analysis is to compress data by eliminating redundancy (Azotea et al., 2015; Gómez-
Palacios et al., 2017; Zeng et al., 2012).

This was run using ArcGIS 10.8 to figure out the principal components of the indices. By using the raster calculator
tool, the percent of eigenvalues were taking into account of the raster with continuous data to find out the total sum
of products of the percent eigenvalues and raster. Hence, the weighted value of the parameters gave output to the
composite indices map.

After conducting PCA for Exposure index, Sensitivity index and Resilience index, the Eigen values for each
parameter were received. The Eigen value of indices were given bellow:

Page 8/26
Table 3
Summary of PCA Results
PC Layer Eigen Value % Of Eigen Values Accumulative of Eigen Values

Exposure Index

Elevation 1.0739 34.2332 34.2332

Slope 0.89236 28.4459 62.6791

Drainage Density 0.83548 26.6329 89.312

Rainfall Pattern 0.20311 6.4746 95.7866

TWI 0.13217 4.2134 100

Sensitivity Index

Geomorphology 3.35414 55.0629 55.0629

WRI 0.83307 13.6761 68.739

NDVI 0.64827 10.6422 79.3813

NDBI 0.45949 7.5432 86.9244

Agricultural Land 0.32352 5.3111 92.2355

Population Density 0.30005 4.9258 97.1613

Dependent Population 0.17292 2.8387 100

Resilience Index

Literacy Rate 0.68241 53.5042 53.5042

Housing Structure 0.50092 39.2744 92.7786

Poverty 0.0921 7.2214 100

3.8 Flash Flood Risk Analysis


The Exposure Index map, Sensitivity Index map and Resilience index map were produced by PCA and the Hazard
Index map was produced from the flood depth data. Flash Flood Risk (FFR) of Sunamganj District was redefined by
calculating the risk map with the assistance of “raster calculator” tool in ArcGIS. The applied equation was given
below:

Flash Flood Risk (FFR) = {Hazard * (Exposure + Sensitivity)} – Resilience

From the previous studies it was found that most of the researchers developed the flood susceptibility class
boundaries based on their own expert opinion and there was no specific rule for classification. In this study, the
resultant flood susceptibility map was classified into very low risk, low risk, moderate risk, high risk, very high risk
zones based on the natural braking method in the ArcGIS environment(Das, 2019; Mahmoud & Gan, 2018; Ullah &
Zhang, 2020).

4. Result And Discussion


Page 9/26
Four types of indexes were studied for the risk assessment. The results of the indexes are:

4.1 Hazard Index Assessment


The flood depth of the years of 2002, 2007, 2017, 2019 and 2020 were analyzed and the result showed that an
average flood depth of 3.61 to 8.74 meter was considered as high-risk zone found in the western part of the
upazilas. Tahirpur, Dharmapasha, Jamalganj, Derai, Sulla upazilas had the majority area in high-risk and very high-
risk zone because of the presence of large Haors densely and comparatively lowland area which made the area
highly vulnerable to flash floods.

4.2 Exposure Index Assessment


The exposure index was calculated by studying elevation, slope, drainage density, average rainfall, TWI. The
undulating topography of Sunamganj showed higher elevation (up to 10 meters) in the northern part in the
Meghalaya border of India but land elevation below 8 meters was found in rest of the area. The elevation and flood
had inverse relationship which resulted higher elevation had lower exposure. The elevation map marked that most
of the Sunamganj region was below 5 meters with very low gradient slope (0 to 3.03 degrees) though the slope
gradient varied from 0 to 55.3. The surrounding hilly area with the low gradient slopes such as flat terrains in
northwest appearance was highly vulnerable to flash flood occurrences compared to higher gradient slopes. Again,
the Surma river flowing through the study area had higher drainage density (> 2 km2) which indicated more surface
runoff associated with high-risk and very high-risk of the flash flood zones. Rainfall occurred more frequently in
current years was found by analyzing 5-year rainfall distribution map. The higher standard deviation values of
northern part showed heavy downpour due to the presence of cyclonic circulation over Sub-Himalayan West Bengal
which made the area high-risk and very high-risk zones by orographic rainfall phenomenon. The higher topographic
wetness index (TWI) areas were more susceptible to flash floods than lower TWI areas because higher TWI was
found in higher density of deep depression. The TWI map illustrated the majority study area had higher TWI values
turned the area into high-risk and very high-risk zones because the higher TWI value represented the rapid runoff
and flash flood.

The spatial distribution of Sunamganj’s flash flood risk exposure was high in the northwestern part and low in the
northeastern part (Fig. 5). The high drainage density, low elevation, low gradient slope, higher TWI were found in
high exposure area. Dowarabazar and Bishwambarpur upazilas showed a very low-risk exposure due to the
presence of higher elevation and vast portion of Dharmapasha, Tahirpur, Derai, Jamalganj areas indicated high-risk
and very high-risk zones for flash flooding.

Page 10/26
Table 4
Correlation Metrix of Exposure Index
PC Layer Elevation Slope Drainage Density Rainfall Pattern TWI

Elevation 1        

Slope 0.38497** 1      

Drainage Density 0.25987** 0.06408** 1    

Rainfall Pattern -0.3416 -0.00056 -0.0215 1  

TWI 0.3201** 0.5772** 0.07131* 0.0099 1

Here, “**” = Highly Significant, “*” = Significant

4.3 Sensitivity Index Assessment


Sensitivity index assessment was discussed with the help of geomorphology, WRI, NDVI, NDBI, agricultural land,
population density and percentage of dependent population. Six geomorphological features such as deep
depression, shallow depression, floodplain, active channel, low land and plain land were studied for
geomorphological map in this study. The map revealed that high density of deep depression, shallow depression,
floodplain and active channel were responsible for high-risk zone. Because heavy rainfall in active river or channel
resulted rapid runoff, sometimes overflowed on the riverbank and onrush of upstream river floods were the reasons
of inundating more areas. The Sunamganj had crisscrossed by numerous rivers coming down from the hills of
India and considerable number of tributaries and distributaries carried out a huge amount of runoff water which
caused flash floods frequently. Numerous distributaries, tributaries, rivers and Haors in most of the unions of
Sunamganj showed high mean WRI value which indicated the deep water. The higher WRI occurred when the
floodwater overflowed very quickly to the surrounding areas which made the area more sensitive to flash floods.
Tahirpur upazila’s three unions named Dakshin Sreepur, Uttar Sreepur and Tahirpur and other three upazlias named
Dharmapash, Derai, and Jamalganj showed higher mean WRI value. Higher NDVI value indicated dense vegetation
which reduced the intensity or speed runoff water of flash floods. But higher density of waterbody, shallow and
deep depression resulted lower NDVI. According to the topographic profile, the northeastern and the eastern part of
Sunamganj had the upland than the western part and this upland showed high NDVI value. Based on the NDVI map,
low land and depression areas with less NDVI value was in very high-risk zone for flash flood. A positive correlation
between NDBI and flash flood was observed where the value ranged from 0.0873 to 0.996 which indicated built-up
area. The downstream and adjacent to the river network area showed higher NDBI value because the riverbank area
had been filling with settlement and the high NDBI containing areas were exposed to the flood risk. According to the
agricultural map, majority agricultural land was lowland and situated beside shallow, deep depression and this
agricultural land submerged during monsoon. This low elevated agricultural land in Sunamganj was most
vulnerable to flash floods because of the rapid flood water movement from the upland to lowland which resulted
middle and western parts of Sunamganj very high-risk zone for potential damage of agricultural land.

The sex ratio of the study area was almost equal which meant more than 50% of population could have high risk
because women and children could face more vulnerability during flood occurrence. The study found that flash
floods had greater effects on densely populated areas than sparsely populated areas and Chatak, Sunamganj
Sadar and Dowarabazar upazilas showed higher population density. Lastly, dependent population refers age group
from 0 to 14 and above 65 years and dependent population is considered most vulnerable for flash flood. The study

Page 11/26
area showed the percentage of the dependent population valued from 42.20–54.30% and the graph revealed that
the northern upazilas had higher dependent population.

The sensitivity index was geomorphology and WRI dominating. The map showed that western part of Sunamganj
was more sensitive to flash flood than eastern part (Fig. 8). The higher WRI value containing areas such as
Dharmapash, Tahirpur, Jamalganj and Sulla were in high-risk and very high-risk zones to flash flood. Because the
Haor orientation accelerated more sensitivity to flash flood in these upazilas.

Table 5
Correlation Metrix of Sensitivity Index
PC Layer Geomorphology WRI NDVI NDBI Agricultural PD DP
Land

Geomorphology 1            

WRI 0.23995** 1          

NDVI 0.14607* 0.83035** 1        

NDBI -0.24454 -0.713 -0.61216** 1      

Agricultural -0.18487 -0.48125 -0.43057 0.33811** 1    


Land

Population -0.13532 -0.37563 -0.35624 0.27354** 0.3286** 1  


Density (PD)

Depending -0.10259 -0.16632 -0.15509 0.1602* 0.16503* 0.20009** 1


Population (DP)

Here, “**” = Highly Significant, “*” = Significant

4.4 Resilience Index Assessment


Literacy rate, poverty rate and housing structure of the study area were taken under consideration for the
assessment. Lack of educational institutions, significant amount of inundated area during monsoon resulted very
low literacy rate of about 35% in average. The upazilas of Tahirpur, Dharmapasha, Jamalganj had the lowest
literacy rates valued 30.4%, 29.2% and 32.5% respectively. The poverty rate scenario of the study area showed that
Tahirpur, Jamalganj, Derai, and Sulla upazilas had the highest rates of poverty at over 22.61% and specially the
Sulla union had highest percentage of 60.9%. The overall poverty condition indicated that 26% inhabitants of
Sunamganj district lived just below of the upper poverty line.

The housing structure of the area were made of four types such as Pucca, Semi-pucca, Kancha and Jhupi whereas
three of the structures were considered highly vulnerable to flash flood except Pucca construction. According to BBS
report of 2011, more than 60% housing structure was Kancha and Jhupi of this district. Specially Sulla,
Dharmapasha, Tahirpur, Jamalganj, Derai, Dakshin Surma had above 80% rate of vulnerable housing structure but
Pucca and Semi-puccha constructed house percentile were very low in this whole district and Jagannathpur and
Chhatak had higher percentage of Pucca house of about 17.1% and 16.1% respectively and other upazilas showed
the existence of Pucca house less than 8%.

The lower literacy rate in mass level resulted unconsciousness of the severity of flash flood and less willingness of
adopting precautions against flash floods. The significant amount of people below poverty line showed their

Page 12/26
inability to fight with flash floods in constructive and non-constructive level and lastly, the vulnerable housing
structure made the weaker resilience to flash flood.

Table 6
Correlation Metrix of Resilience Index
PC Layer Literacy Rate Poverty Housing Structure

Literacy Rate 1    

Poverty -0.15031 1  

Housing Structure 0.3564** -0.07734 1

Here, “**” = Highly Significant, “*” = Significant

The resilience index map showed that Jamalganj, Tahirpur, Dharamapasha and Sulla upazilas showed very low
resilience which made the area high-risk and very high-risk zones for flash flood (Fig. 9). Because these upazilas
were Haor based region and the economic system and ecology resulted less availability of basic right of proper
shelter, food and educational facilities.

4.5 Flash Flood Risk Assessment


The flash flood risk map was generated by integration of four indexes map (hazard, exposure, sensitivity, resilience).
The study revealed that 39% area of the Sunamganj district had very high probability of flash flood while moderate,
low, and very low vulnerable areas were 18.25%, 4.13%, and 38.14% respectively. The risk assessment map
indicated that western part of the study area was highly susceptible to flash flood than eastern part (Fig. 10).
Because the western part had characteristics of very low elevation, lower degree of slope, high drainage density, low
economic status and poor housing structure. The geomorphological feature and vulnerability of human’s social
conditions caused the higher risk in western zone. Besides, the flash flood risk map showed that the zone of high
density of deep and shallow depressions existing places faced the highest number of historical floods.

Page 13/26
Table 7
Area wise Flash Flood Risk
Upazila Very Low Low Moderate High Very high

Bishwambarpur 141.89 3.65 18.39 15.84 9.84

Chhatak 295.89 19.45 61.07 46.23 8.46

Dakshin Sunamganj 61.44 14.66 66.79 109.60 55.36

Derai 65.58 18.71 91.76 113.46 117.14

Dharampasha 72.11 11.17 118.98 137.16 156.37

Dowarabazar 207.84 5.19 30.37 39.90 9.72

Jagannathpur 212.04 19.27 67.84 44.36 24.78

Jamalganj 78.85 18.60 54.23 87.32 93.52

Sulla 27.85 13.29 65.32 88.14 64.66

Sunamganj Sadar 147.70 20.55 42.11 49.19 18.64

Tahirpur 91.65 7.27 54.51 59.17 103.64

Total Area 1402.85 151.81 671.36 790.37 662.14

Area (%) 38.14 4.13 18.25 21.49 18.00

The table showed that Dharmapasha, Tahirpur, Jamalganj, Derai, Sulla and Dakshin Sunamganj upazilas showed
the majority portion of the area at high risk and very high-risk zones due to low land and high density of deep and
shallow depression.

Table 8
Area wise Flash Flood Risk Rank
Risk Rank Upazila Area (%)

1 Dharampasha 59.21

2 Sulla 58.94

3 Derai 56.71

4 Jamalganj 54.38

5 Dakshin Sunamganj 53.58

6 Tahirpur 51.48

7 Sunamganj Sadar 24.38

8 Jagannathpur 18.77

9 Dowarabazar 16.94

10 Bishwambarpur 13.54

11 Chhatak 12.69

Page 14/26
It was found that more than 50% area of these upazilas (Dharamapasha, Sulla, Derai, Jamalganj, Dakshin
Sunamganj and Tahirpur) were in risk zone. Comparatively lower appearance of deep and shallow depression,
higher elevation made Bishwambarpur, Chattak, Dowarabzar, Jaganathpur and Sunamganj Sadar upazilas less
vulnerable to flash flood.

5. Conclusion
The geographical setting of Sunamganj district resulted frequent flash flood. To assess the flash flood risk at
Sunamganj district four indices were used and Principal Composite Analysis (PCA) was applied to composite all the
indicators. The key findings of the study were:

1. The hazard index assessment showed that the western part of the Sunamganj District was mostly at high and
very high risk of flash flooding.
2. The exposure index exhibited that the lower elevation land, lower slope, and high drainage density areas were
responsible for high risk and very high-risk zones.
3. The sensitivity index showed the presence of deep and shallow depressions was high, the WRI value was high
and the percentage of lower agricultural land was high. All these described portions were basically in high and
very high-risk zones for flash floods.
4. The resilience index explained that the eastern Sunamganj part had higher elevation and little presence of deep
depression which showed high resilience and a very low and low risk zones for flash floods.
5. The risk map showed that high and very high-risk zones occupied 39.49% (~ 1452.51 Km2) of the whole
Sunamganj District, while the lowest risk zones and very low and low risk zones was around 42.27% (~ 
1554.66 Km2). Again, the western and northwestern part of Sunamganj District were identified as very high and
high flood susceptibility zones.
6. The area-wise risk ranks were Dharmapasha upazila (59.21%) was in high and very high-risk zone and other
identified high risk zones were Sulla (58.94%), Derai (56.71%), Jamalganj (54.38%), Daksin Sunamganj
(53.58%), and Tahirpur (51.48%).

Low elevation, very gradual slope and high drainage density, presence of high amount of deep depression made
Sunamganj district more vulnerable. On the contrary, eastern parts upazilas of Sunamganj District were classified in
zones of very low to low flood susceptibility. The study would have much more accuracy if sufficient water level
data, precipitation station data in BWDB, updated BBS data were available. However, the study will be useful for
developing susceptibility maps in flash flood- prone regions and identify more possible causes to handle the flash
flood risk at local level.

References
1. UNISDR, C. (2015). No Title. The Human Cost of Weather-Related Disasters 1995-2015.
2. Abdelkarim, A., & Gaber, A. F. D. (2019). Based on the Integration of Geomatics and Hydraulic. Water, 11(9),
1988–2019.
3. Abedin, J., & Khatun, H. (2020). Impacts of Flash Flood on Livelihood and Adaptation Strategies of the Haor
Inhabitants: A Study in Tanguar Haor of Sunamganj, Bangladesh. The Dhaka University Journal of Earth and
Environmental Sciences, 8(1), 41–51. https://doi.org/10.3329/dujees.v8i1.50757

Page 15/26
4. Azotea, M. S. B., Necesito, I. V., Cheong, T. S., Yu, I., & Jeong, S. (2015). Evaluation of Flood Damages Using
Principal Component Analysis. Journal of Korean Society of Hazard Mitigation, 15(4), 215–220.
https://doi.org/10.9798/kosham.2015.15.4.215
5. Bisht, S., Chaudhry, S., Sharma, S., & Soni, S. (2018). Assessment of flash flood vulnerability zonation through
Geospatial technique in high altitude Himalayan watershed, Himachal Pradesh India. Remote Sensing
Applications: Society and Environment, 12, 35–47. https://doi.org/10.1016/j.rsase.2018.09.001
6. CEGIS. (2012). Master plan of haor area. Ministry of Water Resources, Government of the People’s Republic of
Bangladesh, I(April 2012), 1–82.
7. Das, S. (2019). Geospatial mapping of flood susceptibility and hydro-geomorphic response to the floods in
Ulhas basin, India. Remote Sensing Applications: Society and Environment, 14(January), 60–74.
https://doi.org/10.1016/j.rsase.2019.02.006
8. EarthExplorer. (n.d.). Retrieved August 28, 2021, from https://earthexplorer.usgs.gov/
9. Elkhrachy, I. (2015). Flash Flood Hazard Mapping Using Satellite Images and GIS Tools: A case study of Najran
City, Kingdom of Saudi Arabia (KSA). Egyptian Journal of Remote Sensing and Space Science, 18(2), 261–278.
https://doi.org/10.1016/j.ejrs.2015.06.007
10. Gigović, L., Pamučar, D., Bajić, Z., & Drobnjak, S. (2017). Application of GIS-interval rough AHP methodology for
flood hazard mapping in Urban areas. Water (Switzerland) 9.
https://doi.org/https://doi.org/10.3390/w9060360.
11. Gómez-Palacios, D., Torres, M. A., & Reinoso, E. (2017). Flood mapping through principal component analysis
of multitemporal satellite imagery considering the alteration of water spectral properties due to turbidity
conditions. Geomatics, Natural Hazards and Risk, 8(2), 607–623.
https://doi.org/10.1080/19475705.2016.1250115
12. Hossain, A., Rahman, H., Ishaque, F., Uddin, G. T., & Rashid, A. (2008). Flash flood management approach for
flood risk areas using numerical climate modeling data analysis. 1–15.
13. Kamal, A. S. M. M., Shamsudduha, M., Ahmed, B., Hassan, S. M. K., Islam, M. S., Kelman, I., & Fordham, M.
(2018). Resilience to flash floods in wetland communities of northeastern Bangladesh. International Journal of
Disaster Risk Reduction, 31(June), 478–488. https://doi.org/10.1016/j.ijdrr.2018.06.011
14. Kamruzzaman, M., & Shaw, R. (2018). Flood and Sustainable Agriculture in the Haor Basin of Bangladesh: A
Review Paper. Universal Journal of Agricultural Research, 6(1), 40–49.
https://doi.org/10.13189/ujar.2018.060106
15. Khokhar, M. A. H., Javed, S., Awan, H. B. H., Yousaf, I., Iqbal, A., & Waseem, L. A. (2021). Damage Assessment
and Spatio-Temporal Mapping Using Object Based Image Analysis Technique on Floods-2015 in Southern
Punjab, Pakistan. European Journal of Geosciences, 3(1), 9–18. https://doi.org/10.34154/2021-ejgs-
0013/euraass
16. Komolafe, A. A., Herath, S., & Avtar, R. (2019). Establishment of detailed loss functions for the urban flood risk
assessment in Chao Phraya River basin, Thailand. Geomatics Nat. Hazards Risk 10.
https://doi.org/https://doi.org/10.1080/19475705.2018.1539038
17. Kumar Biswas, N., Hossain, F., Bonnema, M., Aminul Haque, A. M., Kumar Biswas, R., Bhuyan, A., & Hossain, A.
(2020). A computationally efficient flash flood early warning system for a mountainous and transboundary
river basin in Bangladesh. Journal of Hydroinformatics, 22(6), 1672–1692.
https://doi.org/10.2166/HYDRO.2020.202

Page 16/26
18. Mahmoud, S. H., & Gan, T. Y. (2018). Multi-criteria approach to develop flood susceptibility maps in arid regions
of Middle East. Journal of Cleaner Production, 196, 216–229. https://doi.org/10.1016/j.jclepro.2018.06.047
19. Mondal, S., Akter, L., Hiya, H., & Farukh, M. (2021). Effects of 2017 Early Flash Flooding on Agriculture in Haor
Areas of Sunamganj. Journal of Environmental Science and Natural Resources, 12(1–2), 117–125.
https://doi.org/10.3329/jesnr.v12i1-2.52007
20. Sajjad, A., Lu, J., Chen, X., & Saleem, N. (2021). Rapid riverine flood mapping with different water indexes using
flood instances Landsat-8 images. 8049. https://doi.org/10.3390/ecws-5-08049
21. Ullah, K., & Zhang, J. (2020). GIS-based flood hazard mapping using relative frequency ratio method: A case
study of panjkora river basin, eastern Hindu Kush, Pakistan. PLoS ONE, 15(3), 1–18.
https://doi.org/10.1371/journal.pone.0229153
22. Y.K. Qiao, & Peng, F. L. (2017). No Title. Land Use Policy.
https://doi.org/https://doi.org/10.1016/j.landusepol.2017.08.037.
23. Zeleňáková, M., Fijko, R., Labant, S., Weiss, E., Markovič, G., Weiss, R. (2019). No Title. Flood Risk Modelling of
the Slatvinec Stream in Kružlov Village, Slovakia. J. Clean. https://doi.org/10.1016/j.jclepro.2018.12.008.
24. Zeng, F., Lai, C., & Wang, Z. (2012). Flood risk assessment based on Principal Component Analysis for
Dongjiang River basin. 2012 2nd International Conference on Remote Sensing, Environment and Transportation
Engineering, RSETE 2012 - Proceedings, 20–23. https://doi.org/10.1109/RSETE.2012.6260577
25. Zhang, D., Shi, X., Xu, H., Jing, Q., Pan, X., Liu, T., Wang, H., & Hou, H. (2020). A GIS-based spatial multi-index
model for flood risk assessment in the Yangtze River Basin, China. Environmental Impact Assessment Review,
83(March). https://doi.org/10.1016/j.eiar.2020.106397
26. Zhao, J.W., Peng, F.Le, Wang, T.Q., Zhang, X.Y., Jiang, B.N. (2017). No Title. Advances in Master Planning of
Urban Underground Space (UUS) in China. https://doi.org/https://doi.org/10.1016/j.tust.2015.11.011.

Figures

Page 17/26
Figure 1

Map of the Study Area

Page 18/26
Figure 2

Framework of Data Analysis

Page 19/26
Figure 3

Flood depth map

Page 20/26
Figure 4

(a) Elevation Map; (b) Slope Map; (c) Drainage Density; (d) Average Rainfall Map; (e) Topographic Wetness Index

Page 21/26
Figure 5

Flash Flood Exposure Index Map

Page 22/26
Figure 6

(a) Geomorphological Map; (b) WRI; (c) NDVI; (d) NDBI; (e) Agricultural Land map

Figure 7

Population Density (Left); Percentage of Depending Population (Right)

Page 23/26
Figure 8

Flash Flood Sensitivity Index Map

Page 24/26
Figure 9

Flash Flood Resilience Index Map

Page 25/26
Figure 10

Flash Flood Risk Map

Page 26/26

You might also like