1 Introduction

Dhaka, the capital city, is the largest and densely populated in Bangladesh. The main natural hazards affecting Dhaka include floods, which are associated with river water overflow and rain water stagnation. After the devastating 1988 flood, the western part of the city was embanked under Dhaka Integrated Flood Protection Project (DIFPP), Phase-I, which protected the area from flood (Mohit and Akhter 2000). The eastern part of the city is yet to be protected. It is observed that some 60% of the eastern Dhaka regularly goes under water every year between June and October due to lack of flood protection (Halcrow Group Ltd 2006a, b). In 1991, Japan International Cooperation Agency (JICA) and Asian Development Bank (ADB) conducted a feasibility study on this area. Afterward, a number of studies were carried out to solve the flood problem under auspices of different authorities and agencies (Dewan et al. 2004). Finally in 2006, Halcrow Group Limited, UK, undertook a study for updating/upgrading the Feasibility Study of Dhaka Integrated Flood Control Embankment. They proposed some structural measures that include construction of embankment, flood wall, pump station and buildup of some pond area. Although the Bangladesh government has invested large amount of resources in protecting Dhaka and its adjoining areas from recurrent floods by constructing embankment, flood vulnerability is increased significantly (Dewan et al. 2007). Experience shows that only structural approach is inadequate to combat recurrent flood problems. From the devastating 1988, 1998 flood monitoring, non-structural approach such as flood hazard map and risk map have been identified to be major issues by the water experts, urban planners and key policy makers of the country (Dewan et al. 2007).

Over the last two decades, remote sensing has played an increasing role in the field of water resource management (Dewan et al. 2005b). Geographic Information System (GIS) is also used extensively to model surface water and flood damage assessment. A flood hazard map and land development priority map for the whole country were developed in 2000 using National Oceanographic and Atmospheric Administration (NOAA) Satellite, Advanced Very High Resolution Radiometer (AVHRR) and Geographic Information System (GIS) data (Islam and Sado 2000; Al-Hussaini 2005). Several studies were undertaken to assess flood hazard of Dhaka city by preparing a flood hazard map (Dewan et al. 2005b, 2007), by estimating flood damages (Dewan and Nishigaki 2004; Dewan et al. 2005a), by delineating flood damaged zones (Mohit and Akhter 2000) or by evaluating historical flood events (Dewan et al. 2004, 2005b; Mohit and Akhter 2000). Different approaches were taken in these studies which were mostly based on satellite images collected from RADARSAT, SAR (Synthetic Aperture Radar), NOAA, AVHRR, etc. without conducting any hydrodynamic simulation. To estimate the extent of flooding associated with a given return period, however, the GIS must be combined with an applicable hydrologic/hydraulic method for estimating stages (Werner 2001).

In this study, an attempt was taken to assess flood hazard of mid-eastern part of Greater Dhaka by developing a flood hazard map through 1D hydrodynamic simulation on the basis of digital elevation model (DEM) data and the hydrologic field-observed data. Finally, to assess the flood risk of that area, a risk map was prepared where risk was defined (Apel et al. 2009) as the product of hazard (i.e., depth of inundation) and vulnerability (i.e., the exposure of people or assets to flood).

2 Study area

The eastern Dhaka extends from Demra to Tongi. A large part of this area is covered by the flood protection and drainage proposals referred to as FAP (Flood Action Plan) 8A, consisting of a continuous zone extending along the eastern side of the city, from Narayanganj in the south to Tongi in the north (Halcrow Group Ltd 2006a, b). Eastern Dhaka has been divided into three compartments in the proposed Dhaka Integrated Flood Protection Project. For this study, the middle part (compartment-2) with an area of 37.16 km2 was selected as study area which is shown in Fig. 1. Geographic location of the study area is in between Latitude N 23°45′50″ to N 23°50′30″ and Longitude E 90°25′09″ to E 90°29′30″.

Fig. 1
figure 1

Location of study area in Dhaka City Map (Modified from Banglapedia 2004)

Full size image

Western part of the area is mainly built-up area, and eastern part is low-lying flood plain area used for cultivation. In recent times, there has been a considerable amount of peripheral development by means of land fill, particularly in the central and southern portions of the area, close to the city center. The study area can be classified into five types according to land-use pattern: built-up, agricultural land, land filled, tree and channel. A detailed land-use map was created (Fig. 2) according to recent satellite image.

Fig. 2
figure 2

Land-use map

Full size image

The study area is exposed to the main urban stream in the west and river Balu in the east. Balu river offtake water from Tongi Khal and outfall to Lakhya river. Average width of the river is 100 meter. Average recorded discharge in dry season (February and March) and in monsoon (August) is 60 cusec and 744 cusec, respectively (BWDB 2005).

3 Data and material

Topographic data that include DEM, satellite image and river cross section and hydrologic data covering discharge and water level were used for this study. Topographic data are crucial for flood inundation modeling, and it is best to use recent and highly accurate topographic data. However, this is not always feasible given time and budget constraints, and therefore, it is of interest to use DEM that can be accessed online and downloaded without charge (Sanders 2007). In recent times, DEM become very important data sources for geoscientists and has been intensively using in a wide range of topographic analysis, flood modeling and other natural studies (Dewan et al. 2004). A DEM is a representation of continuous elevation values usually at a fixed grid interval over the surface of the earth (Lehner et al. 2006). DEM data for this study were collected from the HydroSHEDS (Hydrological data and maps based on SHuttle Elevation Derivatives at multiple Scales) (HydroSHEDS 2010). It provides hydrographic information in a consistent and comprehensive format for regional- and global-scale applications. It offers a suite of geo-referenced data sets (vector and raster), including stream networks, watershed boundaries, drainage directions, and ancillary data layers such as flow accumulations, distances and river topology information (Lehner et al. 2006). HydroSHEDS is derived from elevation data of the Shuttle Radar Topography Mission (SRTM) at 3 arc-second (90 m) resolution during an 11-day mission in February of 2000. Preliminary quality assessments indicate that the accuracy of HydroSHEDS significantly exceeds that of existing global watershed and river maps (Lehner et al. 2006). It may appear quite attractive in cases where high-resolution surveys (such as LiDAR or IfSAR) are cost-prohibitive, and there are other larger factors of uncertainty. It is advantageous for flood modeling that SRTM vertical accuracy is better on relatively flat terrain, such as flood plains, compared with high relief area (Sanders 2007). As Dhaka has flat topography elevation ranges varies from 0.5 m to 12 m, with 70% of the total area within 0.5–5 m (Halcrow Group Ltd 2006a, b), DEM from SRTM was chosen for this study.

Satellite image of the study area was acquired from Google™ Earth (DigitalGlobe image; Date of imagery: 7th March 2007). As the area is rapidly urbanizing, a lot of land development was identified by superimposing the image over the DEM. So for reliable analysis, some modification was done to update the collected DEM which is described in Sect. 4.1.2.

For DEM data processing and mapping, ArcGIS (ESRI 1999) and for hydrologic simulation HEC-RAS (Hydrologic Engineering Center 2002) were used. HEC-RAS, a 1D hydrodynamic model is commonly used to predict flood stage and to map inundation by extrapolating channel centerline flood stage predictions to over bank areas (Sanders 2007).

4 Methodology

4.1 Hazard assessment

Assessment of flood hazard by developing a flood hazard map was carried out by 1D hydrodynamic simulation. The methodology for developing a flood hazard map can be divided into three phases: preparation phase, execution phase and verification and flood hazard mapping phase (flow chart shown in Fig. 3). Important steps of these phases have been briefly described below.

Fig. 3
figure 3

Flow chart of methodology

Full size image

4.1.1 Geo-referencing and projection

Collected satellite image was geo-referenced according to the geographic coordinate system (GCS_WGS_1984). DEM was also in geographic coordinate system. Geographic coordinate systems indicate location using longitude and latitude based on a sphere (or spheroid), while projected coordinate systems use X and Y based on a plane. Projections manage the distortion that is inevitable when a spherical earth is viewed as a flat map. Projected coordinate system used for this study is WGS_1984_UTM_Zone_45 N which is suitable for Bangladesh (ESRI 1999).

4.1.2 DEM modification

When using DEM for flood modeling, there are several limitations to consider. No DEM can resolve channels smaller in width than twice the DEM resolution. Three survey points, one in the middle of the channel and one on each bank, are the bare minimum for a crude definition of the cross section. When channels are not resolved, models are likely to overestimate flood extent and underestimate flood speeds (Sanders 2007). In this study, Balu river (avg. width 100 m) was not identified in the collected 90-m resolution DEM data. Moreover, it is good practice to perform ground surveys of study sites when DEM of questionable accuracy are utilized. Ground survey data can be used to estimate DEM accuracy or can even support additional data processing by providing a basis for ground-truthing the DEM (Sanders 2007). Here, the acquired DEM data were based on satellite mission of year 2000 (Lehner et al. 2006). After that, a lot of land development work has been completed in this area which was observed in recent satellite image. So, the DEM was modified according to current topography. Following steps (Flow chart shown in Fig. 4) were involved in the modification of DEM data: Step 1: Elevation values of DEM were converted from integer to float format. Step 2: The 90-m DEM was re-sampled to 30-m resolution using Bilinear interpolation method. Step 3: Then, the DEM data through the river path were extracted and converted into ASCII format and finally modified the elevation according to actual cross section of the Balu river. Step 4: The DEM data of land filled area were also extracted by observing recent satellite image and then raised the elevation. Step 5: Finally, the modified DEM was merged with the original DEM. The DEM both original and modified is shown in Fig. 5.

Fig. 4
figure 4

Flow chart of DEM modification

Full size image
Fig. 5
figure 5

a Original DEM, b Modified DEM

Full size image

4.1.3 Processing on HEC-RAS

In HEC-RAS, the geometric data were imported which were exported from ArcGIS by HEC-GeoRAS tool. Frequency analyses were carried out with water level data for 32 years (1972–2004) by the Gumbel Distribution method. The maximum water level was calculated for flood of 100-year return period and assigned at upstream and at downstream as boundary condition which were 7.2 and 7.05 m, respectively. Initial flow 100 m3/s was given as initial condition. With this consideration, inundation depth was calculated with 20-m resolution. Then, these data were exported to the ArcGIS.

4.1.4 Flood hazard mapping

Flood hazard map, which can provide information including the past flood track records, flood anticipation, potential evacuation routes, evacuation places, etc. to the local residents, is indispensable for emergency response and for long-term flood disaster management (Osti et al. 2008). In this study, a flood hazard map was prepared using the inundation status, which was found from hydrologic simulation. According to inundation depth, the whole area was divided into five categories. Some evacuation centers were proposed in high places. Some important places such as hospital and police box were also marked in this map.

4.2 Vulnerability and risk assessment

The risk faced by people must be seen as a cross-cutting combination of vulnerability and hazard (Wisner et al. 2004). Disasters are a result of the interaction of both; there cannot be a disaster if there are hazards but vulnerability is (theoretically) nil, or if there is a vulnerable population but no hazard event. These three elements: risk (R), vulnerability (V) and hazard (H), can be written in a simple form (Wisner et al. 2004):

$$ {\text{R}} = {\text{H}} \times {\text{V}} $$
(1)

In this study, an attempt was taken to make a flood risk map. Risk index was calculated by multiplying vulnerability and hazard index. Average depth of inundation was assigned as hazard index. And for calculating vulnerability index, percentage of area covered with house/living place and agricultural land were considered. These steps were followed to prepare the risk map: Step 1: The whole area was sliced into 300 m × 300 m cells resulting in total 624 cells (satellite image of 4 cells out of 624 cells are shown in Fig. 6). Step 2: Average inundation depth in each cells was calculated by re-sampling the 20-m resolution inundation map obtained from inundation simulation to 300-m resolution using Bilinear interpolation method. Step 3: For each cell, an integer value ranging from 0 to 5 was assigned as a hazard index according to inundation depth (shown in Table 1). Step 4: Equation 2 was used for calculating vulnerability index. Weight factors 10 and 2 used for area covered by house and agricultural land, respectively. Step 5: A risk index for each cell was calculated by multiplying hazard and vulnerability indexes (Eq. 3). Step 6: Then, these risk values were converted to raster format and imported to ArcGIS. Step 7: A risk map was prepared by classifying into three distinct risk zones, corresponded to low-, medium- and high-risk areas according to risk index (Table 2).

Fig. 6
figure 6

Satellite Image of 4 cells out of 624 cells

Full size image
Table 1 Assigned hazard index (H) for varying inundation depth
Full size table
Table 2 Area classification according to risk index
Full size table
$$ {\text{V}}_{\text{Index}} = \frac{{10 \times {\text{A}}_{\text{H}} + 2 \times {\text{A}}_{\text{Agri}} + 0 \times {\text{A}}_{0} }}{{{\text{A}}_{\text{Cell}} }} $$
(2)

where VIndex = vulnerability index (ranging from 0 to 10), A H = area covered by house/living place, A Agri = area covered by agricultural land, A 0 = area used for none, A Cell = area of one cell

$$ {\text{R}}_{\text{Index}} = {\text{H}}_{\text{Index}} \times {\text{V}}_{\text{Index}} $$
(3)

where RIndex = risk index (ranging from 0 to 50), HIndex = hazard index (ranging from 0 to 5), VIndex = vulnerability index (ranging from 0 to 10)

5 Results and observations

It is observed from simulation that inundation depth ranges from 1 to 3 m covers most of the area (64% with respect to total inundated area). Southern part of the study area is relatively low-lying where inundation depth is more than 3 m with maximum of 7.55 m. However, built-up area located in western part is mostly unaffected due to higher topography. Extent of inundated area is 20.24 km2 which is 54.4% of whole area. Table 3 shows the results obtained from this analysis. The simulation result has been verified with observed inundation depth of 1988 flood (Halcrow Group Ltd 2006b). Furthermore, the result has been validated by comparing the extent of inundation found from satellite image of just after (November 20, 2007) devastating Category IV (Paul 2009) cyclone “Sidr” (which struck Bangladesh on 16th November 2007). A flood hazard map (Fig. 7) was prepared using this simulation result by classifying into five categories, corresponded to different inundation depth. In order to ameliorate flood-induced damage, the developed flood hazard map would be invaluable. Government can use the map for ensuring the proper development planning of the high hazard zones which is suppose to be urbanized very soon. Urban planner can use this information to make environmentally sound land-use decisions.

Table 3 Percentage area inundated according to varying inundation depth
Full size table
Fig. 7
figure 7

Flood hazard map of mid-eastern Dhaka

Full size image

A risk map (Fig. 8) was prepared by classifying into three distinct risk zones, corresponded to low-, medium- and high-risk areas where risk was calculated as the product of hazard (i.e., depth of inundation) and vulnerability (i.e., the exposure of people or assets to flood). It is observed that high-risk zone covers few areas, and it is located mostly adjacent to river and in transitional zone between western built-up area and low-lying area. In this area, risk is high because area coverage with houses is high which means population density is also high; huge damage for infrastructure and building may occur. High-risk area represents low-lying built-up area where people are more exposed to hazard than those living in other locations. It is also observed that, in the western built-up area, there is no inundation means risk index is zero, is completely risk-free though the area is densely populated. Southern area covered mostly by agricultural land where inundation depth is maximum, falls in medium-risk category though no people living there. For the effective planning of flood defenses and the safety of the people living in high-risk areas, useful information can be provided using this risk map. Priority can be given to the developments necessary in this area. Aid can be provided and necessary advance action taken for a future flood event by understanding the map.

Fig. 8
figure 8

Risk map of mid-eastern Dhaka

Full size image

6 Concluding remarks

In this paper, assessment of flood hazard by developing a flood hazard map for mid-eastern Dhaka of Bangladesh was carried out by 1D hydrodynamic simulation using both topographic remote sensing data and hydrologic field-observed data. The study demonstrates a simple and effective way to modify the collected DEM so that it represents the current topography. The objective of flood hazard map is to provide residents with the information on the range of possible damage and the disaster prevention activities. The effective use of hazard map can decrease the magnitude of disasters. On the other hand, flood risk map represents the current scenario of that area according to degree of risk. The map provides helpful information about flood risk management and should be useful in assigning priority for the development of high-risk areas. In addition, the study may have considerable management implications for emergency preparedness, including aid and relief operations in high-risk areas in the future. These two maps may also help the responsible authorities to better comprehend the inundation characteristics of the flood plains, the protection of which is their responsibility. The general public will be made aware of the imagery of flooding which helps in understanding the risk of flood. Finally, these maps in digital form can be used as a database to be shared among the various government and non-government agencies responsible for the construction and development of flood defense.