Flood Hazard Analysis in Greater Dhaka of Bangladesh Using Remote Sensing Data with Geographical Information System

摘要

Flooding is a regular feature in Dhaka during the monsoon. Dhaka has witnessed many disastrous floods of which the 1987, 1988, 1998 and 2004 ones are notable in recent flood history. The objectives of this paper are, firstly, to estimate flooding in Dhaka using Landsat TM and RADARSAT Synthetic Aperture Radar (SAR) data for three greatest events in 1988, 1998 and 2004, and secondly to assess flood hazard using remote sensing data with Geographical Information System (GIS). A threshold technique was employed to estimate flooding areas from the Landsat TM data while a rule based approach was used to delineate land-water boundaries in the SAR datasets. In order to accurately quantify the flooded area, the images taken during the flood and dry season were classified as water and non-water categories, and were subsequently superimposed. Using flood maps derived from the multi-temporal remotely sensed data, an observed flood frequency map was generated and used as a hydrological parameter to assess the flood as hazard. Elevation heights and land use/cover data were used through GIS approach to construct a flood hazard map. Using a ranking matrix in a two dimensional multiplication mode, a flood hazard map was generated by considering the interactive effect of two thematic layers (land cover and elevation) for observed flood frequency, Flood estimation showed that 47.1% area of Greater Dhaka was inundated in the 1988 flood, while in 1998 and in 2004, the inundation areas were 53.2% and 43%, respectively. The flood hazard map revealed that major portion of the study area comprised medium to high hazard zone. For example, the fringe areas, in which most of the dwellings are housed by the urban poor, are exposed to high hazard category. Using this map, flood preparedness, relief, and rehabilitation programs can be formulated for the successful management of potential flood hazard in Greater Dhaka of Bangladesh.