Remote Sensing of 1998 and 2000 Floods in Greater Dhaka, Bangladesh: Experiences from Catastrophic and Normal events

This paper is an attempt to develop a series of maps that precisely depict flood prone areas in Greater Dhaka, Bangladesh using remote sensing techniques. Multi-temporal RADARSAT SAR data were acquired and employed to delineate open water flood boundary during the floods of 1998 and 2000. Using a threshold algorithm, SAR data is segregated into water and non-water areas. The empirical threshold value was obtained by using visual interpretation technique, local knowledge of the study site and by deriving corresponding pixel values to land/water from each image. The result demonstrated that 53 percent of the study area was heavily inundated in 1998 flood which is the largest submerged area during a catastrophic scenario. In contrast, 35.32 percent area was flooded during the year 2000 which represents the area under water for a normal event. Using the reference data acquired from field visit, derived flood maps were further validated. Moderate accuracy is obtained for all flood maps, however, July 1998 image attained the highest overall accuracy (86%) in the dataset. The derived flood maps are expected to be useful to mitigate losses of lives and property from river water flooding in Greater Dhaka. Furthermore, this information would be worthwhile to develop an efficient flood disaster management system.