Potential and limists of satellite-derived digital surface model data for assessing flood risks in Southeast Coast of India
Résumé
Flood risk assessment in low-lying coastal areas requires efficient spatial observations of land elevation for theimplementation of protection, evacuation and safeguard plans of people and assets. Here we evaluated the potentialof Digital Surface Model (DSM) derived from satellite observations to map flood prone areas with the objective ofearly warning on flood risk in the Cuddalore and Pondicherry region, southeast coast of India. Coastal zonemanagement of this 100 km long coast is particularly challenging. Indeed, the whole region experiences at leasttwo cyclonic storms accompanied with storm surge, heavy rains, flooding and beach erosion every year; the havocswreaked by the 2004 tsunami, flash floods of 2005 and 2015, and the Thane cyclone in 2011 are still closememories. We analyzed Sentinel-1 Synthetic Aperture Radar (SAR) and ALOS World 3D DSM satellite data, andGoogle Earth images. All these data are freely available and we compared them to the population census dataacquired in 2011. Using Sentinel-1 SAR images, we discriminated flooded from non-flooded areas before comparingmaps of low-lying areas derived from ALOS DSM data. The results suggest a good agreement between real floodedareas and low-lying areas. However, the micro-topography reflecting channels and drainage systems could not becaptured with important issue for delineating areas with high risks of flooding. We explained that spatial resolutionof about 2 m in X, Y and 10 cm in Z directions are necessary for identifying areas with high risk of flooding asdemonstrated in many countries of the world. It is time to rethink national Indian spatial policy about high-resolutionimages in order to prepare safety plans of the property and the lives of populations of Tamil Nadu coasts.