Rapid detection of regional level flood events using AMSR-E satellite images

Abstract

Remote sensing plays a prominent role in the rapid detection of the flood event at a regional level. In this paper, the potential of AMSR-E images in regional level flood detection was identified. The study area of the research covers a part of Krishna river basin in the Andhra Pradesh state of India. Spatio-temporal database of daily Land Surface Water Coverage (LSWC) was developed by using Normalized Difference Polarization Index (NDPI). NDPI is calculated using AMSR-E brightness temperature of vertical and horizontal polarizations at 36.5 GHz frequency. The flood anomaly identified from the LSWC database is in strong agreement with actual flood events like Ogni cyclone. To extract the hidden information and similarities in the temporal images, image similarity was calculated by using Bhattacharya distance. Based on the similarity values, all the images in the database are ranked which helps in rapid flood information extraction. Among the various flood events identified by the database, Ogni cyclone is chosen for in-depth analysis. The SAR images acquired during the Ogni cyclone was used to validate the results of AMSR-E outputs.