Real-Time Flood Mapping with Temporal SAR Images Using ESA CloudToolbox Service

Abstract

Flood events are increasing at a drastic rate throughout the globe and became a major threat of increasing concern. The Synthetic Aperture Radar (SAR) system is capable of acquiring images during heavy rains and cloud cover. Many studies have proven the potential of SAR images in disaster management studies. In this study, a recent flood event in Kerala state, India, which occurred during Aug 2018 is analyzed using the European Space Agency (ESA) CloudToolbox service. To map the flood area, the C-band Sentinel-1 temporal SAR images and Sentinel-2 temporal optical images that are available within the duration of flood event are used. To process this massive volume of data, the Sentinel Application Platform (SNAP) with Graph Processing Framework (GPF) is used. The flood mapping approach adopted for each image is automated through XML code which is implemented in GPF. Kittler and Illingworth threshold algorithm is used to obtain the optimum threshold for flood identification. The results are validated against the ground truth data. The results indicate that ESA’s CloudToolbox service is quite efficient in handling the big volume of data and real-time information extraction.