PROBABILISTIC URBAN FLOOD MAPPING USING SAR DATA

In this work we present an automatic algorithm for providing probabilistic flood maps, not only on bare soils, but also within urban areas. The probabilistic flood mapping procedure is based on synthetic aperture radar (SAR) data and the Bayesian inference. Both intensity data and Interferometric SAR (InSAR) coherence feature are used. The approach improves the information content of a binary SAR-based floodwater map, which does not give any indication on the uncertainty in the pixel state. The proposed methodology is tested for the flood event that heavily affected the city of Houston (Texas) during the 2017 hurricane season. Data provided by the Sentinel-1 mission are used, with a geometric resolution of 20m. The algorithm takes fully advantage of the Sentinel-1 mission's repeat cycle of six days and narrow orbital tube to fully exploit the potentialities of InSAR coherence feature to detect floodwater in complex environments. The application of the proposed method to the Houston case study showed promising results.