SAR coherence and polarimetric information for improving flood mapping

Abstract

By providing high quality flood maps a spaceborne SAR can be an effective source of information. These maps support civil protection authorities for disaster risk reduction. Here we propose a methodology for identifying floods that occur on different types of land cover, such as urban areas, bare and poorly vegetated soil and vegetated areas. Our approach takes advantage of polarimetric SAR data and InSAR coherence to better characterize the landscape. Indeed, the Sentinel-1 repeat cycle of six days, and its systematic acquisition of dual-pol SAR data, provides an unprecedented chance to develop automatic, high frequency flood mapping algorithms for complex environments. The algorithm has been tested on two different Sentinel-1 datasets, acquired, respectively over Greece and Italy, showing promising results.