Urban Flood Mapping of the July 2020 Kyushu, Japan Heavy Rain Based on Interferometric Coherence of Sentinel-1 Images

Abstract

This study examined a methodology for urban flood mapping based on discrimination analysis of pre- and co-event interferometric coherences obtained from multitemporal Sentinel-1 SAR images. The method was applied to the areas in Kumamoto prefecture, Kyushu Island, Japan affected by the heavy rain in July 2020. Two pre-event images, one post-event image, and existing land cover map were analyzed with the help of inundation inventory data estimated from aerial photos. We revealed that the co-event coherences in the inundation urban areas were significantly decreased from the pre-event coherences. The result of the discriminant analysis of the pre- and co-event coherences shows the inundation areas were well extracted, and the inundation areas not identified in the inventory were also extracted in the coastal areas.