DETECTION OF FLOODED URBAN AREAS USING SAR: AN APPROACH BASED ON THE COHERENCE OF STABLE SCATTERERS

The utility of synthetic aperture radar (SAR) data to produce flood delineation maps is well established. However, for what concerns urban settlements, flood mapping still represents a challenge, because the radar signatures of flooded urban pixels are generally ambiguous. As a matter of fact, flood mapping algorithms generally do not consider urban areas, thus producing a lot of missed detection errors. To cope with this problem, this study proposes a new method that basically analyzes the complex coherence of urban pixels characterized by high backscatter combined with high temporal stability (stable scatterers). Contextual information is also used to reduce the noise of the maps. The rationale is that, since water surfaces show no coherence in a repeat-pass interferogram, a decrease of coherence may occur even for (some) stable scatterers if floodwater is present in a resolution cell. To develop the algorithm, the inundation that hit Houston (Texas, USA) in April 2016 was considered. This event was observed by Sentinel-1 in Interferometric Wide Swath mode. Results showed that looking at the coherence of stable scatterers could represent a reliable road to at least mitigate the problem of missed detection of flooded urban settlements.