Flood damage assessment with multitemporal earth observation SAR satellite images: A case of coastal flooding in Southern Thailand

Estimation of economic loss due to flood often becomes necessary for flood damage assessment. Conventional practices to estimate damage by postflood survey are laborious and time-consuming. This study presents a framework of rapid estimation of flood damage using earth observation satellite data to minimize the time and efforts for damage assessment. The study encompasses the case of severe coastal flooding in Nakhon Si Thammarat Province in Southern Thailand in January 2019. Land use-land cover (LULC) maps of study sites are prepared using Random Forest (RF) classification in Google Earth Engine (GEE) to classify cropland and built-up areas. For this case study, Sentinel-1C-band synthetic-aperture radar (SAR) data provided by ESA (European Space Agency) were used. The datasets were taken before and after the flood incident and were processed using Sentinel Toolbox. Flood inundation maps generated using Kittlerand and Illingworth’s automatic thresholding algorithm are validated using Joint Research Centre (JRC), European Commission flood datasets. LULC maps are overlaid on flood inundation maps to assess the agricultural and habitat loss in the study area using Sentinel-1 SAR data. This study shows effectiveness of Sentinel-1C-band SAR for flood inundation mapping and land cover classification and cross-validates the previous studies that proved the effectiveness of Sentinel-1 data. Performance evaluation using accuracy assessment proves that this approach can be used for accurate estimation of agriculture losses/risks caused by coastal flooding.