Mozhgan Abbasi - Assistant professor of Natural Resource and Earth Science Faculty, Shahrekord University
Faezeh Khalifeh Soltanian - PhD candidate of Natural Resource and Earth Science Faculty, Shahrekord University
Hamid Reza Riyahi Bakhtyari - Assistant professor of Natural Resource and Earth Science Faculty, Shahrekord University

During flooding it is very difficult and costly to collect information about the location and required damage mitigation for disaster management authorities. In this case, the use of SAR based satellite data due to its ability to penetrate clouds has the advantages for operational near real-time flood monitoring. A large flash flood event took place in late March early April 2019 following an intense and heavy rainfall that occurred in the western of Golestan province, north of Iran. During the flood events, large area including the entire plain around the city of Aqqala includes all agricultural fields as well as the city itself inundated. Two space born radar images, obtained from Sentinel-1 covering the same area, one before and another one during the flood event, were processed. The aim of this study is to evaluate of the performance of SAR-derived Sentinel-1 products to generate flooding map using random forest classifier. The area of flooded pixels based on the results of change detection of two datasets was obtained 233.20 km2. This research provided a useful and rapid reference for flood mapping using accessible and high resolution SAR remote sensing imagery.

Flood mapping, SAR satellite data, Sentinel-1, Random forest, Classification