REMOTE SENSING FOR FLOOD INUNDATION MAPPING USING VARIOUS PROCESSING METHODS WITH SENTINEL-1 AND SENTINEL-2

Abstract. The following paper discusses different remote sensing methodologies for flood mapping. The chosen study area is nearby Karlovo, Bulgaria where torrential rains in August and September 2022 caused devastating damages in the area. The location is still under critical status and damages are not completely evaluated, yet.

The aim of this study is to investigate how open satellite data can be applied in the mapping the results from the floods. The research team worked with radar and optical data from Sentinel-1 and Sentinel-2. Different technological approaches were applied to classify the land cover changes. Given the fact that the flood mass consisted mainly of mud an innovative approach was considered in order for the damages area to be computed.

Climate change predictions and deforestation suggest increasingly frequent floods in the future. For these reasons, rapid and timely flood mapping under specific conditions is very important in flood modelling, hazard and risk analysis. In order to investigate the flooded area in the region of the selected study area, different post-processing methods were used. The research team focused on using satellite imagery from both Sentinel-1 and Sentinel-2. Various spectral indices were computed as well as unsupervised and supervised classifications were applied. The results of the study were presented in the form of statistical analyses and comparison between the computational methods.