Interannual comparison of historical floods through flood detection using multi-temporal Sentinel-1 SAR images, Awash River Basin, Ethiopia

Synthetic-aperture radar (SAR) data from Sentinel-1 satellites provides unprecedented opportunity to evaluate inter-annual flood characteristics, although consensus on best flood detection methods is lacking. This study compared the performance of three flood detection methods to evaluate inter-annual flood characteristics at two sites in the Awash River Basin of Ethiopia. The methods are Change Detection and Thresholding (CDAT), Normalized Difference Flood Index (NDFI) and Root of Normalized Image Difference (RNID). The reference flood map was prepared based on a field survey for the maximum extent of the 2020 flood. Inter-annual flood characteristics were evaluated in terms of flood onset, recession and frequency of occurrence over the analysis period (2017 to 2022) but with a particular focus on the 2020 extreme flood events at Borkena and Dubti sites. Findings showed that the performance of the flood detection methods significantly differed. The RNID method, which allowed manual estimation of threshold, provided the highest flood detection capability at both sites. Flood detection accuracy improved when normalizing signal backscatter intensity of S-1 in change detection method. Flood onset and recession showed noticeable difference across the sites. Findings of this study indicate the potential of the satellite remote sensing methods to evaluate the spatial and temporal characteristics of floods, but further research is needed to evaluate and improve the performance of these methods for other flood affected sites.