Optimizing change detection methods for flood mapping using polarimetric SAR

Citation

Felix IK, Armando M, Berardi A, Bovolo I, Hunter P, Neil C, Perez CS & Silva TSF (2025) Optimizing change detection methods for flood mapping using polarimetric SAR. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. https://doi.org/10.1109/jstars.2025.3628450

Abstract
Flooding is becoming increasingly frequent and severe worldwide, posing significant risks to lives and property. Accurate flood information, particularly regarding its extent and location, is essential for effective mitigation efforts. This study examines the May 2023 Emilia-Romagna flood in Italy to enhance flood detection accuracy. For the first time, we applied optimization of power difference and ratio polarimetric change detection methods to identify the most effective flood detection method. Additionally, we tested various reference images within a time series to determine the most suitable reference image using Sentinel-1 Synthetic Aperture Radar data spanning from 2017 to 2023. Results revealed that the Optimisation of Power Ratio (OPRatio) method was most effective in detecting flooded areas. Notably, we established that in the study area, the optimal reference image is not always the one immediately preceding the flood; instead, an image acquired a month before flood or a composite of images from March across multiple years provided the most accurate results. This approach, combined with the OPRatio detector, achieved the highest accuracy and lowest false alarm rates. When applied to a flood event in Scotland, it similarly reduced false detections. This study underscores the importance of employing polarimetric change detectors alongside optimal reference images to improve the precision and reliability of flood mapping.

Keywords
Floods; Accuracy; Sentinel-1; Synthetic aperture radar; Rain; Indexes; Optical imaging; Detectors; Uncertainty; Satellites

Journal
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

• Published version