Surface Water Mapping from SAR Images Using Optimal Threshold Selection Method and Reference Water Mask

Water resources are an important component of ecosystem services. During long periods cloudiness and precipitation, when a ground-based sample is not available, the water bodies detected from satellite SAR (synthetic-aperture radar) data using threshold methods (e.g., Otsu Kittler–Illingworth). However, such do enable to obtain correct value for backscattering coefficient (?0) relatively small areas in image. The paper proposes substantiates method mapping surface bodies, which makes it possible correctly identify even “water”/“land” class imbalance situations. operates on principle maximum compliance resulting mask with given reference mask. Therefore, enables exploration possibilities searching choosing optimal parameters (polarization speckle filtering), provide quality was applied natural industrial Pohjois-Pohjanmaa region (North Ostrobothnia), Finland, Sentinel-1A -1B ground range (GRD) (ascending descending orbits) 2018–2021. Reference masks were generated based optical spectral indices derived Sentinel-2A -2B data. polarization filtering chosen since they most accurate ?0 (on average all observations above 0.9 according Intersection over Union criterion) resistant random fluctuations. If proposed more than method. Without mask, calculated as thresholds obtained previous observations. In this case, good accuracy It shown that identification under significant conditions, covers only fraction percent area study.