Water channel detection from Sentinel-1 SAR images for large scale river monitoring

In the remote sensing science applied to river systems, Synthetic Aperture Radar (SAR) is able to provide quite good resolution, all-weather, day and night, coherent data which is complementary tooptical data. SAR technology is still poorly exploited in river science, mainly due to the demandingpre-processing tasks and the complexity of the signal interpretation. One of the most intuitive andpromising applications in river science is the delineation of the water channel for example for waterdischarge estimation based on water channel widths. The present work presents a robust framework for the delineation of water channel in medium-large (>20-25 m) river reaches. Themethod is built on a short-stack of multi-temporal SAR data and combines different polarization signatures, the intensity of backscattering and the signal phase to detect the probability of water location. The work is based on the images from Sentinel 1 of the European Copernicus program, a constellation of 2 satellites that carry a C-band SAR that, in its TOPS mode configuration, provides global coverage with a temporal resolution of 6 days and a spatial resolution of about 5 by 20 meters. The case studies are located in the Po river basin (Italy). The method allows fast and all conditions water channel location detection at large scale. In the future, this classification may be refined by combining the results with optical information from, for instance, Sentinel 2 for monitoring water channel width and then discharges every few days.