| Summary: | Ice jams related flooding causes tremendous loss both physically and economically in society by destroying habitats, damaging the infrastructures, water supply, navigation. Hence, it’s of great importance to account this phenomenon. Ice jam study using Synthetic Aperture Radar (SAR) data has gained demand, however only a few studies came out using Sentinel-1 data, especially with vertical transmit & vertical receive (VV) and vertical transmit & horizontal transmit (VH) backscatter polarization for detecting ice jams. This study aims to put forward an automatic algorithm to detect the ice jams using Sentinel-1 backscattering intensities VV and VH for the first time in Kemijoki River system in Finland. 10 days average mean mosaics of Sentinel-1 Ground Range Detected (GRD) products were used as input. The algorithm was successful in detecting ice and ice jams during the study period from December to March between 2018 to 2021. VH backscatter polarization gave the best results in delineating the ice jam. To quantify the results, River Lake Ice Extent product based on Sentinel-2 and for visual validation RGB (red, green, blue) composites of Sentinel-2 Mosaics have been used. Visual validation was the most trusted in this method. The recall and precision were above 70 percent. Cloud cover and lack of good in-situ reference data was also a draw back for the validation of result. The potential ice jam detected in Rovaniemi is opening the wider possibility for applying similar technique for finding ice jams for the entire Finland.
|
|---|