Extraction of Sea Ice Information from SAR for Numerical Sea Ice Prediction

New techniques for automated extraction of sea ice information from synthetic aperture radar (SAR) RADARSAT-2 images with a very high level of confidence are presented. The ice/water detection approach developed based on more than 15,000 SAR images (2010-2016) and corresponding Canadian Ice Service (CIS) Image Analysis products is able to classify 65.4% of SAR data with the accuracy of 99.8%. Introduction of an auxiliary SAR image acquired several hours before allowed to further increase the number of retrievals. Ice concentration is also estimated based on the number of ice/water retrievals. The large number of SAR retrievals with very high accuracy makes it possible to assimilate them in operational Environment and Climate Change Canada (ECCC) Regional Ice-Ocean Prediction System (RIOPS).