Detection of the Leads in the Arctic Drifting Sea Ice on SAR Images

Information about the state of the sea-ice cover (e.g. presence of the open-water fractures) is extremely important under the harsh sea-ice conditions. This is especially valuable when navigation through the Northern Sea Route becomes a year-round possibility. Synthetic Aperture Radar (SAR) satellite images are used to monitor the Arctic sea ice. There are systematic data records available from the last three decades. Quantitative characteristics of the new ice/fracture features in dual co- and cross-polarization properties are very different and maybe effectively used in the development and improvement of SAR detection and discrimination algorithms. We propose a classification method to separate open -water fractures (leads) using polarization ratio and polarization difference together with texture features and Neural Network implementation for sea-ice classification of SAR images.