Characterization of Low-Backscatter Ocean Features in Dual-Copolarization SAR Using Log-Cumulants

The potential of using log-cumulants for discrimination between mineral oil spills and other low-backscatter ocean features in synthetic aperture radar data is investigated here. Radarsat-2 fine quad-polarization data containing experimental oil spills, as well as simulated biogenic slicks and a natural phenomenon, are analyzed. For this data set, the combined information of matrix log-cumulants of first and second order is found to clearly discriminate the majority of the mineral oil spills from the simulated biogenic slicks and the natural phenomenon. These preliminary findings suggest that the proposed method has a potential application in classification of low-backscatter ocean features of unknown origin.