ERS SAR studies of sea ice signatures in the Pechora Sea and Kara Sea region

Results from a statistical analysis of 105 synthetic aperture radar (SAR) images and meteorological data are presented, covering parts of the Pechora Sea and Kara Sea in the Russian Arctic. Wind speed, air temperature, and other data were collected for the SAR sample areas, and a manual sea ice classification of the SAR samples was performed. All variables were input to different multivariate regression analyses, which were used to separate ice and water samples. In subsequent regression analyses the classes young ice and rough first-year ice were separated from water, and the two ice types were separated from one another. In the separation of all ice types from water, correlation coefficients of up to 0.90 were achieved between predicted and actual values. Correlation coefficients were as high as 0.93 between predicted and actual values for open water and young ice and for open water and rough first-year ice. When trying to separate young ice from rough first-year ice, correlation coefficients of about 0.60 were obtained. The study indicated that the mean and standard deviation of the backscattering coefficients and air temperatures were the most important information for separating water and sea ice using regression techniques.