With the potential for a greater human presence and increased development of natural resources, culturally and biologically sensitive shorelines in the Arctic may face increasing risk of environmental emergencies. To establish response contingency plans it is necessary to improve upon available shoreline maps. This study assessed the potential for automated classification of Arctic shore and nearshore land covers using polarimetric Radarsat-2 imagery. Repeatability and optimal acquisition parameters (incidence angle and polarization) were evaluated through comparison of multiple incidence angle images acquired over two study areas in the southern Beaufort Sea, Northwest Territories, Canada. Analyses of backscatter coefficients and decomposition parameters showed that shallow angles are generally preferred, and that discrimination of a number of land covers (including: sand, mixed sediment, herbs, shrubs, and wetlands) is possible. |