Discrimination of open water from sea ice in the Labrador Sea using quad-polarized synthetic aperture radar

Abstract A method to differentiate open water from sea ice using quad-polarization C-band synthetic aperture radar (SAR) imagery is presented, based on the roughness properties of the sea surface scattering facets and the polarization ratio (PR) between VV and HH polarization backscatter from the SAR images. In quad-polarization SAR images, the normalized radar cross section (NRCS), which is linked to sea surface roughness, can be used to derive the orientation angle shift and its standard deviation, as well as to compute the measured polarization ratio, PR. An X-Bragg backscatter model is explored to calculate a theoretical look-up table for the polarization ratio as a function of incidence angle, dielectric constant, orientation angle shift and its standard deviation. Thus, ocean surface backscatter can be identified as open water or sea ice by comparing the measured PR with a theoretical model for PR. The proposed method is validated by comparison to collocated optical video images and manual visual interpretations. Estimates of the size, shape and locations of open water from our method are shown to be in good agreement with observed data. The overall accuracy of our algorithm is about 96%. The classification accuracy for sea ice and open water is about 96% and 94%, respectively.