Characterization of sea ice types using synthetic aperture radar

Abstract

The results of an investigation into the use of synthetic aperture radar (SAR) imagery for sea ice-type discrimination are presented. X- and L-band dual-polarization SAR data of Beaufort Sea ice were examined using manual interpretation techniques to determine which channel provides the most information. Quantitative methods for ice-type discrimination also were explored by statistical parameterization of these data. Various statistical tests, both parametric and nonparametric, were applied to evaluate the utility of the parameters for machine interpretation of SAR ice data. The results obtained indicate that, under winter ice conditions, X-nband is superior to L-band for discriminating various ice types. Also, imagery obtained at small incidence angles shows greater tonal variation between ice types than that obtained at larger angles. Of the quantitative measures evaluated, mean and standard deviation appear to be the most valuable. Examination of quantities involving higher order moments indicates that a better understanding of the SAR imaging process is required before these measures can be utilized successfully.