Sea ice classification during freeze-up conditions with multifrequency scatterometer data

Helicopter-borne radar backscatter measurements are analyzed with respect to a multifrequency classification approach of sea ice. Measurements were carried out over the Arctic Ocean during August and September 2007 and represented unusually warm freeze-up conditions. Radar cross sections (RCSs) of totally ice-free wind-roughened water are used in combination with an ocean surface theoretical backscattering model for the calibration. The calibrated RCS sigma degrees agrees within 1 dB with nearly simultaneous Envisat Advanced Synthetic Aperture Radar measurements and literature values. Sea ice was classified using a Bayesian maximum likelihood approach. By including information from simultaneous infrared and visible video imagery of sea ice, four different surface types of sea ice could be identified in the resulting sigma degrees: old ice, gray ice, nilas, and open water. The most reliable classification was obtained through combination of copolarized C-, X-, and Ku-band data. The results degraded by only 7% in the case where the X-band information was dropped. On the other hand, a combination of the C- and X-bands or the X- and Ku-bands yielded a degradation of 13%. Given the remaining uncertainties in the approach, for sea ice classification during summer/fall conditions, our results suggest the complementary use of two of these three frequency bands instead of relying on just one frequency band