We have compared the time series of C-band HHpolarization backscattering coefficients (σ • ) of the Baltic Sea land-fast level ice with results from a 1-D high-resolution thermodynamic snow/ice model (HIGHTSI). The σ • time series were obtained from ENVISAT synthetic aperture radar (SAR) images. The study period was from the middle of the winter to the early melt season, February 3-April 7, 2004. Due to the large incidence angle range of the SAR images, the σ • values were divided into three subseries. In general, the HIGHTSI results greatly helped to interpret the σ • behavior with changing ice and weather conditions. The modeled snow-surface temperature, cases of snow melting, and evolution of snow and ice thickness were related to the changes in σ • . Equally useful information could not be obtained solely on the basis of large-scale atmospheric models.
Realistic forcing data for HIGHTSI were available in the form of coastal-weather observations and model results of the European Centre of Medium-Range Weather Forecasts (ECMWF). The latter make it possible to apply HIGHTSI in the interpretation of SAR data from all ice-covered seas. There were some cases where detailed ground truth, combined with theoretical σ • modeling, would have been needed for interpretation of the σ • trends. A very interesting observation was the large variation of level ice σ • with changing weather conditions, which complicates automatic classification of the SAR images, and thus, the algorithms must be tuned for different ice conditions. The HIGHTSI model could act as an indicator of various ice conditions for algorithm development.Index Terms-Radar scattering, sea ice, synthetic aperture radar (SAR).