When classes look alike: leveraging multifrequency SAR and TIR for separating early-stage sea ice types

Accurate characterization of sea ice types in synthetic aperture radar (SAR) images remains a challenge due to the dynamic nature of sea ice and the spectral similarity between certain ice classes. This study investigates the potential of multisource satellite data for refining sea ice discrimination, with particular emphasis on separating open water and early-stage sea ice types, specifically new ice and young ice, which often show similar signatures in radar imagery. By comparing C- and L-band SAR data, we examine the complementarity of different data sources. In addition, we incorporate thermal infrared brightness temperatures from Landsat 8/9 and analyse class-wise histograms of radar intensity and thermal signatures. While early-stage sea ice classes and open water exhibit substantial internal variability and overlapping SAR signatures, thermal data provide enhanced contrast between these classes. The comparatively stable backscatter behaviour of level and deformed ice is also characterized to establish a reference framework. Overall, the results emphasize the complementary strengths of diverse satellite sensors in advancing sea ice type characterization.