Comparison of in situ and airborne measurements of multiyear sea ice thickness with dual-frequency, polarimetric SAR observations

Abstract

First results are presented from a comparison of in situ and airborne measurements of sea ice thickness to C-and Xband polarimetric synthetic aperture radar images acquired over the Lincoln Sea in March-May 2012. Operation IceBridge data indicate a moderate correlation between ice surface roughness and thickness in this multi-year ice regime, motivating efforts to derive ice thickness from SAR data. C-band backscatter shows moderate correlations to in situ and airborne ice thicknesses (r ≈ 0.6). At X-band correlations with airborne ice thickness data are much lower. For all polarizations the relationship with ice thickness levels off at thicknesses greater than 5 m, where ice is heavily deformed. Current research efforts are focused on exploiting polarimetric parameters to discriminate thick deformed ice to support future efforts to develop a thickness retrieval algorithm for undeformed ice up to 5 m thick.

A Meta-Analysis of Sea Ice Monitoring Using Spaceborne Polarimetric SAR: Advances in the Last Decade

“…However, there were no strong correlations between ice thickness and the C-band crosspolarized ratio or X-band copolarized correlation as reported in [163]. In contrast, Shannon entropy and Freeman-Durden volume scattering were considered useful features for estimating SIT [163]. In [164], the correlation between fast ice thickness and each of the three simulated HCP features, the correlation coefficient of RH and RV, the conformity coefficient and the degree of polarization, was found to be higher than 0.9.…”

“…In [39], [77], [161], and [162], copolarized correlation derived from X-band SAR data and the cross-polarized ratio extracted from C-band SAR data were found to be ideal features for retrieving thickness of MYI and FYI. However, there were no strong correlations between ice thickness and the C-band crosspolarized ratio or X-band copolarized correlation as reported in [163]. In contrast, Shannon entropy and Freeman-Durden volume scattering were considered useful features for estimating SIT [163].…”