Microwave remote sensing of snow depth on Arctic sea ice

Snow on sea ice affects the radiative balance and knowledge of snow depth it is required when determining sea ice thickness from freeboard measurements based on altimeter observations e.g. from CryoSat-2. Remote sensing of snow depth based on microwave satellite observations was first suggested by Marcus and Cavalieri (1998) for Antarctic sea ice based on SSM/I (Special Sensor Microwave / Imager) data. It is based on the the gradient ratio of the 18 and 37 GHz channels, vertically polarized, i.e., the difference between the two brightness temperatures, normalised by their sum, denoted (GRV18,37). Transferring the procedure to Arctic ice conditions has been attempted, but turned out to be difficult because it is hampered by several complications such as the presence of more distinct ice types in the Arctic, and more sea ice deformation. During the various flights in the context of the Operation Ice Bridge (OIB) campaigns, also snow depth on sea ice has been observed with a snow radar and laser altimeter. Here, we use OIB snow depth data of the years 2009-2011 (about 500 observations) for a systematic investigation of all gradient ratios between daily averages of the AMSR-E (Advanced Microwave Scanning Radiometer for EOS) channels near 6, 10, 18, 23, 37 and 89 GHz, both polarizations for correlation with the snow depths. As initial result we find the strongest correlations with values around -0.7 for the vertically polarized gradient ratios at 6 GHz channel with those at 10,18 and 23 GHz and for GRV(18,10). A similar study is performed for AMSR2 and OIB data 2012-1015 (about 300 observations). Here we find even stronger correlations up to -0.88 at the same channel combinations. These are promising results in view of systematically retrieving snow depth on Arctic sea ice. A multilinear regression and a more detailed physical analysis of the data will be presented, also distinguishing cases of first-year and multi-year ice and discussing potential systematic influences in the OIB data.