Analysis of snow changes in alpine regions with X-band data: electromagnetic analysis and snow cover mapping

High-resolution and high-frequency COSMO-SkyMed images acquired in the period between 26 April 2010 and 5 April 2011 over the test site in South Tyrol (Northern Italy) offer the chance to analyze the snow changes and to infer information about the physical characteristic of the snow. The X-band sensitivity to snow status was analyzed using two different electromagnetic approaches: 1(st) Radiative Transfer model, IEM, and a multi-scattering and multi-layer snow scattering model. It results that the description of the dry snow requires a more detailed information about the underlying layers to extract information about the volumetric and ground contribution of the snowpack. The comparison between multi-scattering and multi-layer model predictions and SAR data indicates a better agreement between the measurements and co-polarized backscattering values with respect to the cross polarized backscattering values which appears to be lower than expected indicating that a detailed description on the land surface parameters might help to generate more accurate simulations. The change detection technique for the detection of wet snow was investigated to obtain snow cover map. By using the threshold of -3dB the two frequency distributions for the snow and no-snow areas, are well-separated only in the case of wet snow areas; on the contrary it results that, at the beginning of the melting season, the frequency distribution still overlaps. From the comparison with LANDSAT 7 ETM+ derived snow map, the omission error of 9.11% and the commission error of 1.84% confirm the typical underestimation of snow cover from SAR images with respect to optical images.