|
In this work, the characterization and extraction of snowpack parameters from X-band SAR imagery has been addressed. A preliminary sensitivity analysis was carried out by exploiting datasets of snowpack parameters (depth, density, snow grain radius, temperature and wetness) collected from the available meteorological stations on two test sites in the Italian Alps. This is a crucial step, since it provides indications on the sensitivity of the input features (i.e., backscattering coefficients and ancillary data) to variations in the target snow parameters. X-band data has been found to contribute to retrieval of the snow water equivalent under specific conditions, i.e., that the snow cover is characterized by a snow depth of roughly 60-70 cm (snow water equivalent <100-150mm) and with relatively large crystal dimensions. After this phase, the retrieval process is addressed. The method is based on a Neural Network retrieval algorithm trained by using a DRTM electromagnetic model in order to estimate the snow water equivalent. The proposed approach also makes use of the threshold criterion for detecting the wet snow cover extent on which the retrieval cannot be performed. The method has been developed and calibrated on the Cordevole plateau located in the Dolomites, Eastern Italian Alps, where ground data collected by the Avalanche Center in Arabba and meteorological data measured by a network of
automatic stations were available. The method was then validated on a second site located in South Tyrol region (Eastern
Italian Alps), where also manual and automatic ground measurements of snow parameters were available. The activity was carried out in the framework of two projects funded by the Italian Space Agency (HYDROCOSMO and SNOX) for the exploitation of X-band satellite SAR data for the analysis and characterization of snow in mountain areas.
|
