Microwave emission analysis over long-term time series of snow data collected in Italian Alps

This paper will investigate the effect of layered snow in alpine regions on microwave emission at Ku and Ka bands, using both experimental data and model simulations. A multilayer dense-medium radiative transfer model (DMRT), was implemented under the quasi-crystalline approximation (ML-QCA), to account for the effects of snow layers on the emission from dry snow covers. The model then evaluated the sensitivity of two microwave indices, based on frequency and polarization combinations, to snow parameters. Model simulations were compared with radiometric dual frequency/polarization measurements of snow covers, collected during long-term experiments carried out over three winter seasons between 2007 and 2011 in the Eastern Italian Alps. This comparison has the twofold purpose of validating the model with experimental data and verifying the influence of snow layering on microwave emission and related frequency and polarization indices. The wide variations of snow characteristics over several winter seasons allowed for an extended validation of the model, which was demonstrated to account for the complex stratigraphy (up to 15 layers) of snow. The measured brightness temperatures at Ku and Ka bands were compared with those simulated through the multi (ML-QCA) and single-layer (SL-QCA) models, by using the observed snow parameters as inputs. In the case of SL, we used the average value of all layers weighted for the layer thickness. The results showed that the ML-QCA model was better correlated to the radiometric the radiometric measurements than the SL-QCA. A direct comparison of measured and simulated data showed that the slope of correlation for the single layer ranged between 0.4 and 0.5, with determination coefficient lower than 0.3. The slope in the multi-layer approach ranged between 0.7 and 1.0, with determination coefficients between 0.51 and 0.75, and Root Mean Square Error (RMSE) between 11K and 15K.