SWE retrieval in Alpine areas with high-resolution COSMO-SkyMed X-band SAR data using Artificial Neural Networks and Support Vector Regression techniques

Abstract

The potential of satellite Synthetic Aperture Radar (SAR) sensors for Snow Water Equivalent (SWE) retrieval in Alpine areas is assessed in this study. X-band HHpolarized SAR backscatter from 2012-2015 images acquired by the COSMO-SkyMed constellation over the South Tyrol province in northern Italy is compared with SWE in-situ measurements and nivo-meteorological station records. The resulting relationship is compared with simulations based on the Dense Media Radiative Transfer – Quasi Mie Scattering (DMRT – QMS) model. Artificial Neural Networks (ANN) and Support Vector Regression (SVR) machine learning techniques are trained and used for SWE retrieval from COSMO-SkyMed data. Good accuracy and small computational cost are observed for both ANN and SVR. The resulting SWE maps agree with snow conditions measured in-situ.