Abstract
Snow accumulation and melt are crucial for the hydrological cycle strongly contributing to runoff, groundwater storage, flooding, wet-snow avalanches and contaminant release. Monitoring of snowmelt timing is challenging, given the scarceness of in-situ sensors. Hydrological models represent a valuable tool, but with intrinsic limitations due to the usage of simplified assumptions and the scarceness of input meteorological data. Remote sensing has shown to be a valuable tool for snow monitoring, providing spatially distributed information in remote and difficult to access areas. In particular, the recent launch of the Copernicus Sentinel satellite missions has opened new opportunities due to the systematic collection of high spatial resolution images with high acquisition frequency. However, the state-of-the-art methods do not properly exploit this large amount of heterogenous information. Hence, we aim at developing novel approaches being able to exploit the rich multi-temporal and multi-source data to monitor the snowmelt status. This is achieved by exploiting the multisensory information separately first and by developing data fusion techniques in order to improve the snowmelt retrieval. We first present a physically-based approach that exploits Synthetic Aperture Radar (SAR) time-series for snowmelt retrieval. This is achieved by observing the relationship between the presence of liquid water content inside the snowpack and the backscattering behavior at C-band (Sentinel-1) [Marin et al., 2019]. Secondly, we investigate changes in snow cover fraction (SCF) by merging low-resolution and high-resolution data. SCF is usually obtained by performing a linear regression given the normalized difference snow index (NDSI) [Salomonson and Appel, 2006]. We propose to improve the SCF retrieval from low resolution data by adaptively exploiting the high-resolution information. All the information obtained by SAR- and multi-spectral data can be fused for obtaining the correct timing of snowmelt on-set, with relevant implications on hydrological evaluations.