Abstract
The seasonal snowpack of the mediterranean Andes of Chile holds a critical role in supplying water for agriculture, ecosystems, industries and drinking water for up to eight million in the city of Santiago alone. Despite the above, its characterization remains highly uncertain due to the scarcity of in-situ observations and the complex topography of the region. In this context, physically based snow models provide an opportunity to produce high resolution estimates of the main snowpack variables: snow water equivalent (SWE) and snow depth (SD). However, from recent modeling efforts in this mountain range, the performance of these simulations is hampered mostly by uncertainty in precipitation amounts, which are currently not well resolved by global reanalysis datasets nor by station-based gridded products. This study aims to address the forcing uncertainty problem and enhance snowpack simulations by assimilating Sentinel-1 C-band synthetic aperture radar (SAR)-derived snow depth and wet snow retrievals. These observations are assimilated into the Canadian Hydrological Model, a variable resolution, physically based snowpack modeling platform. The study domain comprises two high mountain basins of about 600 square kilometres located in the Chilean Andes, within latitudes 33-34 degrees South. To assess the value of the Sentinel-1 data, we conduct a control experiment where only snow-covered area (SCA) is assimilated. Then, we carry out the assimilation of snow depths and wet snow maps. Finally, the assimilation experiment using Sentinel-1 products is compared to the control (SCA only) through evaluation against ground truth data (terrestrial LiDAR, Pléiades retrievals and station measurements of SWE and SD). This allows to answer whether these new sources of snowpack information constitute an improvement over the more traditional SCA assimilation, or if more work is needed to make these Sentinel-1 products usable in data assimilation over the Andes. The combination of continuous, all-condition remote sensing data like C-band SAR and a high-resolution snow model could be a step forward in water resources monitoring, particularly in data scarce regions like the Andes Cordillera.