Assessment of sentinel-1 c-band SAR snow depth assimilation in operational hydrological modeling of high-alpine catchments

Monitoring and forecasting mountainous runoff is essential, as snowmelt provides vital freshwater resources while also posing flood and drought risks to downstream regions. However, spatiotemporal modeling of the high-alpine snowpack remains challenging, due to shortcomings in their process-based descriptions and substantial deficits in input data of operational hydrological models. To address this issue, we evaluate the potential of assimilating 500 m resolution Sentinel-1 C-band Synthetic Aperture Radar (SAR) snow depth retrievals in operational hydrological modeling of several high-alpine catchments in Austria. We first evaluated the Sentinel-1 snow depth product with ground-based measurements from 27 AWS (automatic weather stations) in Austria (r = 0.88, MAE = 22.3 cm) and annual photogrammetric snow depth surveys in the Dischma catchment in Switzerland (r = 0.72, MAE = 69 cm). We then forced the conceptual semi-distributed hydrological model COSERO with the 1 km 2 gridded meteorological INCA product, which are both operationally used for hydropower and flood forecasting, and model thirteen high-alpine catchments in Austria between 2016 and 2024 at 500 m spatial and hourly temporal resolution. The assimilation of Sentinel-1 snow depths substantially improved overall model performance relative to modeling with uncorrected INCA data. For the entire 2016–2024 modeling period, this improvement is indicated by a median increase of NSE by 9.9% (0.75 to 0.82) and KGE by 34.1% (0.62 to 0.83), and a reduction of the median runoff deficit from −23.1% to −3.9%. The data assimilation in the operational modeling setup proved more effective during snow-poor years, a finding that may become more relevant under an increasingly warming climate in the European Alps. The Sentinel-1 snow depth product showed good agreement with ground measurements and snow cover maps (MODIS). Its assimilation remedied the winter precipitation deficit in INCA, while improving spatiotemporal snowpack simulation and enhancing snowmelt runoff prediction. This more realistic representation of the mountainous snowpack may benefit simulation of other hydrological variables including evapotranspiration, soil moisture, and groundwater recharge. The approach also shows promise for modeling remote, ungauged basins globally. These results demonstrate the value of integrating satellite-derived snow information for operational alpine hydrological modeling and water resource assessment.