Regional Mountain Conference

In this study, we present first results of snow simulations using the snow-hydrological model openAMUNDSEN including the ensemble-based assimilation of satellite-based snow cover data. Our ensemble-based snow modelling operates in a sequential prediction–update cycle and is based on a particle filter, enabling repeated integration of different type of snow observations. The snow model ensemble is generated by the perturbation of meteorological forcing variables. The assimilation framework is developed as a Python toolkit with a high-performance software architecture focusing on parallelization, supporting efficient regional-scale ensemble simulations. In a first model experiment, we assimilate wet snow maps based on Sentinel-1 data and fractional snow cover maps from Sentinel-2 imagery. Test site is the alpine headwater catchment Rofental, Tyrol, Austria (98.1 km2). Preliminary results point out that distributed snow cover simulations using openAMUNDSEN benefit from the ensemble-based assimilation of satellite-based snow cover data. Further testing of the assimilation framework is required to optimize existing assimilation parameters and facilitate regional-scale applications.