Assessment of snow model uncertainty using a large number of openAMUNDSEN snow model configurations: A study from the Berchtesgaden National Park (Germany)

While the assessment of climate model uncertainty is well established, the uncertainty originating from the selection of a surface snow model usually only receives little attention. However, a better understanding of snow model uncertainty currently becomes more and more important, as novel climate model data at the kilometer-scale, innovative downscaling techniques, and increasing computational capacities are among the elements that pave the way for a new phase of high resolution and physically based climate change impact studies assessing cryospheric changes in complex mountain areas. To investigate the uncertainty induced by the selection of the snow model configuration, we simulate the seasonal snow cover in the mountain area of the Berchtesgaden National Park (Germany) under historical conditions (10/2013 - 09/2023) and for a 10-year period characterized by a 1°C warming. Therefore we use a large number of openAMUNDSEN snow model configurations (n = 108) with T-Index, enhanced T-Index as well as energy balance based snowmelt methods, varying land cover maps and spatial resolutions. Forcing data for the 10-year warming period is constructed using the stochastic bootstrap resampler (climate generator) available within the openAMUNDSEN modelling framework. Prior to the estimation of snow model uncertainty, we evaluate the snow model results using satellite-based snow data. Our results suggest that differences in key snow metrics such as snow cover duration and snow disappearance day can be in the same range as the impact of a 1°C warming. The results also support the identification of critical snow model settings that need to be considered, in particular, when using energy balance instead of degree-day snow models to investigate climate change impacts on snow hydrological processes in complex mountain terrain.