Evaluation of modelled snow depth and snow water equivalent at three contrasting sites in Switzerland using SNOWPACK simulations driven by different meteorological data input

• Snow depth HS and SWE can successfully be modelled with the snow model SNOWPACK.
• SNOWPACK was driven with input data from standard meteorological measurements.
• We validated the model at three stations in Switzerland at different elevations.
• Incoming long-wave radiation and albedo can successfully be parameterized.
• A proper treatment of precipitation measurements is crucial for modelling HS and SWE.

The knowledge of certain snow indices such as the number of snow days, maximum snow depth and snow water equivalent or the date of snow disappearance is important for many economical and ecological applications. However, snow data are frequently not available at the required locations and therefore have to be modelled. In this study we analyse the performance of the physically based snow model SNOWPACK to calculate the snow cover evolution with input data commonly available from automatic weather stations. We validated the model over several years at three very diverse stations in Switzerland: Weissfluhjoch (2540 m a.s.l.), Davos (1590 m a.s.l.) and Payerne (490 m a.s.l.), where snow depth and the full radiation balance are measured in order to assess the uncertainties induced by the parameterizations of radiation fluxes and by the use of uncorrected precipitation measurements. In addition, we analysed the snow water equivalent at the high-alpine station Weissfluhjoch. The results demonstrate that the radiation balance, which is often measured incompletely, can successfully be parameterized and has an unexpectedly small impact on the modelled snow depth. A detailed analysis demonstrates that an adequate precipitation correction decreases the mean absolute percentage error by 14% for snow depth at the alpine and high-alpine stations and by 19% for snow water equivalent at Weissfluhjoch. The low altitude station Payerne (ephemeral snow conditions) revealed a high sensitivity with regard to the temperature threshold to distinguish solid from liquid precipitation. The analysis further suggested a high sensitivity to ground heat fluxes for ephemeral snow covers. Overall, the daily snow depth could be modelled with a mean bias error of less than − 8 cm at all sites, whereas the mean bias error for the snow water equivalent was less than − 55 mm w.e. at Weissfluhjoch.