SPASS – new gridded climatological snow datasets for Switzerland: potential and limitations

Abstract. Gridded information on the past, present, and future state of the surface snow cover is an indispensable climate service for any snow-dominated region like the Alps. Here, we present and evaluate the first long-term gridded datasets of daily modeled snow water equivalent and snow depth over Switzerland, available at 1 km spatial resolution since 1962 (spanning 60+ years). These climate-oriented datasets are derived from a quantile-mapped temperature index model (OSHD-CLQM). The validation against a higher-quality but shorter-duration dataset – derived from the same model but enhanced with data assimilation via an ensemble Kalman filter (OSHD-EKF) – shows, on the one hand, good results regarding bias and correlation and, on the other hand, acceptable absolute and relative errors except for ephemeral snow and for shorter time aggregations like weeks. An evaluation using in situ station data for yearly, monthly, and weekly aggregations at different elevation bands shows only slightly better performance scores for OSHD-EKF, highlighting the effectiveness of the quantile-mapping method used to produce the long-term climatological OSHD-CLQM dataset. For example, yearly maps of gridded snow depth compared to in situ data demonstrate an RMSE of 25 cm (20 %) at 2500 m and of 1.5 cm (80 %) at 500 m. For monthly averages, these numbers increase to 30 cm (25 %) and 3 cm (100 %), respectively. A trend analysis of yearly mean snow depth from these gridded climatological- and station-based data revealed very good agreement on direction and significance at all elevations. However, at the lowest elevations the strength of the decreasing trend in snow depth is clearly overestimated by the gridded datasets. Moreover, a comparison of the trends between individual stations and the corresponding grid points revealed a few cases of larger disagreements in the direction and strength of the trend. Together these results imply that the performance of the new snow datasets is generally encouraging but can vary at low elevations, at single grid points, or for short time windows. Therefore, despite some limitations, the new 60+-year-long OSHD-CLQM gridded snow products show promise as they provide high-quality and spatially high-resolution information on snow water equivalent and snow depth, which is of great value for typical climatological products like anomaly maps or elevation-dependent long-term trend analysis.