Abstract
The EURO-CORDEX initiative provides a comprehensive set of regional climate simulations for the present and future European climate. We assess how well snow cover, a crucial component of the climate system, is represented in this model ensemble. We consider output from 11 regional climate models (RCMs) at 12 km resolution, driven by ERA-Interim and 10 different general circulation models, and compare it with model- and remote-sensing-based reference data sets for the periods 1990–2008 and 1985–2008, respectively. Elevation-resolved evaluations for four mountainous regions indicate that for mean winter snow water equivalent (SWE), the reanalysis-driven RCM ensemble mean agrees well with reference data. However, model spread becomes substantial (ca. ± 50%) at the highest elevations of the regions. For the number of snow-covered days (SCD), the agreement of individual models with reference data is generally better. With GCM-forcing, simulated mean winter SWE and SCD biases are typically larger in all regions except for Scandinavia. Part of this bias can likely be attributed to an overestimation of winter precipitation in the RCM runs. The analysis also reveals that some models indicate perennial snow accumulation at much lower elevations than suggested by proxy data. This bias is likely related to inaccuracies in atmospheric forcing (aforementioned precipitation biases) and shortcomings in the representation of snow cover in the RCMs. Overall, and after spatial masking of spurious grid cells, both the reanalysis-driven and the GCM-driven EURO-CORDEX 12 km ensemble are able to represent the main spatio-temporal characteristics of European snow coverage. The future changes in snow cover characteristics will be the subject of a follow-up study.