Snow depth in convection-permitting regional climate model simulations over southern Germany - ready for impact-related research?
- Center for Earth System Research and Sustainability (CEN), Universität Hamburg, Hamburg, Germany (benjamin.poschlod@uni-hamburg.de)
Snow dynamics are affecting the climate system, water cycle, ecology, human society, and infrastructure. Furthermore, the representation of snow on the land surface within regional climate models is crucial for the mass and energy balance in the modelled climate. Simulated daily snow depths of two high-resolution regional climate models, the WRF at 1.5 km resolution and the COSMO-CLM (CCLM) at 3 km resolution both driven by ERA5 reanalysis data are evaluated with 83 station observations in southern Germany during 1987 – 2018. Furthermore, based on the atmospheric output of CCLM, the hydrometeorological snow model AMUNDSEN is run at the point scale of the climate stations. In addition, the ERA5-Land dataset (9 km) complements the comparison as state-of-the-art reanalysis land surface product driven by the same atmospheric conditions of ERA5. ERA5-Land shows considerable deviations of grid cell elevation and station elevation (mean absolute error: 93 m) and moderate biases in air temperature (-0.3 °C) and precipitation (+19.1 %). However, there is a strong positive bias in mean winter snow depth (+3.5 cm) and snow cover duration (+33.3 d). The WRF simulation features a cold bias of -1.2 °C and moderate bias in precipitation (-5.2 %) during winter. This results in a slight overestimation of snow depth (+0.4 cm) and snow cover duration (+6.6 d). The CCLM strongly underestimates snow depth (-2.5 cm) and snow cover duration (-19.8 d), albeit a very good reproduction of air temperature (+0.0°C) and precipitation (+9.7 %). AMUNDSEN reverses the underestimations of the CCLM to an overestimation of snow depth (+2.2 cm), however improving the reproduction of snow cover duration (+6.4 d). All models fail to skilfully predict white Christmas.
Extremes of snow dynamics such as annual maximum snow depths, maximum daily snow accumulation and melting are not well reproduced by ERA5L and CCLM. WRF and AMUNDSEN can improve the representation of extremes but still with considerable limitations.
In conclusion, the simulation of snow depths with WRF and AMUNDSEN can benefit from the finer resolution of the topography in the high-resolution climate models compared to ERA5-Land. However, even though driven by the same large-scale atmospheric conditions of ERA5, the four snow depth simulations vary by a huge margin. The high spatial resolution of convection-permitting climate models shows potential in reproducing the winter climate in southern Germany. However, the uncertainties within the snow modelling prevent a further straightforward use for impact research. Hence, careful evaluation is needed before any impact-related interpretation of the simulations, also in the context of climate change research.
How to cite: Poschlod, B.: Snow depth in convection-permitting regional climate model simulations over southern Germany - ready for impact-related research?, EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-1064, https://doi.org/10.5194/egusphere-egu23-1064, 2023.