This study addresses snow depth overestimation and temperature underestimation in the WRF/Noah-MP regional climate model by producing eight sensitivity simulations at a 10 km grid spacing for the 2001-2006 period, driven by the ERA5 reanalysis. The experiments differ either in the parameterization of snow-related processes within the Noah-MP land surface model, the physical parameterization schemes of the atmospheric model (planetary boundary layer, surface layer, microphysics), or the WRF model version (v4.2 or v4.3, with the latter incorporating snow-related code enhancements). The best-performing configuration is further tested by conducting a hindcast simulation spanning an extended climatological period (1985-2010). The study area includes parts of Western, Central, and Eastern Europe, focusing on the Carpathian Basin. Snow depth is evaluated using in situ observations from low-elevation stations and ERA5 reanalysis data. Temperature and precipitation biases are assessed using the CARPATCLIM dataset, consisting of gridded surface meteorological observations. With the appropriate choice of physical parameterizations in the WRF/Noah-MP model, the median snow depth bias is reduced from 3.4 cm/day to 0.7 cm/day across all stations during the 2001-2006 period (November to March), with a corresponding decrease in RMSE from 8.4 to 3.9 cm/day and an increase in the correlation coefficient from 0.68 to 0.82. Concurrently, the underestimation of average daily minimum temperatures decreases from 2.5 degrees C to 1.6 degrees C over a plain subregion within the Carpathian Basin. Further efforts are required to eliminate the cold bias above snow cover in the WRF/Noah-MP model. The results suggest that exploring sensitivities to physical processes within the land surface scheme is equally important as testing different physical parameterizations in the atmospheric model itself.
