Estimated Accuracy of the Simulation of the Precipitation Forming Snowpack in the Urals according to Global Numerical Weather Prediction Models and the ERA5 Reanalysis

The paper considers the spatiotemporal distribution of accumulated cold-season precipitation and the accuracy of its simulation according to the ICON and GFS/NCEP numerical weather prediction (NWP) models. The forecasts of the NWP models were obtained with a daily lead time and aggregated to calculate monthly total precipitation. They were compared with the observational data from 189 weather stations, as well as with the ERA5 reanalysis data. The calculations were performed for the territory of the Urals and adjacent parts of European Russia and Western Siberia for the period from October 2019 to April 2024. It was found that the GFS/NCEP model in spring and ERA5 reanalysis during the entire cold season systematically overestimated the amount of precipitation. The root-mean-square error of the calculation of monthly total precipitation turned out to be the lowest for the ICON model data. Just these data can be recommended for calculating snow accumulation, although the ICON model underestimates the amount of precipitation for the heaviest snowfalls. The NWP models substantially overestimate precipitation for weather stations located in open areas, which may be caused by blowing-related underestimation of the measured precipitation amount. The average absolute forecast error for weather stations located in the Eastern Ural region is lower than for those in European Russia and the Urals.