Application of the WRF model in simulating snow depth in the northern part of Iran

Snow depth modeling serves several purposes, including weather prediction, water storage estimation, flood forecasting, and assessing energy production potential. The WRF model is commonly used for snow depth simulation. The studied area is the northern part of Iran. For a more detailed investigation and to eliminate the effects of land-sea interaction on the implementation of the WRF model, the northern part of Iran was divided into three separate regions. It should be noted that the results obtained from this research are evaluated separately for each area. These three regions include northeast, north and northwest. The simulations were conducted for 48-hours using two nests with 9 and 3 km resolutions, respectively. Also, in the model settings, 41 levels are considered in the vertical direction, and the pressure at the highest level is 50 hPa. In these simulations, the fifth generation ECMWF reanalysis (ERA5) data with a spatial resolution of 0.25 degrees and 6-hours’ time step was used as the initial and boundary conditions. The optimal setup was determined based on the Taylor diagram. It involves specific parameterization schemes for different regions: Tiedtke scheme is used for the north and northeast regions, and OSAS scheme for the northwest region, to parameterize convection. WSM-3 scheme is used for microphysics in all three regions. QNSE/QNSE scheme is applied to the north and northwest regions, while YSU/MM5 scheme is used in the northeast for boundary/surface layer parameterization. For radiation, New Goddard and Dudhia schemes are best suited for long-wavelength and short-wavelength respectively. NOAH-MP scheme is also used for surface parameterization across all three regions. Daily snow depth values from the model compared with the observed station data using statistical indices such as RMSE and Bias. In this study, the data of 62 synoptic stations (12, 20 and 30 stations respectively in northeast, north and northwest regions of Iran) have been used to extract snow depth data. By applying the optimal configuration in all three regions, the results showed that the amount of error in the northwest is lower than the other two regions. The results of the investigation in each area showed that the model performs better in lower snow depth values (north region) and in the northeast, the model performance depends on the station height, which seems to be more accurate in the stations with lower altitudes. In the north and northeast regions, there are overestimates in most of the stations (with the bias of 0.115 and 0.264 m, respectively). In the northwest, unlike the other two regions, the model has underestimated the snow depths in most stations with the bias of -0.016 m. This could be due to an overestimation of snow albedo in the WRF model, as suggested by previous research. The RMSE error for all stations in the northwest is less than 0.018 m, which is lower than the other two regions of the north and northeast where the RMSE is 0.195 and 0.143 m, respectively. The differences between the model and station data could be due to several factors, including the inaccuracy of the model’s input data, the model’s limitations in accurately simulating snow depth, changes in snow albedo, environmental influences like local wind and sunlight, the area’s topography, and the spatial scale differences between the model and the stations.