Comparison of snow cover resulting from satellite and a numerical prediction model in Iran

The prediction of snow cover is fundamental in terms of the snow melting role for runoff prediction. The purpose of this study is the feasibility of using snow cover output of the WRF model with the aim of predicting snow melting. Since there are a few snow monitoring stations in the mountainous areas of Iran, satellite data are suitable widespread observation data. Accordingly, the extraction of snow cover data using MODIS sensor was addressed. Afterwards, it was compared to the snow cover of the WRF model outputs for a case study. Finally, the results were assessed using the contingency table. The studied area is located on Chalous Road, during January 28 to January 30, 2017 with heavy snowfall. Synoptic analysis indicates a strong pressure gradient in surface level associated with cold air in upper layers. Overlapping of satellite snow cover data with model output shows the linear correlation coefficient 0f 0.7 at a significant level of 0.01 in a 90 × 78 points network, 5 km resolution. The probability of detecting snow on 3 day average is 87% and the Probability of false detection is about 20%. Heidke Skill Score is about 0.7 for clear sky days (January 29th and January 30th.) which is the best result. In January 28th. The Heidke Skill Score is reduced to 57%. It may be the result of inaccurate satellite observation due to cloudiness sky or inaccurate forecasting. Since there is no ground station in the selected area, one cannot properly comment on this.