Abstract:Based on the surface historical observation data, the statistical analysis of the snow to liquid ratio of the typical winter snowfall events in China is conducted. Then the snow to liquid ratio and fresh snowfall depth forecast models are built using the Cobb method, ECMWF IFS model forecast and the gridded precipitation and its type grid forecast from the National Meteorological Centre. The statistical analysis shows that the snow to liquid ratio in the snowfall events in China is widely distributed, with a minimum of 0.3∶1 and a maximum of 100∶1. Only about 4% of the snow to liquid ratio records are 10∶1. The snow to liquid ratio has obvious regional and seasonal characteristics. The snow to liquid ratios in Xinjiang, eastern Northwest China, North China and Northeast China are larger than those in other regions. The snow to liquid ratio in winter is larger than that in spring and autumn. Based on the modified Cobb method, which can identify the cloud according to topography, the snow to liquid ratio is firstly constructed based on the statistical relationship between the temperature in the cloud and the snow to liquid ratio and the related vertical velocity. The snow to liquid ratio and fresh snowfall depth forecasted by modified Cobb method can reflect the spatial characteristics of the actual snow to liquid ratio in the precipitation process, compared with the original Cobb method. The fresh snowfall depth forecast performance is obtained by combining the snow to liquid ratio forecast model and the National Meteorological Centre’s gridded snowfall forecast and it significantly improves the fresh snowfall depth forecast compared with that by the ECMWF IFS model. In particular, the ability to predict fresh snowfall depth over 20 cm has been significantly improved, and the TS score increase rate of the fresh snowfall depth forecast built in this paper relative to that by the ECMWF IFS model is basically over 10%. At present, this technology has been applied in the National Meteorological Center, providing strong support for winter snowfall forecasting.