Abstract:Based on surface observation and sounding data from 2001 to 2013, the spatial and temporal distribution of snowfall occurring at surface air temperature of 0-2℃ (hereinafter referred to as critical surface temperature condition) in China is analyzed, and comparative analysis of vertical thermal characteristics is made between snow and rain under the above condition, then a decision tree discriminant method is introduced to discriminate snow from rain. The results show that the frequency of snow under the critical surface air temperature conditions is generally higher than that of rain and sleet, and it mainly distributes in the central and eastern regions between southern North China and northern Jiangnan Area, with an average annual number of 7.69-15.38 stations. Combined soundings of snow and rain show maximum temperature difference around 650 hPa, more obvious when T2 m (surface air temperature) is lower. Maximum humidity difference lies around the surface, more obvious when T2 m is higher. A melting layer above the surface exists more frequently in rain than in snow. The melting layer of rain is mainly located in the middle layer, while that of snow is mainly near 1000 hPa. Discriminative accuracies of different decision tree discriminant models constructed from different factors vary differently. T2 m is especially important, and wet bulb temperature can improve the discriminative accuracy in the process of discrimination. The accuracy of the decision tree discriminant model constructed from cloud top temperature, middle level mel ting parameter and low level wet bulb temperature is the highest among all models, with a discriminant accuracy rate of 91.86%.