Based on the daily ground snow observation data from Weining Station and the second sounding data of L-band radar at 08∶00 and 20∶00 located in the same place, the snowfall prediction indicators of the distribution characteristics of temperature, humidity, wind direction and wind speed of sounding standard isobaric surface from 2008 to 2019 are analyzed by using mathematical statistics method. Then we build a snow forecast model. The results show that: ①The snowy weather in Weining is mainly related to the humidity and wind direction and speed of the middle and low layers, and the temperature near the ground~500 hPa; ②The snowfall forecast model mainly includes the humidity index (H_earth≥80%,H₇₀₀≥80%、H₆₀₀≥60%)、the temperature index（T_earth≤0.5 ℃,T₇₀₀≤ -1.2 ℃,T₆₀₀≤ -4.8 ℃,T₅₀₀≤-8.2 ℃)、the wind direction and wind speed index （V_earth≤5 m·s^-1, V₆₀₀≥17 m·s^-1）, as the altitude increases, the wind speed increases, and the north-south wind shear becomes obvious; When the humidity and temperature indicators are met at the same time, it is directly judged that there will be snowy weather, and the wind direction and wind speed indicators are no longer judged.③The forecast model is used to test the snowfall in winter (December, January and February) of 2020. The prediction accuracy is 70%, and the prediction reliability is high. It has strong guidance for snowfall prediction in Weining.