The phase state of precipitation in winter is relatively complex, and the consistency of transformation forecast between liquid and solid state is a difficulty in the forecast. In order to explore the application of muti-source data in rain-snow phase prediction, the precipitation phase characteristics of a rain-snow weather in Zhangjiakou area from 17 to 19 November 2020 were analyzed based on ERA5 reanalysis data, as well as cloud radar, microwave radiometer and SA dual polarization Doppler radar observation data. The results show that the rain-snow weather occurred under the coordination of upper trough, lower vortex and surface trough, the cold air from northwest behind the upper trough moved southward, which led to a rapid drop of air temperature and the change of precipitation phase. In the early stage of the process, the whole atmosphere was strong warm advection, and the ground temperature was high and the precipitation phase was rain. In the evening of November 18, the cold advection developed strongly, the temperature of each layer decreased rapidly, and the whole layer turned into cold layer, which led to the conversion of precipitation phase to snow. Through the dynamic diagnosis of divergence and vertical velocity, it was shown that the dynamic forcing during the rainfall period was mainly located in the upper layer, while during the snowfall period it was mainly located in the lower layer. The high-resolution data of cloud radar could reflect the change of 0 ℃ layer, and the change of mass center greater than 10 dBZ could indicate the change of precipitation intensity. The maximum basic velocity could reach 6-8 m·s-1 during rainfall stage, while it was less than 2 m·s-1 at snow stage. The high-resolution data of microwave radiometer could accurately judge the time of rain-snow conversion, and it was found that the integral water vapor appeared jump and peak value three to five hours before the precipitation. The combination of dual polarization radar and microwave radiometer could accurately judge the phase of precipitation particles, which could be used in prediction of precipitation phase.