Through the Two hailstorms processes on May 17, 2018 and April 27, 2019, combined with the radar echo evolution characteristics of the individual hail_cloud cells, the characteristics of the polarization parameters such as Zh,Zdr,φdp, and ρhv of the individual hail_cloud cells were analyzed, and The corresponding relationship between the distribution of precipitation particles in different phases and temperature forms environmental conditions.The analysis results show that: (1) If the 45 dBz echo height is above the 0℃ layer height, the echo intensity Z≥55 dBz, and VIL≥20g.km-2 are used as hail identification indicators, the body scanning observation time will be 1-3 times earlier than the hailfall time. , the hail warning time advance is 6-18 minutes.(2) The Zdr of the non-spherical large particle area is between 2.8-5.4 dB, φdp is between 179-191°, and ρhvis between 0.85-0.99, which corresponds to the strong echo area and mainly contains large raindrop particles. Composed of groups, the ρhvof the approximately spherical large particle area is between -2.3-1.2dB, φdp is between 180-235°, and ρhv is between 0.54-0.90. The main distribution is spherical particles such as graupel or hail.(3) When approximately spherical particles (hail) are located on the front side of non-spherical particles (raindrops) and there are positive and negative central areas of Zdr , this is due to the drag effect of wind direction and speed, which makes the large raindrop cluster move smaller. The result of graupel or small hail moving faster in the direction of the wind, placing it at the front of the strong echo zone.(4)Since the particle identification and classification algorithm in this study did not consider the temperature and environmental conditions in which hydrometeor particles exist, solid precipitation particles such as wet snow (WS) and ice crystals (CR) exist in warm clouds, and there is a lack of graupel particle classification and identification products. Therefore, The hydrometeor particle classification and identification algorithm needs to be improved.