In this study,based on the quality-controlled S-band dual-polarization radar gridded observation data,a hydrometeors classification recognition algorithm is established,so as to analyze the spatial distribution and evolution of hydrometeors in the precipitation process by using the fuzzy logic algorithm,as well as the characteristics of hydrometeors scattering and spatial orientation.This algorithm can classify hydrometeors into different types such as liquid,ice,and mixed states,which is helpful in finding the key structures of cloud microphysics which affect the precipitation.First,the membership function is obtained according to the radar echo characteristics of hydrometeors.Second,according to the different contributions of radar observation variables in identifying hydrometeor types,the weight of the membership function value corresponding to each observation value is determined,and,after the weighted average of each function value is obtained,then the logical value corresponding to hydrometeors types is obtained as well.Finally,the integration and defuzzification processing is performed,and the maximum value of the logical value in each grid point is selected,after which the hydrometeor type represented by the value is considered to be the particle type represented by the grid point.When determining the weight of the membership function corresponding to the observed value,the horizontal reflectance factor and ambient temperature are taken as the direct influence factors for calculating the hydrometeor types,and,instead of the weighted average calculation,an algorithm for hydrometeor types recognition based on the parameters of S-band dual polarization radar and the ambient temperature is proposed.Next,the rationality of the algorithm is verified by means of a precipitation process in the pre-flood period of South China using radar and precipitation data.The study results show that the distribution characteristics of the rain-type obtained by the inversion are basically consistent with the distribution characteristics of the actual observed precipitation,which proves that the algorithm is able to reflect the hydrometeor types in the precipitation area,and that the recognition results are basically reasonable.Further research shows that there are a large number of drizzle-type particles present in the air before the occurrence of precipitation,and that there is a negative correlation between the changes of drizzle and rain particles during the process of precipitation,thereby indicating that this precipitation is mainly caused by the collision of drizzle and rain particles.