In order to improve the monitoring and early warning technology of shorttime strong rainfall and its application to mountain areas, selecting the relevant radiosonde data, identifying the type of precipitation, and combined with the automatic correction technology, the classification optimization technique of 〖WTBX〗ZI〖WTBZ〗 relationship is used to estimating rainfall. Besides, the radar reflectivity factor is extrapolated by using the crosscorrelation extrapolation method, and the cumulative precipitation in 30 minutes is predicted. Based on the observation data of nearly three years, the grades of precipitation corresponding to the radar reflection and radar rainfall estimate at Yueyang are determined. Compared with the data of automatic stations, the relative error of the precipitation estimate is smaller; compared with the empirical formula of 〖WTBX〗ZI〖WTBZ〗 relationship, the accuracy of the rainfall estimate by the classification optimization technology of 〖WTBX〗ZI〖WTBZ〗 relationship shows improvement. After a year of application in Yueyang Meteorological Service, the early warning technology has a good result in the precipitation estimation and shorttime strong rainfall warning.