Based on the wind field and vapor field at 700 hPa of NCEP/NCAR daily reanalysis dataset, and the precipitation data of 11 stations in Shanghai, this paper applies the method of the extended period rainstorm forecast in Shanghai. Based on the data of 112 rainstorm days in the past 10 years, the static prediction model of low frequency system is established. One is the type of northern lowfrequency anticyclone and the south lowfrequency cyclone, while the other is the type of northern lowfrequency cyclone and the south lowfrequency anticyclone. Further research shows that rainstorms with a daily rainfall of 100 mm or more are mainly lowfrequency cyclones, accounting for 60% of the total number of rainstorms. Lowfrequency cyclonetype mobile characteristics are very consistent in the 30 days before the occurrence of rainstorms. This consistency can be used to establish the dynamic prediction model of heavy rain. Using this method, the heaviest rainfalls in the flood season in the last 5 years are forecasted accurately, and the lead time is more than 12 days.