Using the WRF (Weather Research and Forecasting) model and its 3DVAR (3-Dimentional Variational) data assimilation system, data assimilation sensitivity experiments on a local torrential rain event in middle Shandong on 8 May 2011 are carried out adopting data assimilation and rapid update cycle of 36 km, 12 km and 4 km nested grid at three-hour intervals. Results show that both surface data assimilation and rapid update cycle are playing key roles in the precipitation forecast. During rapid update cycle, the model cannot forecast the rainfall in Shandong if assimilating no surface observational data or assimilating all surface observational data with cold start. The rainfall area forecast is significantly improved due to surface data assimilation, which can influence the atmospheric elements such as temperature, pressure, humidity, and wind over 700 hPa. Therefore, the temperature and humidity structure of initial atmospheric conditions is altered, the upper humidity and thermal instability are enhanced at around 700 hPa, the lower wind field becomes stronger under 700 hPa, the wind speed is 2 to 4 m·s-1 stronger than that without data assimilation at 850 hPa to the south of middle Shandong, and the upper unstable atmosphere is triggered by the dynamic effect of the lower wind field. As a result, the rainfall appears in Shandong.