A torrential rainfall event occurred in Guangzhou on 7 May 2017. In order to study the impact of different background field data on the simulation of rainstorm process, the NCEP Global Forecast System (GFS) data and CMA T639 data were used separately to develop lateral boundary and initial conditions for the numerical simulation and impact analysis of the rainstorm with the GRAPES_Meso model. The results indicated that background fields have significant impact on the simulation of rainstorm process. The precipitation forecasts using T639 data (T639_run) as initial and boundary conditions seem to be roughly simulated, while the location of the rainfall simulated by GFS data (GFS_run) is obviously northerly. The reason is that there is a deep water vapor transport and strong ascending motion in the rainstorm area of South China using T639 data as the background field, which can produce extremely heavy rainfall. The updraft and water vapor transport are weak in the actual storm area simulated by GFS data, making the heavy precipitation area northerly. Overall, the height of cloud top is higher and the temperature of cloud top is lower simulated by GRAPES_Meso model. Relatively speaking, the simulation result using T639_run is better than using GFS_run. This research result could provide a reference for the improvement and optimization of the calculation about the water substances and cloud fraction in cloud schemes.