In order to explore the performance of WRF (weather research and forecasting) model on the simulation of maximum hourly rainfall process at the rainstorm center of Zhengzhou City, we selected five common microphysical process schemes to simulate the 14 short term rainstorms that occurred in Zhengzhou City from 2011 to 2018, and analyzed the rainfall characteristics in combination with the temporal and spatial distribution, rain pattern, rainfall peak value and rainfall concentration. The results show that the WRF model can simulate the rainfall of the corresponding event, but with a deviation at the position of the rain area, and the time deviations of the rainfall process for more than 60% of the events are less than 180 min. The model performs well at the simulation of the accumulated rainfall, with a high correlation to the rainfall process, but lacking the ability to simulate the process rainfall amount. According to rain pattern analysis produced by the fuzzy recognition method, the WRF model performs well at the simulation of single peak rainstorm pattern in Zhengzhou City, but poorly at the uniform rainfall pattern. When the WRF model is used to simulate the single peak rainfall pattern, the rain peak deviation can be controlled within ±30%. The simulated rainfall concentration of 13 rainfall events are consistent with the actual measurements. Research shows that the overall simulation results of the WRF model are ideal, which can provide numerical method support for the study of short term rainstorms in Zhengzhou City.