Abstract: Numerical simulations with application of the nudging technique in the Weather Research and Forecasting (WRF) model to assimilate Final Operational Global Analysis data (FNL) were implemented to study three convective rainfall processes in Northwest China. Impacts of spectral nudging (SN) and grid nudging (GN) methods on precipitation simulation in this area were also examined. The result shows that the simulations of SN and GN experiments agree better with station-observed data compared with that of the control experiment. It is obvious that the simulation of distribution and magnitude of precipitation have been improved significantly in the two experiments with nudging, while the SN experiment results are better than that of the GN experiment. Analysis of differences in the basic meteorological elements (humidity, temperature, and wind speed) during the precipitation period simulated in the two assimilation experiments, it was found that the surface wind speed and temperature simulated by the GN experiment are closer to observations than that simulated by the SN experiment. However, the simulation of wind speed and relative humidity at 700 hPa and precipitation in the SN experiment are better than that in the GN experiment. Finally, from the perspective of the vertical variation of the diagnostic physical variable, i.e. water vapor flux divergence, the positive and negative distributions of water vapor flux divergence at 700 hPa and 600 hPa in the SN experiment effectively modulate the spatial distribution of precipitation and the simulation agrees well with observation. The increase of water vapor flux at 700 hPa is one of the major reasons for the improvement of precipitation simulation.