Abstract: In order to improve the assimilation of wind profile radar (WPRD) observations, research is carried out to develop the quality control (QC) and assimilation application of WPRD. For the 13 WPRD observations in Guangdong during May 2013, the method based on empirical orthogonal function (EOF) analysis is used to do the QC of WPRD. Compared with the original observations, the new datasets of observations after QC increase (decrease) the contribution of large (small) scales in both space and time, filter the small-scale and high-frequency noises very well, retain the combined effects of large-scale mean states and small-scale local systems successfully, as well as are closer to the ECMWF reanalysis data. After QC, the datasets of observations are thinned in the vertical. The differences between the NCEP 6 h forecasts and the observations datasets before and after QC are calculated, respectively. These differences show that, the distributions of the innovations corresponding to the observations after QC are more Gaussian-like and unbiased. Based on the GRAPES 3D-Var data assimilation system, the WPRD observations after QC are used in a real-case study characterized by severe rainfall, to illustrate the influence of WPRD observations on model analysis and forecasting. Assimilating WPRD observations in the update cycle is showed to improve the descriptions of the wind fields at the lower levels in the initial conditions, and the corresponding forecast performance of precipitation in terms of both location and intensity is improved as a result. A series of the data assimilation experiments for May 2013 shows that, the assimilation with WPRD observations significantly outperforms the one without WPRD observations with regard to short-range precipitation prediction.