Abstract: Geographically weighted regression kriging (GWRK) is one type of precipitation data processing method which is developed from the geographically weighted regression (GWR). GWRK could consider both variables’ spatial autocorrelation and regression’s spatial non-stationary. This study is a case research that compares how the efficiencies of GWRK and GWR work on the Ganjiang River basin precipitation dataset, and the comparison is carried out via three below steps. Firstly, the basin Tropical Rainfall Measuring Mission (TRMM) precipitation data was validated. Secondly, the GWRK and GWR methods were applied to create the “station-TRMM” merging precipitation model separately. At last, the merging precipitation data was used to drive the GR4J hydrological model to perform hydrological projection. The comparison showed that GWRK method could obviously improve the precipitation merging data accuracy at station scale. The improvement of hydrological projection’s accuracy, however, was not as high as that of station precipitation data, that is mostly due to the hydrological model inputting data are homogenized large-scale precipitation data, which are quite different to “point-to-point” precipitation dataset.