In order to reveal the impact of future water resources changes on the Luan-Tianjin Water Diversion Project, we constructed a SWAT model for the Panjiakou Reservoir-controlled watershed in the Luanhe River Basin based on geospatial and hydrometeorological data. Four single indicators and one comprehensive indicator were selected to evaluate the applicability of the selected eight global climate models (GCMs) on a station-by-station basis, and a weighted multi-mode data set was assembled based on the evaluation results. Subsequently, the improved Delta-DCSI was used to downscale and correct the meteorological data, while retaining its trend signal. The corrected meteorological data was then used to drive the SWAT model, so as to predict the water resources changes in the future scenario. The results show that the proposed SWAT model has good applicability, and the accuracy of the multi-mode ensemble data after model optimization can meet the requirements of the SWAT model for future prediction. In the future, the interannual variation of runoff in the three climate scenarios will be dramatic with great fluctuations, showing an overall increasing trend. Compared to 1997-2022, the inflow runoff of Panjiakou Reservoir will increase significantly in the future, providing abundant water for the basin. As a result, the water demand of Tianjin from Luanhe River water diversion will decrease.