Abstract: In this paper, we predict the air temperature and precipitation of the Yangtze River basin in five years with abnormal precipitation by the Regional Climate Model (RegCM4.5). We compare the prediction results of air temperature and precipitation with three land surface parameterization schemes and two spatial resolutions. The results show that high spatial resolution can improve the accuracy of air temperature and precipitation prediction; Different land surface schemes can cause different temporal-spatial distribution of surface net flux and evapotranspiration, leading to uncertainty prediction performance of air temperature and precipitation in the Yangtze River basin. Through the comparative analysis, we found that CLM4.5 with 30 km resolution is the best combination of predicting light rain, and BATS with 30 km resolution is the optimal choice of predicting heavy rain. CLM3.5 with 30 km resolution gives the best performance for predicting air temperature in the Yangtze River basin.