Abstract: The WRF model is used to conduct the high-resolution (5 km×5 km) dynamical downscaling prediction experiments for the climate in Chongqing in the summer (June-August) of 2021 based on the forecast of Global Climate Prediction System 1.0 (NUIST-CFS1.0) initialized from March 1, 2021. The effects of two land surface parameterization schemes (Noah and RUC) on the downscaling seasonal prediction results are compared and analyzed. The results show that WRF dynamical downscaling can improve the predictions of 2-meter air temperature and precipitation in Chongqing during summer. With the increased horizontal resolution, WRF dynamic downscaling can provide more detailed spatial distribution of 2-meter air temperature, but the improvement in precipitation prediction remains limited. From the impact of the land surface scheme, the RUC scheme demonstrates better agreement with observations in precipitation prediction compared to the Noah scheme. However, the Noah scheme predicts a relatively small latent heat flux and a relatively large sensible heat flux, which contributes to alleviating the underestimation bias of 2-meter air temperature in the original NUIST-CFS1.0 forecast. From the prediction effect of precipitation, WRF dynamic downscaling is superior to NUIST-CFS1.0 in predicting the probability of precipitation of different magnitudes. It can not only capture more heavy rain events (>25 mmday⁻¹) in Chongqing in summer, thus reducing the negative deviation of precipitation in NUIST-CFS1.0, but also capture the characteristics of night rain in Chongqing more reasonably. Further analysis demonstrates that WRF dynamical downscaling exhibits superior performance in capturing the diurnal cycle characteristics of low-level circulation and water vapor transport over Chongqing compared with NUIST-CFS1.0, which corresponds to the prediction effect of WRF on precipitation.