Abstract: To evaluate the dynamical downscaling prediction skill on the averaged precipitation in flood season over Chongqing, using WRF model based on Beijing Climate Center (BCC) second-generation seasonal prediction model system (BCCv2) product, we have conducted a dynamical downscaling prediction test based on the hindcast data over Chongqing during the flood season of 2016 from BCCv2, and compared the effect of different PBL parameterization schemes on the precipitation prediction before and after the dynamical downscaling. The results show that the dynamical downscaling can improve the original prediction in which BCCv2 predicts less precipitation than observation, and the spatial distribution of BCCv2 hindcast data has also been improved by the downscaling. Comparison of precipitation prediction results after using the four PBL parameterization schemes, i.e. MYJ, MYNN2, YSU and ACM2, indicates the YSU has a relative less prediction bias against observations compared to other schemes, although the difference among the four schemes is not very significant. Analysis on the bias of atmosphere circulation between prediction and observation and the nudging test results show that dynamical downscaling can improve the prediction of geopotential height over southern Tibet Plateau and southern China and then weaker moisture transfer by westward wind over southern Tibet Plateau, although the northward moisture transfer over Guangdong and Guangxi is still weak. The bias between boundary field and observation may be connected with the less predicted precipitation than observation over Chongqing during the flood season of 2016 in the dynamical downscaling prediction.