Abstract: Aiming at a regional rainstorm weather process in Sichuan Basin in 2021, this paper uses SAL method to test and compare the high resolution forecast products of CMA_GFS, EC, SWC_WARM, SWC_WARR and EC ensemble forecast. The results shows EC ensemble forecast is the best in the structure prediction of rainfall area, and CMA_GFS is better in precipitation intensity prediction. The location prediction errors of all models are all small, and among them the error of SWC_WARM is minimal. There is a certain relationship between TS score and verified results by SAL, and the unilateral prediction advantage of SAL cannot correspond to the high TS score, while the TS score is higher when the three aspects of SAL are balanced. EC and CMA_GFS show better stability in the aspects of structure, intensity and location, and the overall prediction is relatively stable, but the error from 08: 00 is smaller. The prediction error of SWC_WARR is getting smaller and smaller over time, which can provide some reference for the prediction.