For solving inconformity with forecast errors, the Ensemble Transform Kalman Filter (ETKF) method is used to update the perturbations generated by dynamic downscaling in Global/Regional Assimilation PrEdiction System-Regional Ensemble Prediction System (GRAPES-REPS) and capture more mesoscale perturbations in this paper. On this basis, a brand new method named Cosine Analysis Constraint Scheme is designed to optimize the updated initial perturbations which makes the up-to-date perturbations more consistent with the forecast errors. The results show that the perturbations adjusted by Cosine Analysis Constraint Scheme may present the most abundant information for mesoscale compared to other experiments in the area of precipitation. After constraining, the structure and evolution of the perturbations become more reasonable and match better with the development of weather systems at different scales. The features of spread and perturbation energy are also improved effectively, especially in the early period.