Abstract: The fundamental issue of ensemble forecasting research is whether initial perturbation can accurately describe the uncertainty of atmospheric motion. A reasonable structure and amplitude of ensemble perturbation should reflect the characteristics of forecast error about the state of atmospheric motion. For solving the problems of insufficient spread, recognizability, and reliability of perturbation in GRAPES regional ensemble forecast, this paper designs and develops different schemes to constrain the initial ensemble perturbation with analysis increments extracted from data assimilation system using the relationship of structure and evolution characteristics between the ensemble perturbation and forecast errors of weather systems at different scales. Further, this study analyzes the spatial, physical structure and spatiotemporal evolution characteristics of the ensemble perturbation, spread, and perturbation energy to comprehensively assess the quality and performance of the analysis constraints schemes. Results show that the schemes can partly identify and adjust the false perturbation in the original forecast. After constraining, the structure and evolution of the perturbation are in better agreement with the development of weather systems at different scales, with higher accuracy to describe the forecasting uncertainty. The structures of spreads and perturbation energy are more reasonable,and their growth characteristics over time are also more significant, particularly in the early period.