Abstract: Herein, an assimilation interface for temperature, ozone, and water vapor data in the middle and upper atmosphere is developed, and an assimilation model with complete stratospheric processes, including assimilation, weather forecasting, and short-term climate prediction system, is built based on the latest version of the whole atmosphere community climate model (WACCM6) and the data assimilation research testbed. Using the assimilation analysis field as the initial value, the system performed assimilation simulations of the stratospheric atmosphere in March and April 2020 and provided 0–3 days, 4–15 days, and 16–30 days forecast, as well as 1–60 days short-term climate prediction for stratospheric atmosphere changes in May and June. The results show that the system can accurately reflect the time evolution of the very unusual ozone depletion event in the Arctic stratosphere in March and April 2020, which is very similar to the microwave limb sounder (MLS) satellite observations. While simulations without assimilation can simulate the Arctic ozone depletion event, the ozone depletion magnitude is less than that of the MLS satellite. The assimilation model system improves the simulation of not only the chemical composition of the Arctic stratosphere but also the Arctic stratospheric temperature and circulation changes. The simulated March and April and predicted May and June Arctic stratospheric temperature and zonal wind variability are in good agreement with the assimilation and reanalysis results using Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA2). The root mean square error (RMSE) between the predicted temperature and zonal wind and the reanalysis data is about 3 K and 4 m s⁻¹, respectively, for the top of the stratosphere. The RMSE between the temperature and zonal wind of the Arctic stratosphere simulated by the experiment without assimilation in March-April and forecasted in May-June and the MERRA2 reanalysis data reach about 6 K and 5 m s⁻¹ or more in most regions. On a global scale, this system has the most significant improvement in the simulation of the middle and lower stratosphere, with the RMSE in the prediction results reduced by more than 50% when compared to unassimilated simulation experiment prediction results.