The purpose of this project is to maintain the performance of forecasting systems and to facilitate the daily forecast of air quality over Taiwan region. The main work includes the maintenance and operation of the mesoscale numerical forecasting model, the Mainland China dust dynamic numerical model (TAQM/KOSA), the local riverside mineral dust dynamic model (TAQM/TWKOSA), and the Community Multi-scale Air Quality Model (CMAQ), and provides the forecast product query webpage. The characteristics of atmospheric variables and the concentrations of ozone, PM10 and PM2.5 from the model’s outputs and from the observation have been analyzed. To improve the air quality forecasting capability, we have also tried to identify the potential issues leading by the dynamics. In the future, these results can be further used to improve the performance of the dynamic numerical model. Current sensitivity analysis includes comparing the concentrations of O3, PM10 and PM2.5 between the data from EPA monitoring stations and from model simulations. For general pollution analysis, the simulated concentrations, diurnal variations and the seasonal variation of the pollutants are compared with the observation of EPA monitoring stations and seven air quality districts respectively. The poor quality event days in different weather patterns in this past year, and a serious air pollution event due to poor local dispersion, are analyzed. And the adjustment results of the model parameters are also analyzed, including the minimum boundary layer height limit test and the adjustment test of the dust emission distribution in Taiwan (WRFDUST). To help the daily forecasting in EPA, special-duty manpower is assigned to work for organizing and integrating all of the information to support the pre-warning decision. This project also provides the expert opinion from project director during the Mainland China severe dust events forecasting meeting.