Abstract:A numerical chemical weather forecasting system was established and operationally implemented based on the WRF-Chem Model, an online coupled regional chemical transport model. Performance of the modeling system on daily maximum 1-hour and 8-hour ozone (1h and 8h O3) concentrations was evaluated between May 1st and September 30th, 2013. The results showed that the numerical forecasting has generally good performance. There is no substantial systematic bias in 1h and 8h O3 concentrations and correspondent IAQI in forecasts of 24h, 48h, and 72h. The correlation coefficients (R) are ~ 0.8, and the mean and median biases are around 1′10-9~2′10-9. The forecasted O3 attainment vs. pollution days as well as primary pollutants are also in good agreement with observations. The performance of 48h forecast is slightly better than that of 24h and 72h forecast, and these of the later two are generally close to each other. Meanwhile, further improvement is still needed. For example, model shows substantial biases in O3 concentrations or IAQI forecasts in some cases, and the accuracy of O3 IAQI level forecast is substantially lower than that of concentration and IAQI value forecast. In general, the numerical forecasting system shows relatively good performance in O3 forecasts during May to September, 2013, and it has the capability to support the air quality forecast over Shanghai.