Abstract: Based on the visibility and air quality results from January to March in 2015 forecasted by four environmental meteorology operational numerical models from the North China,East China and South China Regional Meteorological Centers and the National Meteorological Center,respectively,their forecasting quality was verified.The results show that with the decrease of visibility,these models' forecasting quality gradually reduces.When the visibility is less than 1 km,only models from the North China Regional Meteorological Center and the national fog and haze numerical forecasting operational system-CUACE (CMA Unified Atmospheric Chemistry Environment) exhibit a certain forecasting skill;the TS to 24 h visibility prediction in Beijing are 0.20 and 0.10,respectively.Generally speaking,the visibility prediction error of the CUACE model is larger than those of regional meteorological centers' models,and the correlation coefficient of observations with the predictions is less than 0.6.Similarly,with the decrease of air quality,these models' prediction quality to AQI (Air Quality Index) gradually reduces.When the AQI is at the excellent rank,the TS are higher;and due to the good air condition in South China,its TS score for AQI prediction is the highest.In general,for the 24 h AQI prediction,the forecasting qualities of the regional centers' models are superior to that of the CUACE model,and for the 48 h and 72 h ones,the situation is opposite.The values of PM_2.5 concentration forecasted by these models are normally lower than those of observations.The error of 24 h PM_2.5 concentration forecasted by the South China Regional Meteorological Center's model is relatively small,and those forecasted by the North China and East China Regional Meteorological Center's models are comparatively larger.For the prediction of PM_2.5 concentration,the performance of the CUACE model is the worst among these models,with a lower correlation coefficient between the predictions and observations.