A Models-3 Based Self-correcting Air Quality Forecast System and the Estimation

A Models-3 based self-correcting air quality forecast system was discussed in this paper, the automated system was developed with Gambas, Yabasic and R language, which integrates 3 models including WRF, SMOKE and CMAQ. The forecast system captures monitor data from network, corrects the concentrations of different pollutants, and then public the results via web server. The hardware requirements of forecast system is relatively low and it was deployed on a DELL Optiplex 9010 workstation with a 6km-2km nested domain, giving operational air quality forecast for Chengdu. A estimation of the system was performed with 2014 forecasted concentrations and AQI, the results showed that the system well reflected the air quality variations in 2014, the hit rate of 24h direct forecast on air quality grads was 58.27% with a correlation coefficient of 0.71, and the corrected 24h forecast had a hit rate of 64.9% with a correlation coefficient of 0.89, the selfcorrecting method can improve the 24h forecast of Chengdu.