基于MOS的杭州秋冬季空气污染预报和霾诊断

利用由数值预报模式WRF和辨识理论实时迭代统计方法RTIM组成的MOS方法对杭州市2013年2~3月和11~12月期间的空气污染物日平均浓度做预报,预报值与实测值之间相关系数都超过0.75 ,PM2.5、PM10、SO2、NO2、CO 24h平均浓度和O3 8h平均浓度分类预报临界成功指数(CSI)分别为89%、87%、100%、93%、100%和100%,命中率(POD)分别为93%、95%、100%、100%、100%和100%.分析表明,研究期间杭州地区气溶胶以细颗粒为主.根据PM2.5浓度、相对湿度及能见度预报值做霾日分类预报,临界成功指数为89%,命中率为93%.说明该MOS系统对污染物浓度及霾天气预报性能良好,可以为业务化预报提供参考.

Autumn and winter air pollution forecasting and haze events diagnosis in Hangzhou: a MOS-based approach

A MOS method, comprised of WRF and RTIM, is used to forecast daily average air pollutant concentration in Hangzhou from February to March and from November to December, 2013. The correlation coefficients between the predicted and observed values are all above 0.75. The resulting CSI for 24-hour average concentration of air pollutants is highly promising (89% for PM2.5, 87% for PM10, 100% for SO2, 93% for NO2, 100% for CO) and even reaches 100% for 8-hour average concentrations of O3, and the resulting POD is 93%, 95%, 100%, 100%, 100%, and 100% for each of the above pollution measures respectively. The analytic results indicate that PM2.5 is the major contributing pollutant in the aerosol in Hangzhou during the research period. The CSI and POD for haze events diagnosis, which was based on the classification of predicted PM2.5, relative humidity, and visibility, reach 89% and 93%, respectively. The high accuracy of air pollutant forecasting obtained in this study indicates that the MOS model performs well during the study period and has great potential to be applied to regional air pollution prediction in operation mode.