摘要
利用2017-2019年滁州市6种空气污染物质量浓度和AQI资料、气象要素及ECMWF数值预报作为输入参数,构建基于随机森林算法的污染物质量浓度和AQI预报模型,其中AQI预报按季节划分为4个模型.结果表明:6种污染物中O3的预报效果最好,相关系数为0.84,PM2 5、PM10和NO2相关系数分别为0.76、0.72和0.72,SO2和CO预报效果略差;基于分季节模型AQI的24 h预报结果TS评分为0.77,空报率(FAR)和漏报率(PO)分别为15％和12％,相较于非季节模型预报效果更好;春季和秋季的TS评分分别为0.88和0.86,相较于冬季和夏季预报效果更好.
Abstract
Using the mass concentration of six air pollutants,AQI data,meteorological elements and ECM-WF numerical forecasts of Chuzhou from 2017 to 2019 as input parameters,a prediction model of pollutant mass concentration and AQI was built based on random forest algorithm,in which the forecast of AQI was divided into four models according to seasons.The results showed that O3 had the best predictive effect among the six pollu-tants,with a correlation coefficient of 0.84.The correlation coefficients of PM2.5,PM10,and NO2 were 0.76,0.72,and 0.72,respectively.The predictive effect of SO2 and CO was slightly poor.The TS score of 24-hour AQI forecast results based on seasonal models was 0.77,the false alarm rate(FAR)and underreporting rate(PO)were 15％and 12％,respectively,which were better than non-seasonal models.The TS score in spring and autumn reached 0.88 and 0.86,which were better than those in winter and summer.
基金项目
安徽省气象局创新发展专项基金资助项目(CXM202206)
滁州市气象科研基金资助项目(CZQXKY201903)
维普
万方数据