摘要
2012年环境保护部发布的《环境空气质量标准》实施后,贵阳市空气质量状况发生了变化。利用贵阳市空气质量指数和常规气象要素等资料,分析空气质量特征及其与气象要素的关系,通过多元线性逐步回归和BP神经网络方法,分季节建立空气质量指数预报模型,并同CUACE模式进行对比检验。结果表明:近3年贵阳市空气质量状况良好,优良天数增多,污染天数减少且污染天气多出现在冬季,首要污染物为PM_(2.5)、PM_(10)和O_3;各季相关因子不同,但主要与相对湿度和风速有关;两种模型预报效果均表现为夏季评价最高,等级TS评分超过85%,指数准确率近99%,冬季预报效果相对最差,TS评分接近或达到70%,指数准确率超过或接近80%,而春、秋季效果指标差距不大;对2015—2016年AQI的预报效果回归模型的优于CUACE模式的,TS评分和预报准确率分别相差16.2%和20.0%。
The air quality in Guiyang has changed since the implementation of the new ambient air quality standards Ministry of Environmental Protection in 2012.Utilizing the air quality index(AQI)and conventional meteorological elements in Guiyang,the air quality characteristics and its relationship with meteorological elements are analyzed.AQI forecast models are developed according to different seasons,using the stepwise multiple linear regression and the BP neural network method,and compared with CUACE model.The results show as follows.The air quality in Guiyang city in recent 3 years has improved,with light pollution days increased and heavy pollution days decreased.The air pollution is mostly in winter,while the primary pollutants are PM2.5,PM10 and O3.The correlation between AQI and meteorological factors is different in different seasons,but it's mainly related to the relative humidity and wind speed.The two AQI forecast models are season-dependent,with the highest(lowest)forecast skill in summer(winter).Level TS scores are over 85%and the accuracy of index are nearly 99%in summer,while level TS scores close to or reach 70%and the accuracy of index are over or near 80%in winter.The model effect indexes are middle level and close in spring and autumn.Stepwise regression model shows a better comprehensive skill score than the CUACE in 2015 2016,and the TS score and the prediction accuracy differ by 16.2%and 20.0%.
作者
宋丹
夏晓玲
何玉龙
张蕾
杜正静
Song Dan;Xia Xiaoling;He Yulong;Zhang Lei;Du Zhengjing(Meteorological Service Center of Guizhou Province,Guiyang 550002,China;Rural Comprehensive Economic Information Center of Guizhou Province,Guiyang 550081,China)
出处
《气象与环境科学》
2019年第1期93-100,共8页
Meteorological and Environmental Sciences
基金
贵州省科技支撑计划项目(黔科合支撑[2018]2779)
贵州省气象局气象科技开放研究基金项目(黔气科合KF[2015]04号)
中国气象局预报员专项(CMAYBY2016-064)
关键词
空气质量
预报
多元线性逐步回归
BP神经网络
贵阳市
air quality
forecast
stepwise multiple linear regression
BP neural network
Guiyang city
