摘要
为探讨安徽省PM_(2.5)时空分布特征,文章基于2015-2020年地基观测PM_(2.5)、AOD、植被覆盖产品以及气象要素数据等,对比了多尺度地理加权回归、随机森林、全连接神经网络3种模型的精确度,并采用全连接神经网络模型反演了PM_(2.5)浓度,分析了PM_(2.5)浓度的时空变化特征,以及各因子对PM_(2.5)浓度的影响力。结果表明:3种模型中,全连接神经网络模型的精确度最高;2015-2020年PM_(2.5)浓度从平均51.29μg/m^(3)递减至36.71μg/m^(3),季节尺度上,PM_(2.5)浓度冬季>春、秋季>夏季,受政策及疫情影响,2018年的秋冬季、2020年的春夏季PM_(2.5)浓度下降同比最快;空间上表现为皖北>皖中>皖南。10个影响因子两两交互,均比单一因子对PM_(2.5)浓度的影响大,其中AOD对PM_(2.5)的影响力最大,各因子在不同的季节会对PM_(2.5)产生不同程度的影响力。
In order to explore the spatial and temporal distribution characteristics of PM_(2.5) in Anhui Province,based on ground-based PM_(2.5),AOD,vegetation cover products and meteorological elements data from 2015 to 2020,the accuracy of three models,namely,multi-scale geographically weighted regression model(MGWR),random forest and fully connected neural network,was compared.The full-link neural network model was used to invert PM_(2.5) concentration,analyze the spatiotemporal variation characteristics of PM_(2.5) concentration,and the influence of each factor on PM_(2.5) concentration.The results show that the fully connected neural network model has the highest accuracy among the three models.From 2015 to 2020,PM_(2.5) concentration decreased from an average of 51.29μg/m^(3) to 36.71μg/m^(3).On a seasonal scale,PM_(2.5) concentration was higher in winter than in spring and autumn than in summer.The spatial pattern was northern Anhui>central Anhui>southern Anhui.Any interaction of two of the 10 influencing factors has a greater impact on PM_(2.5) concentration than a single factor.AOD has the greatest impact on PM_(2.5),and each factor has different degrees of influence on PM_(2.5) in different seasons.
作者
赵月娇
赵萍
徐凯健
周鹏
申鹏举
于婉婉
陈国旭
ZHAO Yuejiao;ZHAO Ping;XU Kaijian;ZHOU Peng;SHEN Pengju;YU Wanwan;CHEN Guoxu(School of Resources and Environmental Engineering,Hefei University of Technology,Hefei 230009,China;Nanjing Xinda Institute of Meteorological Science and Technology Co.Ltd.,Nanjing 210044,China)
出处
《环境科学与技术》
CAS
CSCD
北大核心
2022年第6期171-178,共8页
Environmental Science & Technology
基金
国家自然科学基金面上项目(41972304)
安徽省自然科学基金项目(2008085QD193)
平安煤炭开采工程技术研究院有限公司委托项目(W2018JSFW0551)。
