地理探测联合地理加权岭回归的PM_(2.5)驱动因素分析  

Identification of Driving Factors of PM_(2.5) Based on Geographic Detector Combined with Geographically Weighted Ridge Regression

作　　者：赵锐 詹梨苹 周亮 张军科 ZHAO Rui;ZHAN Liping;ZHOU Liang;ZHANG Junke(Faculty of Geosciences and Environmental Engineering,Southwest Jiaotong University,Chengdu 611756,P.R.China;School of Environment and Municipal Engineering,Lanzhou Jiaotong University,Lanzhou 730070,P.R.China)

机构地区：[1]西南交通大学地球科学与环境工程学院,四川成都611756 [2]兰州交通大学环境与市政工程学院,甘肃兰州730070

出　　处：《生态环境学报》2022年第2期307-317,共11页Ecology and Environmental Sciences

基　　金：国家自然科学基金项目(41571520);四川循环经济研究中心课题资助(XHJJ-2002,XHJJ-2005);成都市软科学研究项目(2020-RK00-00240-ZF,2020-RK00-00246-ZF);中央高校基本科研业务费专项资金(2682021ZTPY088)。

摘　　要：开展PM_(2.5)的驱动成因分析,对大气污染防治具有重要意义。利用2015—2018年PM_(2.5)地面监测数据,结合地理探测器和地理加权岭回归方法,探测了全国282个城市PM_(2.5)空间分异的关键驱动因素,分析了各关键驱动因素对PM_(2.5)影响的时空异质性。结果表明,气象参数和社会经济活动可更好地解释PM_(2.5)呈现的空间分异性;在2015—2018年间,所建地理加权岭回归模型的R^(2)分别为0.698、0.724、0.656和0.712,AICc分别为1317.533、1234.400、1256.107和1110.740,2种指标均优于全局回归模型和地理加权回归模型,说明地理加权岭回归模型可更好地解释PM_(2.5)产生空间分异的关键影响机制;模型拟合结果进一步显示,气温、比湿度、地区生产总值、年平均人口和工业企业数是引起PM_(2.5)空间分异的关键驱动因素,各因素的影响既存在正向效应也存在负向效应,其对应的回归系数具有明显的时空异质性。The analysis of the driving causes of PM_(2.5) is of great importance to the prevention and control of air pollution.Using the PM_(2.5) ground monitoring data from 2015 to 2018,combined with geographic detectors and geographically weighted ridge regression methods,the key driving factors of PM_(2.5) and their spatiotemporal heterogeneity in 282 cities in China were investigated.The results indicate that meteorological parameters and socioeconomic activities can better explain the spatiotemporal heterogeneity of PM_(2.5) distribution.The R^(2) of the proposed model for each year(2015‒2018)are 0.698,0.724,0.656,and 0.712 and AICc of the model for the four years’data are 1317.533,1234.400,1256.107,1110.740,respectively.Based on these indicators,the proposed model has better fitting results than the global regression model and the geographically weighted regression model.In addition,the model-fitting results show that temperature,specific humidity,GDP,annual average population,and the number of industrial enterprises are the main driving factors of PM_(2.5).These factors have either a positive or a negative effect on PM_(2.5) and their regression coefficients have obvious spatiotemporal heterogeneity.

关 键 词：PM_(2.5) 影响因素 地理探测器 地理加权岭回归 时空异质性 

分 类 号：X16[环境科学与工程—环境科学]