基于特征指标的气象因子对PM_(2.5)浓度的影响分析  

The Analysis of the Influence of Meteorological Factors on PM_(2.5) Concentration Based on Characteristic Index

作　　者：卫星君[1] 赵晓萌[2] 王琦[2] 肖敏敏[1] WEI Xingjun;ZHAO Xiaomeng;WANG Qi;XIAO Minmin(Shaanxi Energy Institute,Xianyang 712000,China;Shaanxi Provincial Climate Center,Xi’an 710014,China)

机构地区：[1]陕西能源职业技术学院,陕西咸阳712000 [2]陕西省气候中心,陕西西安710014

出　　处：《中国环境监测》2022年第6期90-100,共11页Environmental Monitoring in China

基　　金：陕西省自然科学基础研究计划项目(2021SF-493);陕西省气象局秦岭和黄土高原生态环境气象重点实验室开放研究基金课题(2019Y-7);陕西能源职业技术学院重点科研项目(19KYZ02)。

摘　　要：以西安为研究区域,为探究气象因子对PM_(2.5)浓度的影响,采集2017—2019年空气质量与气象因子数据,改进k-Means聚类算法,形成严重污染、重度污染、中度污染、轻度污染共4个PM_(2.5)浓度与气象因子样本簇集。分析簇集数据分布,选择Spearman相关性分析方法,确定影响PM_(2.5)浓度的气象因子;定义PM_(2.5)凸显性条件,给出幅度特征FOA、浮动特征FOF和凸显特征FOH,构建三维空间,确定气象因子对PM_(2.5)影响的大小,进而建立气象因子对PM_(2.5)浓度的影响分析方法。比较多元线性回归和随机森林回归方法,结果表明:该方法提高了分析效率,且无需考虑因子选取和表达,能有效确定影响PM_(2.5)浓度的气象因子种类及影响程度。在低温、高湿、高压和相对静风的气象条件下,空气中颗粒物难以扩散和输送,使西安市PM_(2.5)浓度升高。严重污染、重度污染和中度污染中,PM_(2.5)浓度与相对湿度呈显著正相关,与风速、气温呈显著负相关,且影响大小依次为相对湿度>风速>气温;轻度污染中,PM_(2.5)浓度与相对湿度、气压呈显著正相关,与风速、气温呈显著负相关,且影响大小依次为气温>相对湿度>气压>风速。In order to explore the impact of meteorological factors on PM_(2.5)concentration,Xi’an is selected as the research area.Based on the data of air quality and meteorological factors in Xi an from 2017 to 2019,the K-Means clustering algorithm is improved to form four PM_(2.5)concentration and meteorological factor sample clusters of severe pollution,heavy pollution,moderate pollution,light pollution.The distribution of cluster data is analyzed and spearman correlation analysis method is selected to determine the meteorological factors affecting PM_(2.5)concentration;the salience condition of PM_(2.5)is defined,the amplitude feature FOA(Feature of Amplitude),floating feature FOF(Feature of Float)and salient feature FOH(Feature of Highlight)are given,and the three-dimensional space is constructed to determine the impact of meteorological factors on PM_(2.5)concentration,and then the analysis method of the influence of meteorological factors on PM_(2.5)concentration is constructed.Compared with multiple linear regression model and random forest regression,the experimental results show that this method simplifies the analysis and calculation process,avoid factor selection and expression,and can effectively determine the influence of meteorological factors and their magnitudes on PM_(2.5)concentration.Through this method,under the meteorological conditions of low temperature,high humidity,high pressure and relatively quiet wind,the difficult diffusion and transportation of particulate matter in the air is obtained,so that the PM_(2.5)concentration is increased in Xi an.In severe pollution,heavy pollution and moderate pollution,PM_(2.5)concentration is significantly positively correlated with relative humidity,and significantly negatively correlated with wind speed and temperature,and the order of influence is relative humidity>wind speed>temperature.In light pollution,PM_(2.5)concentration is significantly positively correlated with relative humidity and air pressure,and significantly negatively correlated with wind s

关 键 词：PM_(2.5) 气象因子 特征指标 多元线性回归 随机森林回归 

分 类 号：X823[环境科学与工程—环境工程]