主成分分析法在合肥市空气质量评估中的应用

Application of principal component analysis in air quality assessment of Hefei City

基于2019年1月1日至2020年12月31日地面国控站点的重要污染物日均浓度监测数据和气象数据,利用主成分分析法对合肥市城市空气质量进行了综合评估。其中污染物数据包括细颗粒物(PM_(2.5))、可吸入颗粒物(PM_(10))、二氧化氮(NO_(2))、二氧化硫(SO_(2))、一氧化碳(CO)和臭氧(O_(3))等重要大气污染物的日值数据;气象数据包括平均气温、平均风速、日照时数、20-20时累计降水量、平均本站气压和平均相对湿度。首先通过逐步回归法筛选出与空气质量指数有显著影响的指标,再通过主成分分析法对显著指标进行数据降维。根据主成分分析理论,从5个显著影响指标中提取出2个主成分,所提取的两个主成分累积方差贡献率达到82.9%。评估合肥市城区空气质量的第一主成分为PM_(2.5)、CO和PM_(10)组成的综合指标,表明合肥市空气质量受PM_(2.5)、CO和PM_(10)污染物的浓度变化影响最为显著;第二主成分为日照时数和O_(3)组成的综合指标,表明日照时数和O_(3)浓度的变化是影响空气质量的第二重要因素。主成分综合得分与空气质量指数具有较好的一致性(R^(2)=0.78),对比分析发现,主成分分析法对于优良空气质量及冬季的评估效果最好。研究结果表明,运用主成分分析法提取的主成分结合综合得分可对城区空气质量进行有效评估。

Based on the daily average concentration monitoring data of important pollutants and the meteorological data at ground national control stations from January 1,2019 to December 31,2020,a comprehensive assessment of urban air quality in Hefei City,China,was conducted using the principal component analysis method.The pollutant data includes daily values of important atmospheric pollutants such as fine particulate matter(PM_(2.5)),inhalable particulate matter(PM_(10)),nitrogen dioxide(NO_2),sulfur dioxide(SO_2),carbon monoxide(CO),and ozone(O_3),and the corresponding daily mean meteorological parameters include temperature,wind speed,sunshine duration,precipitation,atmospheric pressure,and relative humidity.Firstly,the indicators that have a significant impact onthe air quality were selected through stepwise regression,and then these significant indicators were reduced to a lower dimension through the principal component analysis.According to the theory of principal component analysis,two principal components were extracted from five significant impact indicators,and the cumulative variance contribution rate of the two extracted principal components reached 82.9%.The first principal component was characterized by high loadings of PM_(2.5),CO,and PM_(10) concentrations,implying the significant impact of these parameters on the air quality of Hefei.The second principal component was characterized by high loadings of O_(3) concentration and sunshine duration,demonstrating their important role on the air quality of Hefei.Overall,good correlation was observed between the results from principal component analysis and the air quality index with a correlation coefficient of 0.78.In addition,it is found that the proposed method is more suitable to assess the air quality in winter and under good air quality conditions.These findings highlight the good performance of principal component analysis method on the assessment of urban air quality.