Variation characteristics and influencing factors of PM2.5 and O3 based on machine learning in Fenwei Plain
Based on the concentrations of PM2.5 and O3 in Xi'an,Fenwei Plain from 2017 to 2021,this study analyzed the change characteristics and trend of PM2.5 and O3,and discussed the interaction effects of pollutant gases(NO2,SO2,CO and HCHO)and meteorological factors(temperature,RH,wind speed,atmospheric pressure,boundary layer height and solar radiation)on PM2.5 and O3 by using machine learning method.Theil-Sen trend analysis found that PM2.5 and O3 decreased by 6.03％and 2.06％per year from 2017 to 2021,respectively.For the single influencing factor GAM models,the model explanation rate of the effects of NO2,SO2 and CO on PM2.5 is higher,temperature,solar radiation and pressure on O3 is higher.In the multiple influencing factors GAM model,all factors exhibited a non-linear relationship with PM2.5 and O3,and the contributions to the change of PM2.5 and O3 are 84.9％and 75.0％with significant impact,also suggesting a good model fit.Contour map were used to analyze the pairwise interaction of several meteorological factors and polluting gases on the concentration of PM2.5 and O3,respectively,which found that temperature and pollution gases(NO2,SO2,CO and O3)have considerable impact on PM2.5 concentration.For O3,the influence of temperature and solar radiation is greater.NO2 and CO interact with meteorological conditions,PM2.5 increased with the increase of NO2 and CO,however O3 showed opposite trends.According to the local pollutant source inventory,it is suggested to strengthen the emission of pollutants from industrial sources and mobile sources,which helps to reduce the concentration of PM2.5 and O3.
NETL
NSTL
万方数据
