The Research Of Analyzing PM2.5 Concentration Based On Ground-based GNSS And MODIS

In recent years,with the rapid development of China’s economy,urban air pollution has become one of the outstanding problems affecting the ecological environment,and PM2.5 pollution is the most serious.As the main source of air pollution,the PM2.5 concentration is usually monitored by the ground stations.However,because of the rare quantity and uneven distribution of stations,this way of obtaining the PM2.5 concentration has faced with many difficulties,for example,it is not possible to carry out in real time on-line monitoring.The ground-based GNSS technology and remote sensing satellite images has been widely used to monitor the PM2.5 concentration.This paper combined the advantages both of ground-based GNSS technology and remote sensing,comprehensively using the Precipitable Water Vapor(PWV),Aerosol Optical Depth(AOD)and air quality,weather,geographic location data to construct the Geographically Weighted Regression(GWR)model for analyzing the distribution characteristics of PM2.5 in Shaanxi.Analyzed and evaluated the applicability and simulation accuracy of the GWR model,found that the proposed GWR model considered multi-factors and spatial differences which is more suitable for the spatial agglomeration characteristics of PM2.5 in Shaanxi area.Furthermore,the GWR model can predict higher accuracy of PM2.5 concentration in the uncovered area of the station.The main research contents of this article are as follows:(1)PM2.5 concentration analysis based on ground-based GNSS technology.Using the 2018 PWV distribution data in Shaanxi,analyzed the changing trends and correlation of Zenith Tropospheric Delay(ZTD)and Air Quality Index(AQI),ZTD and PM2.5 concentration,PWV and PM2.5 concentration by seasons,and constructed a multiple linear regression model of PWV,meteorological data and PM2.5.The results showed that during the period of air pollution,the PWV and PM2.5 concentration change trends are consistent and have a significant positive correlation.To a certain extent,PWV could influence the change of PM2.5,and the PWV change could be used to study the PM2.5 concentration.At the same time,meteorological factors have different effects on the concentration and transmission of PM2.5,the multiple regression model of PM2.5 concentration based on PWV and meteorological data has satisfying effect,which can provide a reference for the study of PM2.5 concentration.(2)PM2.5 concentration analysis based on MODIS images.Using the MODIS data in Shaanxi to extract the AOD value,analyzed the changing trend and correlation of AOD and PM2.5 by seasons,and constructed a regression model of AOD and PM2.5.The results show that the correlation coefficients of AOD and PM2.5 concentration in spring,summer,autumn and winter were 0.631,0.305,0.873,0.689 respectively and presented a significant positive correlation which will aggravate the PM2.5 concentration to a certain extent.That said,the higher the regional AOD value is,the more serious the aerosol distribution will be.Therefore,it is meaningful to study the PM2.5 concentration according to the aerosol changes.(3)Using the GWR model to simulation PM2.5 concentration.Combining the GNSS PWV,MODIS AOD and air quality,meteorology,geographic location data to construct the geographically weighted regression model which could be suitable for obtaining the characteristics of PM2.5 concentration by seasons and analyzing the influence of PWV and AOD at different positions on PM2.5 concentration.The results show that the degree of influence of PWV and AOD on PM2.5 concentration varies with location and seasons.The GWR model that comprehensively considers PWV,AOD,meteorological factors,geographic location,and other multi-influencing factors is superior to the PM2.5 prediction model based on PWV and meteorological data and the PM2.5 prediction model based on AOD,and also performs better than those that do not consider spatial differences Ordinary Least Squares(OLS)global regression model.Among them,R2 of GWR model were 0.794(spring),0.607(summer),0.884(autumn),0.937(winter),compared with OLS regression model,R2 increased by 20.852%,21.888%,7.805%,12.892%and fitting precision increased by 31.256%,17.165%,24.818%,45.939%,the fitting degree and precision of GWR model had been greatly improved.At the same time,the relative accuracy of the PM2.5 concentration predicted by the GWR model reached 90.586%,88.731%,87.878%,92.425%,all exceeding 80%,the model has good performance,high prediction accuracy and strong practicability which is more suitable for the simulation of PM2.5 concentration,and can provide a reference for the prediction of PM2.5 concentration in the uncovered area of the monitoring station.