MGTWR Beijing-tianjin-hebei Region PM2.5 Concentration Modeling And Estimation Analysis

In recent years,PM2.5 has seriously affected people’s health and life,and air quality has attracted much attention.Since the Aerosol Optical Depth(AOD)is closely related to the PM2.5 concentration,many studies have been carried out based on this.In order to improve the accuracy of the model,auxiliary factors such as meteorology and land are also added to the model.The influencing factors of different seasons are not exactly the same,so it is necessary to select significant characteristic variables for different seasons to reveal the different spatiotemporal processes of PM2.5 concentration in different seasons.The current Geographically and temporally weighted regression(GTWR)model can better solve the problem of spatiotemporal non-stationarity and obtain the "average bandwidth" in space and time,but this may ignore the difference of spatiotemporal heterogeneity among each factor.In response to the above problems,this paper adopts the stepwise regression method oriented to the spatiotemporal relationship to select the significant characteristic variables of each season,establishes a reliable spatiotemporal model,and uses a multiscale spatiotemporal geographic weighted regression model(Multiscale geographically and temporally weighted regression,MGTWR)modeling analysis..It adds multi-scale effects on the basis of GTWR,and can calculate the independent spatiotemporal bandwidth of each factor,revealing the spatiotemporal heterogeneity of different influencing factors.The paper takes the Beijing-Tianjin-Hebei region as an example to create a spatiotemporal non-stationary model of MGTWR.The results show that summer has the least significant influencing factors,and the goodness of fit(R2)of MGTWR is higher than that of GTWR,all reaching above 0.94.The mean absolute error(MAE)and root mean square error(RMSE)are 11%～34% lower than GTWR.This method further proves that MGTWR is superior to single-scale GTWR in revealing M2.5 concentration in different spatiotemporal processes in different seasons.The results of scaling analysis show that,relatively speaking,AOD and meteorological factors are local short-term influencing factors,vegetation index(NDVI)is a global long-term influencing factor,and population density(PD)is a local longterm influencing factor.The MGTWR model was used to study the temporal and spatial heterogeneity and influence scale of the factors affecting PM2.5 concentration,and the influencing factors were classified according to their differences.Government departments can make different strategic plans for different types of influencing factors,and the research results can provide effective reference for urban air pollution prevention and control.