Study On The Temporal And Spatial Variation Of PM_2.5 In China Based On LUR Model

Accurate spatial information of PM2.5 is critical for air pollution control and epidemiological studies,and land use regression(LUR)models have been widely used to predict the spatial distribution of ground PM2.5.However,due to limited ground observations,the predicted PM2.5 spatial model of the LUR model has not been fully studied,and aerosol optical depth is used in the study of obtaining continuous PM2.5 concentrations in large areas.In order to explore the adaptability of LUR model in China’s national air pollutant simulation,the spatial and temporal distribution of PM2.5 is accurately understood,the internal mechanism and transmission law of its distribution is revealed,and the temporal and spatial variation characteristics of air fine particles in China in 2015 are explored.Interrelated with different geographic elements.This study is based on the idea of traditional land use regression modeling,with monitoring site PM2.5 concentration as the dependent variable,land use type,altitude,population,road traffic,aerosol optical depth,normalized vegetation index,and 7 types of meteorology.The different geographical factors as independent variables and the LUR model with ordinary least squares is constructed.Then the geographically weighted regression algorithm is introduced to construct the geographically weighted land-use regression model using the effective variables of the model obtained after stepwise regression.Then the fitting results of different types of models are obtained.The paper analyzes the evaluation index and comprehensively verifies the spatial pattern of the LUR model.Finally,it calculates the spatial distribution of the monthly average and annual average PM2.5 concentration in China in 2015 and analyzed the time and space of China’s 2015 PM2.5 concentration with the Hu Huanyong line as a reference line.Change characteristics,analysis of changes in PM2.5 concentration in key control areas of air pollution prevention and control.Through the construction of the national scale PM2.5 concentration distribution model,regression mapping,evaluation of the model,and the analysis of the temporal and spatial distribution of pollutants,there are four conclusions:(1)In the national scale PM2.5 concentration distribution fitting study,the stepwise regression and OLS model evaluation indicators indicate that the model is significant overall,the fitting results of geographically weighted regression,Moran’s I index,and residual distribution are evaluated.The indicator indicates that the model fits well.The LUR model introducing geographically weighted regression better expresses the relationship between PM2.5 concentration and various geographical elements,and reveals to some extent the effects of various geographical elements in different time and different regions on PM2.5 concentration.Shows greater adaptability.(2)There are significant differences in the influence of different variables on PM2.5 concentration in different time and space.Land use type,aerosol optical depth,and meteorological factors are the decisive factors affecting PM2.5 concentration change.Urban and rural industrial and mining residential land types participate in the construction of most models are important to influence factors for analyzing and predicting the change of PM2.5 concentration.Aerosol variables have relatively high relative weight values in all models and are closely related to PM2.5 concentration,which is an important factor for simulating PM2.5 concentration.The influence degree of different meteorological factors on PM2.5 concentration in different months varies with the seasons.The average weight of meteorological elements in the 12 model reaches 42.85%,indicating that the influence of meteorological elements on PM2.5 concentration is remarkable.(3)The research shows that the distribution of PM2.5 concentration at the national scale has significant spatial and temporal differences.The concentration of pollutants on both sides of the Hu Huanyong line shows an indiscriminate distribution patterm.The concentration of PM2.5 in the northeastern plains with high population density,high industrialization level and agricultural development in the south of Hu Huanyong line is higher;the concentration of PM2.5 in winter is higher,the concentration of PM2.5 is lower after entering the spring,and the pollution situation is gradually improved.(4)Compared with the single aerosol product using remote sensing satellite imagery,the PM2.5 concentration data fitted by the multi-source element based on the ground monitoring site data and introduced by the geographically weighted regression modeling method is more accurate and scientific.