Influence Factors Analysis, Prediction And Simulation Of Pm_2.5 Concentration In The Atmosphere Of Changsha

The study takes the central urban area of Changsha as the research area.And the monitoring data of PM_2.5 concentration from 2013 to 2017,which come from ten air-quality monitoring points in Changsha,are used to analyze the temporal-spatial distribution characteristics of PM_2.5concentration in the atmospheric environment of the City.Then the impact factors which significantly relate to PM_2.5 concentration are obtained based on the bivariate correlation analysis between the monitoring data of PM_2.5 concentration and its influence factors,such as land use,road traffic,population density,topography,meteorology,dust,industrial pollution and restaurant sources.Finally the land use regression（LUR）model for the four seasons of the research area is established on the basis of stepwise regression analysis.And with the help of GIS spatial analysis platform,the predicted and simulated PM_2.5 concentration in2018 is carried out on the spatial-temporal scale.The main conclusions are as follows:（1）According to the analysis of the temporal and spatial distribution characteristics of PM_2.5 concentration,the temporal changes showed that the annual average concentration of PM_2.5 was 81.83μg/m³、74.85μg/m³、60.16μg/m³、53.52μg/m³and 52.13μg/m³ respectively from2013 to 2017,which showed a obvious downward trend;in different seasons,it presented the change track of“high in winter and low in summer,fall in spring and rise in autumn”;the hourly average curve of PM_2.5 concentration in different seasons was roughly bimodal and double valley.In terms of spatial variation,the concentration of PM_2.5 is relatively low in areas with large vegetation area coverage and those close to waters,while it was relatively high in areas with dense traffic roads,high population density and those close to industrial parks.（2）Through the bivariate correlation analysis between PM_2.5concentration and its impact factors,the influence factors with large correlation coefficient were obtained.As for the road traffic factors,the main road density,the secondary trunk density were positively correlated with the PM_2.5 concentration,and the buffer zone with strong correlation was in the range of 500 to 1000 meters.In terms of the land use factors,the vegetation and water had a negative correlation with the PM_2.5concentration,and the buffer zone with strong correlation was in the range of 500 to 1000 meters.Among meteorological factors,the precipitation had a significantly negative correlation with PM_2.5concentration.Besides,as for social economic factors,the industrial pollution factors were the largest correlation coefficient showing a positive correlation.（3）The LUR model of the spring,summer,autumn and winter in the research area was ideally constructed.The adjusted R² values of the seasonal models were 0.914,0.721,0.928,0.779 respectively,which were all above 0.7.And the average absolute error rate of the test samples was below 15%.So the model’s independent variable could account for more than 70%of the variation of PM_2.5 concentration.（4）The LUR model was used to predict and simulate the PM_2.5concentration in 2018 on a spatio-temporal scale.The results showed that the average error rates of each season were 0.046,0.058,0.038,0.021respectively and most of the error values were between zero and fifteen.Therefore there was a higher fitting degree between the predicted value and the measured value.On the spatial scale,the high concentration aggregation of PM_2.5 in the study area was more prominent than the interpolation results based only on the monitoring data of the air quality monitoring points.As a whole,the PM_2.5 concentration gradually decreased from the central area of the city to the surrounding direction.