| In recent years,air pollution has become an important factor affecting human health and sustainable development.Fine particles(PM2.5,aerodynamic diameter≤2.5μm)are important indicators to measure air environment quality,and have been proved by a lot of epidemiological research and are closely related to harmful health diseases.However,because of the small number of ground monitoring stations,uneven distribution,high cost of the established stations and time-consuming maintenance in the later period,it is difficult to obtain the spatial distribution of large area PM2.5concentration.Satellite remote sensing has the advantages of wide coverage and low acquisition cost.It has become a hot research topic at home and abroad to use aerosol optical depth（AOD）to monitor PM2.5concentration in a large range.Taking Sichuan Province as the research object,this paper used MODIS satellite remote sensing data,6S radiative transfer model and land aerosol dark pixel method to retrieve the aerosol optical thickness in the study area from 2014 to 2017,and used MODIS aerosol products to verify the accuracy of the retrieved AOD.The results showed that the accuracy of the retrieved AOD is better.On this basis,the influencing factors of AOD size distribution and the temporal and spatial variation of aerosol concentration in the study area in the past four years were analyzed.According to the different characteristics of AOD monitored by satellite remote sensing data and PM2.5data monitored by ground,the vertical correction and humidity correction of AOD are carried out by using the height of boundary layer and relative humidity respectively.The relationship between AOD and PM2.5corrected by vertical humidity was analyzed,and the fitting relationship between them has been established.Because the meteorological and population density data will affect the change of PM2.5concentration,these factors are added into the fitting relationship between AOD and PM2.5to improve the fitting accuracy of the model.Based on the PM2.5concentration estimated by AOD,the temporal and spatial variation characteristics of PM2.5concentration in the study area in the past four years were analyzed.Finally,the correlation between PM2.5concentration and natural factors（temperature at 2m,total precipitation,wind speed,air pressure,elevation）and human factors(PM10,SO2,NO2,CO,O3,population density)is analyzed to provide scientific basis for the future air pollution control of relevant departments.The main conclusions of this paper are as follows:（1）It was found that the AOD mean is decreasing year by year in time change.In2014,the average AOD is the highest,the lowest is 0.468 in 2017;in spatial distribution,AOD distribution is characterized by high east and low in west,There is a negative correlation between the AOD and the terrain distribution of the study area,that is,the AOD is lower in the high-lying area and higher in the low-lying area.Seasonally,the mean value of AOD was as follows:Spring>Summer>Winter>Autumn,and the values were 0.636,0.546,0.324 and 0.296 respectively.From month to month,the mean value of AOD in April was the highest（0.619）,and that in December was the lowest（0.209）.（2）The results showed that AOD is positively correlated with air pressure;AOD is negatively correlated with total precipitation;AOD is negatively correlated with elevation;AOD is negatively correlated with NDVI;AOD is positively correlated with population density;AOD is positively correlated with regional GDP.（3）After getting the daily average AOD distribution in the research area from2014 to 2017,the AOD was corrected vertically and humidity.The revised AOD was matched with PM2.5at the time of transit on that day.The AOD and PM2.5models were established according to the cities.Because PM2.5concentration is affected by meteorological factors,population density,elevation and other factors,this paper adds eight factors including temperature,relative humidity,boundary layer height,total rainfall,wind speed,air pressure,population density and elevation into AOD PM2.5relationship model,and constructs multiple regression model,geographically weighted regression model and linear mixed effect model of each city.The results showed that the linear mixed effect model has the highest fitting accuracy,and the fitting accuracy R of Chengdu,Zigong,Luzhou,Deyang,Mianyang,Suining,Leshan,Nanchong,Meishan,Yibin,Dazhou,Bazhong and Ziyang are all above 0.950;the fitting accuracy R of Guangyuan,Neijiang,Guang’an,Ya’an and Liangshan are between 0.900-0.950;the fitting accuracy R of Panzhihua,Aba and Ganzi is between0.900-0.950 Degree R is between 0.800and 0.900.（4）The annual distribution characteristics and regional differences of PM2.5concentration in the study area from 2014 to 2017 are as follows:PM2.5concentration has a downward trend year by year,among which,the average annual PM2.5concentration in 2017 has a slight increase on the basis of 2016;in winter,PM2.5concentration is the highest,and the lowest in summer;the spatial distribution of PM2.5concentration is similar to that of AOD,and the high value area is mainly distributed in the east of Hu-Huan line,The low value area is mainly distributed to the west of Hu-Huan line.（5）The correlation between PM2.5and natural factors in 2014-2017 is:temperature>total precipitation>relative humidity>air pressure>wind speed>elevation>boundary layer height at 2m;the correlation between PM2.5and human factors is PM10>NO2>SO2>CO>population density>O3.PM2.5concentration and distribution are mainly affected by human factors.The production of human beings will burn a large amount of fossil fuels,which leads to the increase of PM2.5emissions.Among the natural factors analyzed in this paper,the temperature has the greatest influence on the distribution of PM2.5concentration,because the higher the temperature is,the faster the air flow rate is,thus the faster the particle diffusion speed is,and the lower the PM2.5concentration. |
