Study On Satellite Remote Sensing Inversion Method Of Ambient PM_2.5 In Chengdu

With the improvement of people's daily life quality,production and lifeway for human being has caused different degrees of environmental pollutions.As far as air pollution is concerned,the ambient particulate matters such as PM2.5 and PM10 were still one of the emphasis and difficulty for air pollution prevention projects.Presently,using satellite remote sensing to quantitative retrieving aerosol particle distribution has become a rising measurements on monitoring atmospheric particulate matter concentration,so that the spatial and temporal distribution characteristics of ambient pollutants for large scale can be observed.This measure not only overcome the monitoring range limitations by few ground sites,the spatial distribution rules of remote area without sites also be well monitored and make a further step to provide reliable technical support for local government departments in the formulation and decision-making on air pollution prevention strategies.Taking Chengdu as the research area,this paper uses the 6S(Second Simulation of the Satellite Signal in the Solar Spectrum)with local atmospheric,aerosol model and 6S midlatitude atmospheric,urban aerosol model to construct look-up table(LUT)respectively,then quantitative retrieval of aerosol optical depth(AOD)in Chengdu by MODIS L1B remote sensing data products,and revised the two model's AOD by vertical and humidity correction,finally the quality concentration inversion model of ambient PM2.5 in Chengdu has been established and the correlation analysis was also carried out.The main results for this study are as follows:(1)Using ozone monitoring data from Jiaotong university station and water vapor data retrieved through the model to calculate the total amount of ozone and water vapor in the satellite transit period of each day,and then input into 6S model to collect the local atmospheric model parameters in Chengdu city.According to the observation data of aerosol particle number concentration from the particle size spectrometer,the distribution spectrum function curve of the monthly average aerosol particle size distribution in Chengdu is fitted out.The aerosol size distribution spectral function in Chengdu follows Junge power law distribution in the month of Jan.,Feb.,Jun.,Jul.,Aug.and Sep.,besides,the spectral function in Chengdu follows modified Gamma distribution in the month of Mar.,April,May and Oct.Then put the distribution parameters into 6S model and obtained the loop-up table of local model.(2)Based on the look-up table of local model and 6S midlatitude atmospheric,urban aerosol model,then programing interpolation inversion 57days AOD in Chengdu,2017,by using the remote sensing development software ENVI+IDL with those models.Comparing the AOD with near ground PM2.5 in Chengdu found that the correlation between AOD and particulate matter concentration from both of models were low,however,the correlation from local model(R~2=0.2305)is better than that of 6S(R~2=0.1676).(3)The vertical correction and humidity correction of the AOD inversion under the two models are carried out with the aerosol boundary layer and humidity data from Jiaotong university station,the overall correlation of the corrected AOD and PM2.5 particle concentration is up to 0.4026 and 0.2678,respectively.According to the seasonal analysis,the correlation from both models in the autumn is highest,reaching0.7652 and 0.7203 respectively.The correlation coefficient in winter for local model and in spring for 6S midlatitude atmospheric,urban aerosol model is low,reaching0.3042 and 0.1882 respectively.The linear mode of PM2.5 concentration inversion has established in two models and the best fitting statistic model is quadratic equation.(4)The statistical map of polygonal line has been made and shows the two-model's trend of concentration among the inversion concentration of PM2.5 with it real-time concentration values in spring,summer,autumn and winter.Compared with 6S incidental model,the inversion value of local model is more closed to PM2.5 real-time concentration values in spring and autumn.In general,the accuracy of the PM2.5 inversion model in the local mode is relatively higher than 6S incidental model.