Spatial And Temporal Analysis Of Aerosol Optical Depth Inversion In Guizhou Province

With the acceleration of urbanization and the rapid development of social economy,air pollution has become a hot issue of common concern.Aerosol,as a basic element to change the income and expenditure of ground gas radiation and the formation of cloud droplets,affects the global climate change.AOD(Aerosol Optical Depth),as an important optical characteristic parameter of aerosol,is the integral of extinction coefficient of atmospheric medium in vertical direction.High precision AOD can not only accurately reflect the situation of air pollution,but also provide basic data support for the prevention and treatment of air pollution by relevant functional departments.Because of the advantages of large space scale and good continuity of time series,satellite remote sensing has become one of the important means of AOD monitoring.On the basis of MODIS L1B satellite remote sensing image data from 2010-2019,firstly uses Dark Dense Vegetation method to inverse aerosol optical depth in Guizhou Province for10 years,and uses MOD04 aerosol product as a reference to verify the accuracy of inversion results.Then based on the AOD inversion results,the variation characteristics of aerosol optical depth time domain and space domain in Guizhou Province are revealed.Finally,meteorological data,ground digital elevation and vegetation data were selected as the influencing factors,interannual and seasonal estimation models of PM_2.5 concentration were constructed by random forest and multivariate linear regression analysis respectively,and with the measured value of PM_2.5 as a reference,the accuracy of the predicted value of PM_2.5 concentration was verified.The main work and conclusions are as follows:(1)The Pearson correlation coefficient between AOD inversion results and MOD04aerosol products is 0.84,indicating that the aerosol optical depth obtained from MOD021KM inversion has high reliability.(2)In terms of time domain,the optical depth of aerosol in Guizhou province fluctuated between 0.091-0.7285,But from the overall downward trend;AOD seasonal differences are significant,performance:summer>spring>autumn>winter.There are some differences in the spatial distribution of aerosol optical depth in Guizhou province.The high value area is mainly concentrated in the middle part of Guizhou province,and the AOD gradually decreases from the middle part to the outer part,but occasionally the AOD value increases in a few areas,such as the south and southeast of Guizhou province.(3)The correlation between AOD and PM_2.5concentration of the foundation is strong,which indirectly proves the correctness of the AOD inversion results and lays a foundation for the model construction based on the PM_2.5 concentration of the AOD estimation.The correlation analysis between PM_2.5 concentration and modeling factor showed that PM_2.5 was only positively correlated with relative humidity,negatively correlated with wind speed,ground digital elevation and enhanced vegetation index,while the correlation between PM_2.5 and air pressure and temperature varied with season.(4)Based on the AOD inversion results,the interannual and seasonal prediction models of PM_2.5 concentrations were constructed using random forest and multiple linear regression.Experimental results show that compared with multiple linear regression,the interannual and seasonal models of PM_2.5 concentration based on random forests have higher prediction accuracy,and the prediction accuracy of spring and summer section models is higher than that of autumn and winter section models.