Research On Inversion Of Aerosol Optical Thickness Based On Landsat8 OLI Image

Atmospheric aerosols affect the global atmospheric radiation balance,changes in the ecological environment,and human health,and are a hot topic of research at home and abroad.In recent years,with the acceleration of my country’s urbanization process and rapid economic development,the problem of air pollution in urban areas has become increasingly severe.The existing aerosol optical depth(AOD)products have low spatial resolution,which cannot meet the small-scale and refined air pollution monitoring needs in urban areas.The urban area surface cover is complex and changeable,and it is challenging to retrieve high-precision AOD.The research uses Landsat-8 Operational Land Imager(OLI)images as the data source,and Beijing is the research area to carry out aerosol inversion research,aiming to obtain high-precision,high-spatial resolution AOD products.According to the different land cover types in the study area,two methods are used to determine the surface reflectance,combined with the aerosol characteristic data obtained from the AErosol RObotic NETwork(AERONET)site in the study area,to define the aerosol type,to achieve high spatial resolution AOD inversion,and to retrieve the results It has been cross-compared and verified with AERONET and Moderate-Resolution Imaging Spectroradiometer(MODIS)AOD products.The main findings are as follows:(1)The surface reflectance relationship of vegetation areas is inherited from the mature MODIS Dark Target(DT)algorithm.The research uses the spectral response function of the sensor and the measured spectral curve in the ASTER surface object spectrum library to correct the band difference,and then combines the conversion relationship between NDVIMODIS1.24-2.12 and NDVIOLISWIR to re-derive the surface reflectance suitable for OLI relationship.Based on this relationship,the correlation coefficients between the red and blue band surface reflectance and Landsat-8 Collection2 Level2 surface reflectance products are both 0.92,the RMSE are 0.006 and 0.003,and the MAE are 0.002 and 0.004,respectively,with high accuracy.(2)The urban built-up area is a typical high-bright surface,showing a relationship of surface reflectance that is different from that of vegetation areas.The research is based on the 6SV model and AERONET aerosol characteristic data to perform atmospheric correction on OLI images to obtain the true surface reflectance in the red and blue bands.The least squares method was used to refit the functional relationship between the red and blue bands and 2.2μm,short-wave infrared vegetation index and scattering angle.The surface reflectance calculated according to this function has high consistency with the "benchmark" surface reflectance obtained by atmospheric correction.The correlation coefficients are 0.88 and 0.92,the RMSE is 0.012 and 0.015,and the MAE is 0.008 and 0.01.(3)Aerosol type is one of the key factors affecting the inversion of aerosol optical thickness.Based on the monthly average birefringence index,particle average radius,radius standard deviation,and volume concentration of 5 AERONET sites from 2015 to 2020,the study defines the bimodal log-normal spectrum distribution aerosol types divided by month,and the monthly aerosols The type is more in line with the actual distribution of aerosol particles in the study area,and can effectively improve the accuracy of aerosol optical thickness inversion.(4)Verify the accuracy of inversion results based on MODIS Collection 6.1 AOD products and AERONET observation data.OLI-AOD is highly correlated with the observations from the AERONET site(R=0.96),the RMSE is 0.1,and 72.27%of the value lies within±0.2*(AODAERONET+0.05)expected error range.The fitting relationship is AODOLI=0.8925*AODAERONET+0.0343.It is also in good agreement with MODIS DT and DB AOD product,with correlation coefficients of 0.72 and 0.74,and RMSE of 0.109 and 0.056.The fitting equations are AODOLI=1.116*AODDT+0.0485 and AODOLI=1.106·AODDB-0.0044.The above results show that the inversion algorithm proposed in this study has high accuracy and stability,and can be applied to small-scale air pollution monitoring in urban areas.