Multi-source Satellite Data Synergy For The Retrieval Of Aerosol Optical Depth

This research centers on the multi-source satellite data synergy method,which is one of the research focus in the scientific community,for the retrieval of Aerosol Optical Depth(AOD).In view of the characteristics of multi-source satellite data and its demand for scalability,a novel N-Dimensional Cost Function(NDCF)method is proposed in this research based on the radiative transfer model considering BRDF effect.A systematic sensitivity study is conducted on the NDCF method.The proposed method is further instantiated for specific data sets,i.e.multi-sensor satellite data sets of MODIS and VIIRS,and multi-temporal Himawari-8 geostationary satellite data sets.Comparison,validation and analyses are conducted for each instantiation of the NDCF method.The main achievements and innovations of this research are as follows:(1)A novel N-Dimensional Cost Function(NDCF)method is proposed.The NDCF method is well fitted for the demand of multi-source satellite data synergy.The proposed method allows AOD variation among different observations,and we build the Look Up Table(LUT)for different sensors and different bands seperately for the NDCF method.The cost function,which is constructed based on the assumption that the BRDF shape function is independent of wavelength,can take full use of the multi-angle and multi-band information.Moreover,the dimension number of the cost function is flexible,which extends the scalability of the NDCF method.(2)To fit the BRDF with stability and accuracy,in the NDCF method,a novel Shape Function Constrained BRDF Retrieval(SFCBR)method is proposed.The SFCBR method fits the BRDF under the constrain of the a priori knowledge of BRDF shape function.The constrain function we constructed can tie the satellite observation and the a priori knowledge together and only the shape of the a priori BRDF knowledge is considered rather than its reflectance magnitude.The retrieval result of the NDCF method can be significantly improved by adopting the SFCBR method especially in the situation when the redundant observation data is insufficient.(3)A series of sensitivity studies of the proposed multi-source satellite data synergistic AOD retrieval method are conducted systematically.The sensitivity study is conducted on,respectively,the dimension number of the cost function,the real AOD,the variation of BRDF shape function and the aerosol model to prove the robustness of the proposed method.The relative error over standard vegetation region is found lower than that over standard sand region and the dust aerosols generate higher AOD relative error than non-dust aerosols.(4)The proposed method is applied to the multi-sensor satellite data sets of MODIS and VIIRS.Compared to the MODIS and VIIRS official product,the proposed method shows comparative accuracy at the four selected AERONET sites in East Asia,and the Expected Error(EE)of the retrieved AOD from the proposed method is estimated as ? = ±0.05 ± 0.24.Besides,an obvious improvement in the temporal resolution of the BRDF parameters is achieved by adopting the SFCBR method.The proposed method is also applied to the multi-temporal Himawari-8 geostationary satellite data sets.Over North China Plain,the NDCF method improves the AOD retrieval accuracy of the Himawari-8 data and can compensate the noticeable underestimation phenomenon of the Himawari AOD official product,with the EE of ? = ±0.08 ± 0.19.As the number of satellites in the outer space increases and the instruments get updated,the proposed multi-source satellite data synergistic AOD retrieval method should have huge application potential and strong flexibility compared to traditional AOD retrieval method.This work is instructive for further study on multi-source satellite data synergetic AOD retrieval which is the trend for remote sensing.