Study On The Applicability Of Atmospheric Correction Methods For Hyperspectral Remote Sensing Data Of Different Surface Type

The differences in physical properties such as composition and roughness of different surface types determine their different abilities to reflect solar radiation,which leads to significant differences in the radiation values at the sensor pupil of different surface types.Therefore,the atmospheric correction effect is also different for different surface types.Atmospheric correction is one of the very important processes in remote sensing image processing,and the effect of atmospheric correction has a certain influence on the accuracy of subsequent quantitative remote sensing inversion of each parameter.In this paper,GF-5hyperspectral data is used as the data source to analyze the applicability of atmospheric correction methods for different surface types(water,vegetation and soil),and the QUAC model,dark image element method and FLAASH model as well as the atmospheric correction method of AOD and water vapor synergistic inversion proposed in this study are selected to study the applicability of different surface types,and the main results of the study are as follows.(1)Based on the sensitivity analysis of the parameters of the atmospheric correction algorithm,it was found that AOD and water vapor have a large influence on the atmospheric correction results,so the atmospheric correction method of synergistic inversion of AOD and water vapor was proposed.Based on the GF-5 data,the AOD and water vapor contents were co-inverted and analyzed with the simultaneous MODIS aerosol and water vapor products,and the relative errors of the inverse mean values were small,14.02% and 6.91%,respectively.The calibration results of the atmospheric correction method using synoptic inversion of AOD and water vapor are relatively good,and the surface reflectance data are relatively stable;(2)The information entropy and mean gradient indexes were used to analyze the information richness and clarity of the corrected images and different surface types of the QUAC model,the dark image element method and the FLAASH model,as well as the atmospheric correction method of the synoptic inversion of AOD and water vapor in this paper.The results show that: overall,the two indexes of the atmospheric correction method of synoptic inversion in this paper are higher than other methods;in two types of features,water and vegetation,the information richness and image clarity of the synoptic inversion method are better than other algorithms;in this type of features of soil,the FLAASH model is better than other methods;(3)The calibration results of the QUAC model,the dark image element method and the FLAASH model,as well as the atmospheric calibration method for the synoptic inversion of aerosol optical thickness and water vapor in this paper,are cross-validated with MODIS and Landsat8 OLI surface reflectance products.The results show that in the soil type,the atmospheric correction method of synoptic inversion of this paper has less errors,and the relative errors in the blue,green,red and near-red bands with MODIS and Landsat8 OLI surface reflectance products are 0.09%,11.47%,29.83%,3.70% and 9.77%,4.92%,6.01%and 1.38%,respectively;in the vegetation type,the The FLAASH model calibration results were better,with relative errors of 12.08%,15.18%,32.29%,6.75% and 21.61%,3.16%,8.86%,1.08% in the blue,green,red and near-red bands with MODIS and Landsat8 OLI surface reflectance products,respectively.