Atmospheric Correction Of Gf-5 Ahsi Data And Accuracy Evaluation

The Advanced Hyper-spectral Imager (AHSI) data on-board GF-5 satellite fills up China's lack of hyper-spectral earth observation technology. However, using GF-5 AHSI data for remote sensing applications, especially quantitative remote sensing applications, a challenging problem is to retrieve surface reflectance through atmospheric correction. In this paper, the MODerate spectral resolution atmospheric TRANsmittance algorithm and computer model (MODTRAN) radiative transfer model was used to perform atmospheric correction experiments (AHSI-AC) for GF-5 AHSI data. And we validated the atmospheric correction results based on measured surface reflectance and Landsat 8 surface reflectance data. Using field-measure spectra for validation the AHSI-AC method has a relative error of about 10% in most bands. Compared with the ENVI-FLAASH algorithm, the accuracy of the blue-green bands and the short-wave infrared bands are significantly improved, and the red and near-infrared bands are similar. Using Landsat 8 data for intercomparison, the large homogeneous, non-vegetated pixels were selected, and the overall accuracy of the 515 selected validation points reached 81.7%.