蚀变信息提取是高光谱遥感地质应用的重要内容。 基于特殊吸收峰的蚀变矿物提取是蚀变信息提取的重要手段。 由于大气的吸收和散射作用, 为了获得更为真实的地物反射光谱, 必须进行大气校正。 目前, 国内外针对大气校正的对比研究主要集中在大气校正前后图像的质量改善、 地物分类效果的提升以及校正图像像元光谱与实际地物光谱的相关关系等方面, 而对不同校正方法获得的像元光谱与实际光谱吸收峰位的对应情况则很少讨论, 这对于依赖吸收峰特征进行蚀变矿物提取的地质遥感极为不利。 利用CASI-SASI航空高光谱成像系统, 采集了甘肃龙首山地区的航空高光谱遥感数据, 并运用ASD光谱仪, 对该地区实际地物光谱进行了测量。 以此为基础, 开展了FLAASH、 快速大气校正（QUAC）、 经验线（EMPL）等方法大气校正结果的对比研究。 通过对比分析, 发现FLAASH, QUAC和EMPL均能在一定程度上消除大气的影响, 改善航空高光谱遥感的图像质量, 但EMPL方法得到的反射率与实际反射率相关性最好。 此外, 运用人工目视方法开展了实际地物反射光谱的吸收峰位与不同校正方法得到的对应像元反射光谱的吸收峰位的对比研究, 发现不同校正方法得到的像元光谱的吸收峰位与实际峰位均存在不同程度的差异, 虽然EMPL对吸收峰位的保留效果最好, 但依然有“漏峰”的现象。 据此, 提出运用多种大气校正方法开展综合研究, 以提高不同类型的蚀变带定位准确度。

Alteration information extraction is one of the most important aspects of hypersepctral remote sensing application, and alteration mineral identification based on special absorption peaks is an important means of alteration information extraction. Due to the absorption and scattering of the atmosphere, atmospheric correction must be performed in order to obtain a more realistic reflection spectrum of the ground object. At present, comparison researches on atmospheric correction mainly focus on the improvement of image quality before and after atmospheric correction, the improvement of the classification effect of different features, and the correlation between the corrected image pixel spectrum and the actual spectrum. In contrast, the correspondence between absorption peaks of pixel spectrum obtained using different atmospheric correction and actual spectrum is rarely discussed, which is extremely unfavorable for geological application of hyperspectral remote sensing that is based on identification of mineral absorption peaks. In this study, the CASI-SASI aeronautical hyperspectral imaging system was used to collect the airborne hyperspectral data of the Longshoushan area of Gansu Province. Additionally, spectra of terrain objects were obtained using ground based ASD spectrometer. Based on these datasets, a geological application oriented comparison research on FLAASH, QUAC, and EMPL atmospheric correction methods was conducted. The results suggest that FLAASH, QUAC, and EMPL can eliminate the effects of the atmosphere and improve the image quality of aerial hyperspectral remote sensing, and EMPL shows the best effect. Besides, the absorption peaks of terrain objects’ reflection spectra and corresponded pixel reflection spectra obtained with different atmospheric correction were compared using visual interpretation. It is found that the absorption peaks of pixel spectra have different degrees of differences from those of the real terrain object spectra. Although EMPL has the best retention, some absorption peaks are still error corresponded, suggesting that multiple atmospheric correction methods should be used and a comprehensive research should be carried to improve the alteration extraction accuracy.