抄録
Possible techniques for lithologic discrimination using spectral bands in the short wavelength infrared (SWIR; 1.3-2.5, μm) region have been tested. Image data of Queensland, Australia, obtained by Geoscan AMSS (Airborne MultiSpectral Scanner) MkI were used in this study. As this scanner has quite similar spectral band characteristics to the optical sensor of the JERS-1 (Japanese Earth Resources Satellite), which is scheduled to be launched in 1992, the techniques proposed for the JERS-1 data were applied to the AMSS MkI data.
The proposed techniques enhance inter-band response patterns in reflectance space, so that it is necessary to convert original digital numbers (DN) to reflectance of surface materials. The linear regression method using field reflectance measurements could not be applied in this case, because it was difficult to find good calibration targets in the study area. Instead, the Log Residual technique, which normalizes the DN values using geometric means of each pixel and each band, succeeded in converting DN values to band responses resembling to reflectance patterns.
Two spectral indices; Alunite Index (ALI) and Calcite Index (CLI), were calculated by linearly combining the three SWIR bands normalized by the Log Residual method. The Perpendicular Vegetation Index (PVI) was also calculated using two visible and near infrared bands. The images of these spectral indices successfully showed an epidote-rich zone, sericite-rich zones in hydrothermally altered areas, and distribution of vegetation. These techniques have the advantages that the processed results can be easily interpreted by a geologist for the purpose of lithologic discrimination.
