Accurate identification of alteration minerals in remote polar regions remains a significant challenge due to limited field accessibility, extreme environmental conditions, and the inherent spectral complexity of exposed lithologies. Satellite-based multispectral remote sensing, particularly using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), provides an efficient means of investigating surface mineralogy. However, ASTER data are often affected by spectral redundancy, sensor noise, and overlapping absorption features that reduce mineral classification accuracy. This study introduces a hybrid dimensionality reduction and spectral classification workflow integrating Minimum Noise Fraction (MNF), Independent Component Analysis (ICA), and Spectral Angle Mapper (SAM) algorithms to enhance lithological mapping in polar environments. The workflow was applied to ASTER imagery from South Victoria Land, Antarctica, where exposure of metamorphic and hydrothermally altered rocks offers an ideal setting for method evaluation. The MNF transform was first used to suppress noise and extract high-variance components, while ICA separated independent spectral sources representing unique mineralogical signatures. Subsequently, SAM classification calibrated with United States Geological Survey (USGS) reference spectra enabled precise identification of alteration minerals. The integrated MNF–ICA–SAM approach effectively discriminated key alteration minerals including alunite, kaolinite, jarosite, chalcedony, opal, and hematite, corresponding to diagnostic Al–OH, Fe–OH, and hydrous silica absorptions across the VNIR–SWIR spectrum. Comparative analyses demonstrate that the hybrid workflow significantly improves spectral separability and classification accuracy compared with single-method techniques. These results highlight the potential of integrated spectral processing as a robust, transferable, and data-driven framework for mineral mapping in polar terrains. The proposed methodology not only enhances the geological interpretability of ASTER imagery but also establishes a foundation for future integration with hyperspectral, UAV, and machine-learning-based systems in remote and data-limited environments.