Integrating Ancillary Topographic and Spectral Indices for Enhanced Glacier Terrain Classification Using the Maximum Likelihood Classifier

Accurate classification of glacierized terrains is essential for effective glacier monitoring and environmental assessment. However, challenges persist due to spectral similarities among classes and the limited spectral sensitivity of remote sensing data. This study addresses these challenges by employing the Maximum Likelihood Classifier (MLC) enhanced with ancillary topographic and spectral indices to improve classification accuracy. The research focuses on the Kolahoi Glacier region in the Kashmir Valley, using Digital Elevation Model-derived topographic attributes (slope, aspect, plane curvature, and profile curvature) and transformed spectral data from LISS-III imagery (Red/SWIR ratio, NIR/SWIR ratio, and Normalized Difference Snow Index). A total of nine glacier terrain classes in the form snow, ice, mixed ice, debris, supraglacial debris, periglacial debris, water, vegetation, bare land, and shadow were delineated. Multiple classification scenarios integrating ancillary datasets with spectral bands were evaluated to identify the optimal combination for terrain mapping. The highest classification accuracy (90.75%) was achieved by combining the Red/SWIR ratio with spectral data, validated against an Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) image. These findings highlight the efficacy of integrating spectral and ancillary data with MLC for detailed mapping of glacier terrain features. This approach offers significant advancements for remote sensing applications, particularly in monitoring complex glacierized regions under changing climatic conditions.