<![CDATA[Tropical glaciers, such as those in Puncak Jaya, Papua, are among the most climate-sensitive ice masses on Earth, yet their small size, complex topography, and persistent cloud cover hinder accurate monitorin. Conventional threshold-based mapping methods, including the Normalized Difference Snow Index (NDSI), often misclassify debris-covered ice and bright bedrock, limiting their applicability in tropical mountain environments. This study develops and evaluates an integrated Linear Spectral Unmixing Analysis (LSUA)–Geographic Information System (GIS) methodology for high-fidelity mapping of glacier extent and surface composition in Puncak Jaya. Multispectral Landsat 8 OLI imagery was processed using LSUA to generate fractional abundance maps of clean ice, debris-covered ice, supraglacial water, and surrounding terrain. These outputs were integrated with Digital Elevation Models (DEMs) in a GIS framework for glacier area computation, elevation-based change detection, and spatial context analysis. Accuracy assessment using confusion matrices and Root Mean Square Error (RMSE) metrics against high-resolution reference imagery demonstrated that the LSUA–GIS workflow outperformed conventional NDSI mapping, particularly in detecting debris-covered ice, with an overall classification accuracy exceeding 90%. Results revealed continued glacier retreat, with the most significant ice loss occurring at elevations 4.884 MASL. The proposed workflow offers a reproducible and scalable approach for mapping small, fragmented tropical glaciers, providing critical data for climate impact assessment, hydrological planning, and long-term monitoring in remote mountain regions.]]>
