<![CDATA[Precise and detailed land mapping is essential for sustainable land management, environmental conservation, and regional planning, especially in complex and diverse landscapes. This study aims to present an innovative framework for the development of Land Mapping Units (LMUs) at a detailed scale (1:20,000), through the integration of Random Forest (RF) analysis and high-resolution remote sensing data. This study was conducted in the South Malang Plateau, Indonesia (the area characterized by karst, tectonic, volcanic, and alluvial landforms) from June to December 2024. As part of the methodology, the study utilized a combination of geospatial data, including geological maps, DEM-derived topographical indices, and remote sensing indices (Normalized Difference Soil Index/NDSI, Soil Adjusted Vegetation Index/SAVI, Normalized Difference Water Index/NDWI, Modified Soil Adjusted Vegetation Index/MSAVI). A total of 10,903 field observation points were analyzed, with 70% used for model training and 30% for validation. The results show that RF-based LMUs achieved R2 of 0.93 and Root Mean Square Error (RMSE) of 0.645, which is reliable to use. The LMUs provide a comprehensive understanding of landform-specific characteristics, including soil fertility linked to parent material, erosion sensitivity, and slope variability. These insights support applications in precision agriculture, disaster mitigation, and environmental planning. Moreover, the result can guide informed decision-making to prioritize sustainable land management that effectively prevents land degradation in the South Malang Plateau region, as stated in the Sustainable Development Goals (SDGs). The study demonstrates the potential of combining machine learning and remote sensing to refine spatial analysis and address the limitations of manual mapping methods. The proposed framework is scalable and adaptable to other diverse landscapes, making it a valuable tool for advancing sustainable land management in a rapidly changing world.]]>
