<![CDATA[Global potato production reached approximately 383 million metric tons in 2025, with Indonesia contributing around 1.22 million metric tons (0.32% of global output). However, the sustainability of Indonesia’s potato production is increasingly threatened by soil quality degradation in key growing regions. Existing predictive studies have largely focused on soil physical properties, with limited incorporation of remote sensing technologies. This study investigates the potential of Unmanned Aerial Vehicle (UAV) as a high-resolution, non-invasive tool to estimate potato yield through vegetation index transformations. Utilizing a split-plot experimental design across elevation gradients, we integrated soil physical properties with UAV-derived vegetation indices—Visible Atmospherically Resistant Index (VARI), Green Leaf Index (GLI), and Normalized Green-Red Difference Index (NGRDI). Results reveal that Total Nitrogen (TN), Base Saturation, and Bulk Density significantly influence yield variability, and can be accurately estimated using NGRDI, GLI, and a modified GLI (GLI CS), respectively. A multiple linear regression model was developed to predict potato yield: y = 24.22 + 7.26(NGRDI) + 9.87(GLI) + 28.42(GLI CS). This research demonstrates the efficacy of UAV-based spectral analysis in enhancing yield prediction models, offering a scalable and precise approach for sustainable potato cultivation. Future work should incorporate machine learning to improve model robustness and assess applicability across varied agro-ecological contexts.]]>
