Phenology-adapted potato mapping index (PMI) for ground sample-free identification

• Develop a novel potato mapping index (PMI) for robust potato identification. • PMI accurately identifies potatoes across diverse regions and sowing dates. • PMI enables efficient potato mapping without extensive training data. • PMI has high potential for scalability and global potato mapping. Potato is the world’s most important non-cereal food crop, crucial for global food security and agricultural sustainability. Large-scale potato mapping faces persistent challenges, primarily due to scarce training samples, complex background crop compositions, and limited spectral distinctiveness compared to staple crops. To overcome the persistent limitations in ground data dependency, we propose a novel Potato Mapping Index (PMI) that leverages Sentinel-2 temporal phenology to enable scalable potato identification without field sampling. The PMI operates dynamically based on crop phenological stages, utilizing Sentinel-2 bands (B8 (NIR) and B12 (SWIR-2)) and indices (normalized difference built-up index (NDBI) and red edge normalized difference vegetation index (RENDVI)) to accentuate spectral differences between potatoes and background crops. This approach effectively isolates the unique spectral signature of potato crops, facilitating accurate and efficient mapping. Validated across multiple years and in six key potato-producing countries (Belgium, China, Denmark, Germany, the Netherlands, and the United States), the PMI achieved an average overall accuracy (OA) of 92% and Kappa coefficients exceeding 0.8. These performance metrics are notably higher than those of the traditional supervised classifier, random forest (RF) model, which yielded an average OA of 90% and an average Kappa coefficient of 0.72. The resulting maps show high agreement with reference datasets, demonstrating robust spatiotemporal generalization. These findings offer a novel and valuable approach for global-scale potato mapping.