Samuel Daniel Lup Haruta, Pablo López and Caleb Gustavo De Bernardis
NAX Solutions, C. Granja de Rocamora, 6, 03015 Alicante, Spain; slup@naxsolutions.com
Efficient harvest timing is critical for optimizing sugarcane yield and resource use. This study integrates multispectral and Synthetic Aperture Radar (SAR) satellite imagery with advanced machine-learning algorithms to develop a scalable model for real-time monitoring and forecasting of sugarcane ripening. By tracking the temporal evolution of physiological phenomena measured through neural networks over satellite imaging, the NAX ripening model enables precise prediction of optimal harvest windows. The model was validated through laboratory analyses of field data collected in Argentina, Colombia, Mexico, and Guatemala. Strong correlations between satellite-derived metrics and in-field measurements of stalk water content and sucrose content underscore the robustness of this approach. Condition-specific ripening models enhance the utility of this methodology, enabling variable-rate application of ripening agents, real-time harvest planning adjustments, and significant cost reductions in laboratory expenses for harvest planning. This satellite and data driven system provides a transformative alternative to conventional laboratory-based assessments, reducing chemical agent use and its environmental impact. The results demonstrate the model’s scalability and applicability across diverse geographic and climatic conditions, offering a solution for the sugarcane industry’s evolving needs. This work paves the way for sustainable, precision agriculture practices, revolutionizing the approach to ripening management and harvest optimization.