Satellite-guided crop phenology modeling for North and South America

Crop phenology, representing the physiological development stages of crop growth, alters surface energy, water, and carbon budgets, thereby modulates land-atmosphere interactions. It is also crucial for estimating crop production and designing agricultural management, making it a key to food and water security. However, phenology varies across different crops and regions, and growth stage information is often sparse, limiting our understanding and modeling capabilities. Current process-based crop models typically determine phenology by accumulating growing degree days by using site-specific parameters, which limits its application to large-scale studies. Advances in remote sensing offer tools to bridge this data gap and enhance model performance.

In this study, we combined NASA MODIS reflectance data (NIRv) and USDA Crop Data Layer to create high-resolution, multi-decade crop phenology maps for four major crops (maize, soybean, wheat, and rice) in the US. Five phenology stages—emergence, vegetative, reproductive, mature, and harvest—were identified using a curve-based algorithm. Our estimates align well with USDA county-level crop progress reports, demonstrating the robustness of our method. Our approach also accurately captures interannual variability in crop phenology, such as late emergence dates in 2019 due to increased spring rainfall in the Corn Belt. We then used our product to guide a process-based crop model (Noah-MP crop), improving phenology parameterization of growing seasons. Extending our method to a global scale, we showcased its capability in regions lacking ground surveys, such as South America. These satellite-based phenology maps have significant potential for understanding crop responses to climate variability, enhancing model parameters, and fostering sustainable agricultural development, especially in data-scarce regions.