Quantifying Crop Phenology for Food Sustainability under Climate Change: Integration of Remote Sensing, Machine Learning, and Ecosystem Modeling

Climate change is projected to continue and accelerate significantly in the future if global greenhouse gas emissions are not curbed. Previous studies have shown that crop growth and production in agroecosystems are primarily determined by the weather conditions over the growing season. Crop phenology represents one of the most critical indicators in determining crop yield and adjusting the adaptation of crops to climate change. It provides essential information for monitoring and modeling crop growth dynamics and productivity. Therefore, understanding and quantifying the impacts of climate change on crop phenology and then agricultural production is crucial to formulate feasible climatic adaptation strategies. This work examined the crop phenology dynamics and their effects on crop production under changing climates using a synthesis approach that integrated remote sensing, machine learning, and ecosystem modeling. First, I applied an improved remote sensing-based method to detect historical crop phenology dynamics at the regional and continental scales over the past two decades. Then, I predicted future crop phenology dynamics in the United States under future climate scenarios by incorporating machine learning, remote sensing, and climate model projections. The spatiotemporal patterns of crop phenology and their correlations with climatic variables were further analyzed. Moreover, I used a process-based agroecosystem model (DLEM-Ag, Dynamic Land Ecosystem Model-Agriculture) to predict the effects of future phenology dynamics on crop productivity in the US corn cropping system. Finally, I explored how a warming climate would exacerbate crop phenological stages and influence agricultural production