Climate-smart Agriculture: Potentials, Tradeoffs, and Cost-Benefit Analyses in US Corn-Soybean systems

Abstract

The UN Climate Change Conference 2023, or known as COP 28, highlighted with the first “global stocktake” of the world’s efforts to address climate change under the Paris Agreement, concluded that the current progress was too slow across all areas of climate action. It calls on governments to speed up actions by 2030, from reducing greenhouse gas emissions, to strengthening resilience to a changing climate, and to providing financial and technological support (UNFCCC, 2023). In 2022, the US government announced a five-year investment of billions of dollars in climate-smart agriculture through the Inflation Reduction Act (IRA), which has thrust regenerative agricultural practices, such as the adoption of cover crops (CC) and no-till (NT), into the spotlight. Yet there are significant knowledge gaps on the magnitude, distribution, and economic feasibility of those mitigation opportunities, and where and how to use public subsidies. Climate-smart agricultural practices mitigate climate change via two main paths, the sequestration of soil organic carbon (SOC) and/or the reduction of Nitrous oxide (N2O) emissions. Great efforts in the scientific community have been made towards measuring N2O emissions and SOC sequestrations and understanding their responses to different management practices. However, challenges remain in closing knowledge gap in understanding (1) determinative processes and underlying drivers of spatial-temporal heterogeneity in agricultural climate mitigation potentials, (2) the integrative impacts of conservation practice adoption on GHG emissions and crop production, and (3) how to identify efficient and actionable climate-smart practices in a scalable manner. In this dissertation, I aim to explore the responses of N2O emissions, SOC sequestration, and crop production to various management practices, deliver integrative frameworks of social-environmental evaluations through cost-benefit analysis, and develop efficient and trustable predictive tools for assessing climate-smart practices targeting the corn and soybean cropping systems in the US, which is a globally representative agricultural production area. The climate mitigation potentials associated with cover cropping, no-till, and different fertilizer application strategies will be assessed across field and regional scales. Multiple modeling and inference approaches including process-based model simulations and casual analysis, meta-analysis models, and knowledge-guided machine learning (KGML) models were utilized to tackle unanswered questions regarding the spatial and temporal drivers, potentials and pathways, and quantification methods towards climate change mitigation in the US agroecosystems. In the first study, I simulated and quantified variations of N2O emissions over space, time, and management practices in the Midwest by a process-based model, ecosys, and manifested the impacts of freeze-thaw cycles on N2O emissions in overlooked non-growing seasons using causal inference. In the second study, I built meta-random-forest models and a framework of cost-benefits analysis upon hundreds of paired field observations of N2O emissions, SOC change, and crop yield to visualize the spatial distribution of climate change mitigation potentials of adopting CC and NT. In the third study, I developed KGML models based on advanced machine learning algorithms for predicting crop yield and SOC change under various management scenarios to overcome the hassle and complexity in using process-based models, shedding light on the future of an efficient and trustable SOC simulator. Overall, results and findings presented in this dissertation deliver critical information on the climate change mitigation pathways in agriculture and approaches/frameworks for modeling and evaluating their environmental and economic effectiveness in a timely manner.