Crop modeling for assessing and mitigating the impacts of extreme climatic events on the US agriculture system
The US agriculture system is the world’s largest producer of maize and soybean, and typically supplies more than one-third of their global trading. Nearly 90% of the US maize and soybean production is rainfed, thus is susceptible to climate change stressors such as heat waves and droughts. Process-based crop and cropping system models are important tools for climate change impact assessments and risk management. As data- science is becoming a new frontier for agriculture growth, the incoming decade calls for operational platforms that use hyper-local growth monitoring, high-resolution real-time weather and satellite data assimilation and cropping system modeling to help stakeholders predict crop yields and make decisions at various spatial scales. The fundamental question addressed by this dissertation is: How crop and cropping system models can be “useful” to the agriculture production, given the recent advent of cloud computing and earth observatory power? This dissertation consists of four main chapters. It starts with a study that reviews the algorithms of simulating heat and drought stress on maize in 16 major crop models, and evaluates algorithm performances by incorporating these algorithms into the Agricultural Production Systems sIMulator (APSIM) and running an ensemble of simulations at typical farms from the US Midwest. Results show that current parameterizations in most models favor the use of daylight temperature even though the algorithm was designed for using daily mean temperature. Different drought algorithms considerably differed in their patterns of water shortage over the growing season, but nonetheless predicted similar decreases in annual yield. In the next chapter of climate change assessment study, I quantify the current and future yield responses of US rainfed maize and soybean to climate extremes with and without considering the effect of elevated atmospheric CO2 concentrations, and for the first time characterizes spatial shifts in the relative importance of temperature, heat and drought stresses. Model simulations demonstrate that drought will continue to be the largest threat to rainfed maize and soybean production, yet shifts in the spatial pattern of dominant stressors are characterized by increases in the concurrent stress, indicating future adaptation strategies will have trade-offs between multiple objectives. Following this chapter, I presented a chapter that uses billion-scale simulations to identify the optimal combination of Genotype × Environment × Management for the purpose of minimizing the negative impact of climate extremes on the rainfed maize yield. Finally, I present a prototype of crop model and satellite imagery based within-field scale N sidedress prescription tool for the US rainfed maize system. As an early attempt to integrate advances in multiple areas for precision agriculture, this tool successfully captures the subfield variability of N dynamics and gives reasonable spatially explicit sidedress N recommendations. The prescription enhances zones with high yield potentials, while prevents over-fertilization at zones with low yield potentials.
Zhuang, Purdue University.
Geology|Geophysics|Climate Change|Agriculture|Atmospheric sciences
Off-Campus Purdue Users:
To access this dissertation, please log in to our