Irrigation plays an important role in food production and the water cycle worldwide by enhancing agricultural yields, buffering climate variability, and appropriating 70% of total human freshwater use. Maintaining and even expanding irrigated areas is required to address increasing global food demand and climate-induced water stress. Over the latter half of the twentieth century, however, non-renewable groundwater use more than tripled to comprise ~1/5 of global irrigation water. As a result, key agricultural regions around the world are on unsustainable trajectories due to aquifer depletion. With limited water resources defining the 21st century, finding ways to maximize water use and operate within system boundaries is crucial. Crop and hydrology models can support decision making in the face of these challenges by simulating alternative management pathways under a range of resource conditions. In many cases, however, critical input datasets are missing or lack the precision and accuracy to fully parameterize landscape models. Recent rapid advances in large-scale satellite remote sensing can address these data gaps by quantifying landscape characteristics at previously infeasible spatial and temporal resolutions. In this dissertation, I present new methodologies that translate Landsat satellite observations into annual irrigation maps needed to understand and manage agricultural water resources. Maps are then analyzed and integrated into crop models to better understand historic water use, evaluate novel stakeholder-driven groundwater management, and support future planning. I focused on the High Plains Aquifer (HPA) in the central United States, where a $20 billion agricultural economy is threatened due to extensive depletion over much of the aquifer. In Chapter 1, I used Google Earth Engine and the full Landsat archive from 1999-2016 to generate annual, moderately high resolution (30 m) irrigation maps for the Republican River Basin portion of the HPA from 1999-2016. I found considerable interannual variability in irrigation location and extent, largely driven by annual precipitation, commodity prices, and increased irrigation efficiency over time. Chapter 2 extended this method to the full 450,000 km2 HPA from 1984-2017, addressing additional challenges from satellite data gaps and a wider range of climate, crop types, and management. I estimated that up to 24% of currently irrigated area could be lost by 2100 if aquifer depletion continues along recent trends.With increasing resource scarcity, a diverse set of groundwater management approaches have emerged across the HPA to slow depletion. In Chapter 3, I combined the satellite-derived irrigation maps, detailed well records, and national crop maps to assess the efficacy of innovative stakeholder-driven groundwater management in northwest Kansas referred to as the Local Enhanced Management Area (LEMA) program. I found that farmers surpassed targets for reduced water use without compromising irrigated area through adaptive cropping choices and increased irrigation efficiency. Chapter 4 extends the LEMA analysis with process-based crop models to robustly quantify impacts to the full water budget along with trade-offs in crop yield. Integrating remote sensing into this modeling framework allowed me to estimate quantities that are difficult or impossible to measure. As aquifer depletion threatens crop production in many parts of the world, approaches that integrate models with in-situ and remotely sensed data can improve understanding and help inform economically and hydrologically sustainable management strategies.
Read
