ADVANCING CLIMATE RISK ASSESSMENT FOR AGRICULTURE AND WATER RESOURCES MANAGEMENT

Supriya  Savalkar

Global population growth and intensifying climate variability are expected to place unprecedented pressure on food and water systems. Meeting future demands will require protecting the three pillars of sustainable agriculture—crop yield, crop quality, and water availability—while enhancing productivity and developing more effective strategies for managing water resources. Climate-induced variability adds uncertainty, highlighting the need for robust frameworks that more comprehensively capture and address these risks. A major challenge lies in the fact that models guiding decision-making in agroecosystems and water resources carry multiple layers of uncertainty and bias, stemming from data inputs, assumptions, structural formulations, and parameterizations. This dissertation tackles these challenges by developing approaches to better quantify, reduce, and interpret uncertainty in agroecosystem and hydrologic models, with the overarching goal of strengthening climate risk assessments that underpin food and water security.The studies are set up at different spatial scales, where each case study addresses some aspects of uncertainty that challenge the food and water security of a region. The first component evaluates earlier planting as an adaptation strategy to mitigate crop exposure to extreme heat under climate change. Focusing on major U.S. spring wheat growing regions, I showed that while earlier planting can lower exposure to damaging temperatures during reproductive stages, it does not consistently replicate historical production conditions, and the planting window narrows considerably. These results challenge assumptions related to the effectiveness of earlier planting and emphasize the need to look at complementary adaptation strategies. The second component of this work investigates errors arising from the temporal disaggregation of daily meteorological data, an often overlooked but influential source of bias in agroecosystem modeling. Using high-resolution agricultural weather station networks, I quantified the impacts of disaggregation errors on two biophysical models and demonstrated their potential to distort risk estimates—most notably in sunburn risk assessments. Importantly, simple adjustments using observed statistics could reduce these biases substantially, highlighting the value of integrating insights from observational networks into model workflows. The third component centers on hydrologic modeling of snowmelt-dominated river basins, where accurate precipitation phase partitioning is essential for projecting water availability. Replacing the commonly used surface-temperature-based method with an approach that incorporates both surface temperature and relative humidity, I demonstrated that the change in snowmelt contribution to runoff is less drastic in the future than previously assumed. Hydrology model applications in the Columbia River basin with the new precipitation phase partitioning reveal higher historical snowmelt fractions and suggest that earlier assessments may have overstated climate-driven losses in snowpack and streamflow. These refinements carry significant implications for managing competing demands across irrigation, hydropower, environmental flows, and flood control. Collectively, this dissertation advances the understanding of uncertainty in agroecosystem models and provides practical pathways to reduce errors that shape risk assessments. By addressing uncertainties from data handling, adaptation evaluation, and model representation, this doctoral research improves the foundation for decision-making at the intersection of food, water, and energy security in a changing climate.

ADVANCING CLIMATE RISK ASSESSMENT FOR AGRICULTURE AND WATER RESOURCES MANAGEMENT

Files and links (1)

Abstract

Metrics

Details