UC Davis

Adaptation to the multi-scale impacts of climate change in water resources systems is challenged by substantial uncertainty in future hydroclimatic projections, specifically regarding snowpack decline, flood and drought risks, and long-term transient trends in projections. This dissertation develops new approaches to address these challenges, specifically for the northern California water resources system. A new, daily timestep simulation model of the system, is introduced. Model simulation over an ensemble of climate scenarios provides a baseline system response for each chapter. In Chapter 1, a statistical analysis of these baseline simulations links several vulnerabilities directly to snowpack decline and the shift of snowmelt-fed streamflow earlier in the water year. To adapt seasonal reservoir management to snowpack decline in the region without the cost of additional infrastructure, the study proposes and tests adaptations that parameterize the structure of existing operating policies: a dynamic flood control rule curve, and revised snowpack-to-streamflow forecasting methods to improve seasonal runoff predictability given declining snowpack. These adaptations are shown to mitigate the majority of vulnerabilities caused by snowpack decline across the scenario ensemble. To address the issue of adapting to future flood and drought risk, Chapter 2 introduces a scenario selection framework. Scenarios are clustered by hydrologic properties, including full natural flow and snow water equivalent, along with a baseline-regret metric- the difference between the status quo and a perfect foresight adaptation. Findings suggest that reservoir operating policies should be trained to scenarios with a high baseline regret value in order to be most robust to climate uncertainty. Lastly, Chapter 3, develops framework for for dynamic climate adaptation based on multi-objective policy tree optimization, a heuristic search method that combines relevant indicators, actions, and thresholds in a flexible policy structure. Analysis of a robust set of policy trees identifies the most common indicators, actions, policy structure, and timing that produce robust policies. This represents a new and transferable problem framing for adaptation under uncertainty in which indicator variables, relevant actions, and policy structure are identified simultaneously.