CMIP6-informed Flood Hazard and Uncertainty Assessment for Dam Safety Evaluation

Aging dams in the United States are vulnerable to the increasing magnitude and frequency of weather extremes under a changing climate. A significant increase in flood hazards compromises the flood regulation capacity of dams and poses a threat to the population, infrastructure, and national energy security. Ensuring the safety of dams and the nearby communities require a comprehensive assessment of future flood hazards. We implement a hierarchical modeling framework that integrates atmospheric, hydrologic, and statistical models for selected dams across the state of Michigan. Six sets of global climate models from the latest Climate Model Intercomparison Project Phase 6 (CMIP6) are downscaled by both dynamical and statistical approaches and used to simulate future runoff response by two calibrated hydrologic models, Variable Infiltration Capacity (VIC) and Precipitation-Runoff Modeling System (PRMS). The simulated runoff is then routed through a high-resolution river network routing model, RAPID, to produce ensemble streamflow projections to support risk assessment. Each ensemble member covers 1980–2019 in the historic period and 2020–2059 in the near-term future period under the high-end (SSP585) emission scenario. We extract annual maxima from ensemble streamflow projections, derive statistical distributions by three widely used probability density functions, Generalized Extreme Value (GEV), Log-Pearson III (LP-III), and Log-Normal (LN), and produce flood frequency estimates (e.g., 100- and 500-year return period floods) at dam locations for both periods. The change in future flood hazards and the main source of uncertainty are then evaluated. The results from the study can benefit the risk assessment needs and support overall decision-making by multiple stakeholders. The framework and tools developed are generic and scalable for assessing future flood hazards for other dams in the United States.