An Atmospheric-Hydrologic Modeling Framework to Evaluate the Sensitivity of the Probable Maximum Flood (PMF) in a Changing Climate

Abstract:

The Probable Maximum Flood (PMF) is a key design standard intended to ensure that dams and other high-risk structures can safely pass exceptionally large flood flows. Yet due to (1) climate non-stationarity, (2) evidence that the magnitude of extreme floods is controlled not only by heavy rainfall but also successive storms and snowmelt, and (3) a shift in the dam safety community’s thinking towards risk-based decision-making, a re-evaluation of existing PMF estimates and their associated uncertainty is needed. Using the WRF-Hydro model, we reconstruct three major storms (December 1964, January 1974 and January 1997) over the Feather, Russian and Santa Ana watersheds (California) under historic and projected climate conditions. We drive WRF-hydro with ERA5 reanalysis at a regional boundary with nested domains at 15 and 3 km horizontal resolution. To evaluate the sensitivity of the PMF to a wider set of flood drivers (individually and in combination) than has previously been considered, we examine multiple scenarios, including different storm temporal and spatial patterns, temperature sequences, and antecedent precipitation, soil moisture and snowpack. This approach is intended to improve our understanding of the uncertainty associated with and increase our confidence in PMF estimates, and ultimately help better assess the vulnerability of critical infrastructure to hydrometeorological extremes.