The civil infrastructure of our country is by large built by stationary standards to address weather and climate related risks. However, with increasing effects of climate change being manifested, the project team is interested in re-assessing the risk of projects that are based on stationary intensity-duration-frequency curves to manage events such as extreme precipitation and flooding via runoff. The objectives of this project are to (1) develop protocols for incorporating non-stationarity into extreme value theory, which can be applied to flood frequency as well as extreme precipitation events; and incorporate ongoing and future projections of climate warming on flood frequency estimates for watersheds affected by (2) rain-on-snow events and (3) atmospheric rivers.

The technical approach includes two main elements to address the first objective and a third element to address the second objective. The first element incorporates non-stationarity as observed in historical records of extreme precipitation into the protocols for intensity-duration frequency relationships via the examination of the underlying probability distribution function. The second element involves the incorporation of regional climate models into flood risks; the project team analyzed the effects of atmospheric rivers on flooding/extreme precipitation in the western United States. The third element is specific to flood frequency estimation in watersheds affected by rain-on-snow events and snowmelt. The project team employs a hydrological model to track both the contributions of rain-on-snow events and snowmelt to runoff and assess the sensitivity of this relationship to increases in temperature.

The project team divided the results into three main sections, each of which address different aspects of flood risk with climate change via different hydrological processes: extreme precipitation, atmospheric rivers, rain-on-snow events. The project team found the tradeoff in the increasing variability versus decreasing bias from including a time-varying parameter in extreme precipitation distributions for both stationary and nonstationary environments. They conducted a study on the patterns of behavior of landfalling atmospheric rivers along three sections of mountains in the west coast. They separate the contribution of snowmelt versus rain-on-snow events on flooding events across the United States and discuss the differences.

The work incorporates time-dependent relationships into the best current extreme precipitation and flood frequency estimation methods. The results of this work benefit risk-based flood design not only at Department Of Defense and other governmental facilities, which will surely be impacted by changing flood risk in the face of climate warming, but also across the field of hydrologic engineering design.