A performance-based approach to quantify atmospheric river flood risk

Abstract

Atmospheric rivers (ARs) are a class of meteorologic phenomena that cause significant precipitation and flooding on the US West Coast. This work presents a new Performance-based Atmospheric River Risk Analysis (PARRA) framework that adapts existing concepts from probabilistic risk analysis and performance-based engineering for application in the context of AR-driven fluvial flooding. The PARRA framework is a chain of physically based models that link the atmospheric forcings, hydrologic impacts, and economic consequences of AR-driven fluvial flood risk together at consistent “pinch points”. Organizing around these pinch points makes the framework modular, meaning that models between pinch points can be updated without affecting the rest of the model chain, and it produces a probabilistic result that quantifies the uncertainty in the underlying system states. The PARRA framework can produce results beyond analyses of individual scenario events and can look toward prospective assessment of events or system changes that have not been seen in the historic record. The utility of the PARRA framework is demonstrated through a series of analyses in Sonoma County, CA, USA. Individual component models are fitted and validated against a historic catalog of AR events occurring from 1987 to 2019. Comparing simulated results from these component model implementations against observed historic ARs highlights what we can learn about the drivers of extremeness in different flood events by taking a probabilistic perspective. The component models are then run in sequence to generate a first-of-its-kind AR flood loss exceedance curve for Sonoma County. The prospective capabilities of the PARRA framework are presented through the evaluation of a hypothetical mitigation action. Elevating 200 homes, selected based on their proximity to the Russian River, was sufficient to reduce the average annual loss by half. Although expected benefits were minimal for the smallest events in the stochastic record, the larger, more damaging ARs were expected to see loss reductions of approximately USD 50–75 million per event. These results indicate the potential of the PARRA framework to examine other changes to flood hazard, exposure, and vulnerability at the community level.