Leveraging Internal Variability to Assess Regional Vulnerability to Atmospheric Rivers Using a High-Resolution Atmospheric Simulation

Atmospheric rivers striking the west coast of the United States lead to extreme precipitation and flooding. The vulnerability of existing assets in the region is not well understood because we have a relatively limited (~50+ year) record of flooding in the region, yet atmospheric rivers (ARs) may occur in a wider variety of sizes, orientations, sequences, and magnitudes than have been observed. Global climate models provide a larger library of conceivable atmospheric rivers spanning thousands of simulated years, but their spatial resolution prevents their use for regional system vulnerability assessments. Here we present work downscaling the global models using the Intermediate Complexity Atmospheric Research model (ICAR) on a 6 km grid. We show that simulated variability in ICAR provides a realistic representation of the local climate; that ICAR's precipitation response to integrated vapor transport (IVT) is highly correlated with observed responses, though it has a persistent low bias. We then explore the variability in AR driven precipitation from 5,000 years of downscaled simulation to identify regions for which the observational record may fail to capture the most intense precipitation that is likely to occur. These meteorological sequences are used to run a hydrologic model (SUMMA) to simulate the aggregated hydrologic response. After first order bias corrections are applied, the 5000 years of simulation yields a collection of AR events with more extreme precipitation than has been observed even in the "historical" period of the global models. This results in new sequences of extreme flooding that water managers at the US Army Corps of Engineers will use to assess the vulnerability of the regional system.