Assessing the Predictability and Probability of 21st Century Rain-on-Snow Flood Risk for the Conterminous U.S.

The U.S. faces challenges and bears risk related to flood prediction and the protection of life, property and infrastructure. For much of the Nation, the timing of heavy rainfall can coincide with seasonal snow-cover. The combined rainfall and melt during so-called rain-on-snow events has historically contributed to some of the Nations most destructive and costly floods. Decision-makers lack guidance on ROS flood risk. Assessment requires accurate estimates of antecedent soil and snow conditions, rain-snow height levels, the energy exchange between the atmosphere and snow surface, and rainfall intensity. In a future climate, each of these variables and the integrated response that impacts flood risk is expected to change in complex and often contradictory ways. In this presentation, we present results from downscaling seven different CMIP5 models for historical and future 52-year-periods with both mid-century and end-century future horizons, each for RCP 4.5 and 8.5 emissions scenarios. We downscale GCM output to 12 km CONUS-wide resolution using the computationally efficient Intermediate Complexity Atmospheric Research (ICAR) model. We verify simulations against available observations and examine the frequency and intensity of rain-on-snow events, how they vary in the historical simulations, and the mechanisms by which they are projected to change across diverse future climate realizations. Results are presented by river basin and aggregated to National Climate Assessment regions. Highly variable regional and local changes, particularly in the East and Midwest, in runoff during rain-on-snow events are attributed to an increase in magnitude of extreme events despite an overall reduction in rain-on-snow frequency, particularly at the geographic margins of historical seasonal snow-cover. Changes in the West vary regionally and by elevation with increases in colder regions and decreases in warmer regions. The large suite of downscaled climate model output permits an assessment of compound changes communicated as a function of warming, time horizon, and/or cumulative carbon dioxide emissions.