Abstract
Atmospheric rivers (ARs) are narrow filaments of high water vapor content that have considerable influence on the western United States (US) hydroclimate. ARs provide significant amounts of annual precipitation and snowfall and affect mountain snowpack via snow water equivalent (SWE) accumulation and ablation. With ARs projected to become increasingly key players in western US hydrology, water resource managers will rely progressively more on AR seasonal forecasts to infer flood/drought risks and make informed decisions about water supply allocation. However, precisely how well current seasonal climate prediction systems capture ARs and their associated hydrologic variables is still an open question. Here, we evaluate the ability of high (HR) and low resolution (LR) CCSM4 and CESM1 seasonal global retrospective forecasts to characterize precipitation, snowfall, and SWE changes associated with western US landfalling ARs. HR forecasts more accurately represent hydrologic variables than LR forecasts, however, CCSM4-HR underestimates AR-related snowfall, causing enhanced AR-related SWE ablation. Further investigation reveals amplified onshore positive temperature advection by south-southwesterly biased AR winds causes ARs in CCSM4-HR to be embedded within thicker air columns, yielding increased freezing level heights, reduced snowfall, and increased SWE loss. Results suggest both HR and LR global seasonal forecast models are capable of characterizing AR distribution and frequency, but HR models are needed for proper precipitation, snowfall, and SWE representation. Furthermore, models used to assess AR-related hydrological processes must contain accurate wind fields, as even minor biases can have a profound effect on a model's ability to simulate AR precipitation and SWE accumulation/ablation rates.