Reduced Southern Ocean Warming Enhances Global Skill and Signal-to-Noise in an Eddy-Resolving Decadal Prediction System
Steve Yeager, UCAR
11:00 am – 12:00 pm MDT
Seasonal-to-decadal prediction research has historically focused on advancing understanding and model representation of predictability mechanisms at work in the Atlantic and Pacific sectors. Recent work with CESM prediction ensembles has revealed promising potential to leverage initial condition memory in the Pacific for multiyear forecasts and in the Atlantic for decadal forecasts, while also exposing coupled model fidelity issues that would otherwise go unnoticed. New CESM high-resolution prediction ensembles are yielding important insights into predictability mechanisms and model behavior that will likely have broad impacts and influence future coupled climate modeling and prediction research priorities. The impact of increased model horizontal resolution on climate prediction performance is examined by comparing results from low-resolution (LR) and high-resolution (HR) decadal prediction simulations. There is general improvement in global skill and signal-to-noise characteristics, with particularly noteworthy improvements in the eastern tropical Pacific, when resolution is increased from order 1° in all components to order 0.1°/0.25° in the ocean/atmosphere. A key advance in the ocean eddy-resolving HR system is the reduction of unrealistic warming in the Southern Ocean (SO) which we hypothesize has global ramifications through its impacts on tropical Pacific multidecadal variability. The results suggest that accurate representation of SO processes is critical for improving decadal climate predictions globally and for addressing longstanding issues with coupled climate model simulations of recent Earth system change.