[A-AS02] 高性能計算が拓く気象・気候・環境科学

[AAS02-02] Decadal predictability in a high-resolution eddy-resolving model: a signal-to-noise paradox perspective

★Invited Papers

*Wei Zhang^1,2、B. Kirtman、A. C. Clement、L. Siqueira、J. He、J. Xia、N. Perlin、J. Infan (1.University of Miami、2.National Oceanic and Atmospheric Administration (NOAA) )

Recent studies suggest the widespread existence of the signal-to-noise paradox in seasonal-to-decadalclimate predictions. The essence of the paradox is that the signal-to-noise ratio in models can beunrealistically small and models may make better predictions of the observations than they predictthemselves. Underestimated decadal predictability has been identified in current global climate models(e.g., IPCC-class models) and based on a multi-model assessment of CMIP5/6 models, we find that models tend to underestimate decadal predictability in regions where it is likely for the paradox to exist. These models fail to fully resolve mesoscale ocean features (with length scales on the order of 10km), such as the western boundary currents, potentially contributing to the signal-to-noise paradox andthus limiting climate predictability over decadal timescales. To test this hypothesis, we perform a suiteof CESM model experiments incorporating high-resolution eddy-resolving ocean (HR: 0.1°) in thatcapture these important mesoscale ocean features with increased fidelity. Compared with the eddy-parameterized ocean model (LR: 1°), the paradox is less likely to exist in HR, particularly over eddy-richregions. These also happen to be regions where increased decadal predictability is identified in HR. Weargue that this enhanced predictability is due to the enhanced vertical connectivity in the ocean. Thepresence of mesoscale ocean features and associated vertical connectivity significantly influencedecadal variability, predictability, and the signal-to-noise paradox.
Moreover, we detect a better representation of the air-sea interactions between SST and low-levelatmosphere over the Gulf Stream, thus improving low-frequency rainfall variations and extremes overthe Southeast US. The results further imply that high-resolution GCMs with increased ocean modelresolution may be needed in future climate prediction systems.