Mesoscale ocean turbulence is the best-known expression of Chaotic Intrinsic Variability (CIV), which spontaneously emerges from the unstable ocean circulation. A substantial amount of CIV is also found up to the scale of basins and decades, potentially driven by large-scale baroclinic instability or resulting from spatiotemporal inverse cascade processes.

A 56-year atmospherically-forced 50-member ensemble simulation of the global eddying ocean/sea-ice system has been performed to explore these phenomena in the context of the OCCIPUT research project, on the basis of the DRAKKAR 1/4° NEMO model configuration. This large ensemble experiment shows that monthly to multi-decadal CIV is largest in western boundary currents and south of about 30°S, with imprints over most of the globe; this low-frequency large-scale CIV competes with (and in certain zones exceeds) the atmospherically-forced ocean variability (AFV) in terms of strength.

This presentation summarizes recent studies of the low-frequency, large-scale impacts of AFV and CIV on the variability and trends of climate-relevant oceanic indices, such as heat content, volume and heat transports, sea level, EKE, or air-sea CO2 fluxes. We finally propose new ways to analyse this complex atmospherically-modulated oceanic “chaos” without separating CIV and AFV, by adopting concepts from dynamical system and information theories.

The partly random character of climate-relevant ocean fluctuations in the eddying regime questions the attribution to atmospheric drivers of observed signals, the turbulent ocean predictability and its potential influence in high-resolution coupled simulations.