CoAuthors

Stéphanie Leroux (LGGE, France); Laurent Bessières (CERFACS, France); Guillaume Sérazin (LGGE-CERFACS, France); Pierre-Vincent Huot (LGGE, France); Alexandre Jaymond (LGGE, France); Sandy Grégorio (LGGE, France); Jean-Marc Molines (LGGE, France); Bernard Barnier (LGGE, France); Laurent Terray (CERFACS, France)

The CHAOCEAN OST-ST project has shown that in the eddying regime, the ocean circulation spontaneously generates a low-frequency chaotic intrinsic variability (LFCIV), which reaches basin scales and 1-100-year timescales, and strongly imprints climate-relevant variables like SLA, SST, Oceanic Heat Content (OHC), or AMOC. This LFCIV is almost zero in laminar ocean models used in most recent climate projections (CMIP5, AR5 IPCC report) but gives a marked stochastic flavor to the low-frequency variability in eddying ocean models, which are being coupled to the atmosphere for the CMIP6 exercise (and future IPCC climate projections). These results motivated the companion OCCIPUT project, which adopts a fully-probabilistic modelling approach: we performed the first-ever long (1960-2015), large (50-member) ensemble of global ocean/sea-ice 1/4° hindcasts driven by a full (reanalyzed) atmospheric forcing. Ensemble members were started from slightly different initial conditions and driven by the same forcing. Resulting ensemble statistics give access to the time-varying forced and chaotic oceanic variabilities, and to the time-varying atmospheric imprint on the ocean chaos over a wide range of spatiotemporal scales.
CHAOCEAN and OCCIPUT show that the oceanic variability should be seen as a broadband atmospherically-modulated “chaos”. Chaotic ocean dynamics shapes the SLA statistical structure (i.e. PDF) in very contrasting ways, and the atmospheric modulation of instantaneous PDFs (i.e. the probability that SLA takes a given value) varies in time and space. This blend of external and internal sources of oceanic variability raises new and important detection/attribution issues for satellite/in-situ observations: in key regions, the oceanic chaos can mask atmosphere- and climate-related SLA, AMOC, SST and OHC evolutions, even at large spatiotemporal scales. This chaos indeed can “contaminate” for instance [1] the observed 20-30 year SLA/OHC regional trends due to climate change, which are locally dominated by LFCIV; or [2] the interannual-to-decadal SLA/OHC/SST/AMOC regional variability that is partly random (uncorrelated with the forcing) in key areas of the global ocean.
In short, these probabilistic studies show that daily-to-decadal oceanic variability in this turbulent regime is influenced by the atmospheric variability but is substantially chaotic in key regions (which is new for large time and space scales). We will discuss potentially major consequences for the interpretation of observational oceanic data, and interesting perspectives for the analysis and use of SLA, satellite, and in-situ data.