A large ensemble of 1/4° ocean/sea-ice hindcasts to characterize the stochastic character of the ocean variability at interannual-to-decadal timescales.

Abstract:

In the perspective of the future CMIP exercise, turbulent Ocean Global Circulation Models (at about 1/4° resolution) are progressively replacing laminar OGCMs (2° to 1°) in ocean-atmosphere coupled systems. Unlike in laminar models, a substantial intrinsic and chaotic variability spontaneously emerges in turbulent OGCMs under climatological forcing, cascading up to multidecadal timescales.
To which extent may this chaotic low-frequency intrinsic variability (LFIV) be modified or paced by an interannually-varying atmosphere ? The objective of the OCCIPUT project is to perform an ensemble of 50 global 1/4˚ ocean hindcasts (1958-2014) driven by the same reanalyzed atmospheric forcing, after perturbation of their initial conditions. This ensemble provides the first opportunity to extract and study both the intrinsic and the forced components of the oceanic variability, and to measure the actual constraint exerted by the atmosphere on this variability.
We have analyzed a smaller regional ensemble hindcast (10 members, 20 years) run over the North Atlantic as a "pilot experiment" and we will illustrate the imprint of the LFIV, and compare its amplitude with the atmospherically-forced variability, on various ocean quantities relevant for the climate system: Sea-Surface Height (SSH) and Temperature (SST), Heat-Content and Atlantic Meridional Overturning Circulation (AMOC). We find that:
- in the turbulent areas like the Gulf Stream, the intrinsic part of the variability on interannual timescales dominates the variability driven by the atmosphere,
- for a basin-wide integrated quantity like the AMOC, at 26.5N, the intrinsic variability is about 30 % of the magnitude of the forced variability on interannual timescales.
Some preliminary results will also be presented from the first years of the global 50-member ensemble which is currently being performed.