Turbulence homogène de l'océan pour les simulateurs climatiques

The original CONTaCTS proposal described plans for developing parameterizations for the ocean components of the next generation of climate simulators. Ocean variability falls into two categories; the predictable component that responds in a deterministic way to the external ‘forcing’, and the unpredictable, or chaotic, turbulent, component that arises due to instabilities inherent in the predictable component and that feeds back on the latter. The question of parameterization involves capturing the essence of this feedback based on the mean flow characteristics, and including them as part of the mean flow forcing. One way we have progressed on this front is through the application of the so-called 'thickness-weighted averaging' (TWA) technique [7, 10, 16], which elegantly encapsulates eddy feedbacks on the mean in the momentum equations. Generalization of this technique was required for it to be applied to a realistic ocean simulation. Most importantly, we have generalized the buoyancy variable, key to the formulation of TWA, in order to accommodate a nonlinear, sea-water equation of state.  Beyond that, we analyzed an ensemble of runs, rather than rely on stationarity, which is almost certainly an invalid assumption. The results demonstrate that the leading existing eddy parameterization misses roughly half of that feedback [10]. A second approach focusses on quantifying eddy-feedbacks from a cross-scale perspective, as this is at the heart of eddy parameterization [4, 12, 17, 28]. We have been able to move beyond the classical assumptions of spatial homogeneity and temporal stationarity in several ways, including  by means of a wavelet analysis [17, 28]. We have found the spectral structure of the eddies exhibits a power law different from those predicted by theory.

In an exciting development, we have been able to propose an explanation for this spectral structure based on the richer dynamics of our governing equations relative to the reduced equations previously used to justify classical theoretical results. We have also been analyzing submesoscale resolving ensembles and demonstrated the existence of scale dependent nonlocality in eddy energy transfers.  This is essential to the development of so-called 'scale aware' eddy parameterizations [12].  In an interesting study of the Southern Ocean, we demonstrated regional distinctions in the eddy variability and identified unique characteristics of various dominant frequency bands [13].  We have contributed to the development of software tools based in the emerging cloud-based environment for large scale model analysis [14]. Last, we recently analyzed and criticized a widely adopted eddy transport parameterization [16]. 

Our MOPGA funding has also supported studies in small scale mixing [23], a novel form of near equatorial mixing [11], air-sea submesoscale interaction [15] and coupled model development [18].