The relation between baroclinic adjustment and turbulent diffusion in the two-layer model

Baroclinic adjustment and turbulent diffusion are two popular paradigms used to describe the eddy-mean flow closure in the two-layer model, with very different implications for the criticality of the system. Baroclinic adjustment postulates the existence of preferred equilibrium states, while the turbulent diffusion framework predicts smooth variations of the mean state with the forcing. This study investigates the relevance of each paradigm over a wide range of the parameter space, including very strong changes in the diabatic forcing. The results confirm the weak sensitivity of the criticality against changes in the forcing noted by baroclinic adjustment studies but do not support the existence of preferred equilibrium states. The weak sensitivity of the mean state when the forcing is varied is consistent with the steepness of the diffusive closure predicted by homogeneous turbulence theories. These turbulent predictions have been tested locally against observed empirical diffusivities, extending a previous study by Pavan and Held. The results suggest that a local closure works well, even at low criticalities when the eddy momentum fluxes are important, provided that the criticality is generalized to include the effect of the meridional curvature potential vorticity (PV) gradient. When friction is weak, the development of this curvature may be important for halting the cascade and making the flow more linear. A remarkable difference from previous homogeneous results is that the empirical closure does not appear to steepen at low criticality. This may be due to the use of a generalized criticality or to the distinction between a local and domain-averaged closure.