Invariant parameterization of geostrophic eddies in the ocean

TL;DR

This paper extends invariant parameterization to higher-order schemes for geostrophic eddies, preserving symmetries of the barotropic vorticity model, and demonstrates improved turbulence modeling results.

Contribution

It introduces the first invariant higher-order closure schemes for geostrophic eddies that preserve model symmetries and shows their effectiveness in turbulence simulations.

Findings

Invariant schemes outperform standard models in turbulence decay.

First examples of invariant higher-order closure schemes.

Preservation of infinite-dimensional symmetry algebra.

Abstract

The framework of invariant parameterization is extended to higher-order closure schemes. We also define, for the first time, generalized invariant parameterization schemes, where symmetries of the corresponding original model are preserved as equivalence transformations of related classes of closed system of differential equations. As a particular problem, we consider invariant parameterization schemes for geostrophic eddies in a barotropic ocean. Here the initial model is the barotropic vorticity equation, which is equivalent to the system of incompressible inviscid two-dimensional Euler equations on a midlatitude beta-plane. The maximal Lie invariance algebra of this model is infinite-dimensional, and we intend to preserve it in the course of invariant parameterization, at least partially. The parameterizations proposed for the eddy vorticity flux and the energy flux are of order one and a half since we explicitly consider the equation for the turbulent kinetic energy in the closure models. These parameterizations are therefore the first examples of invariant higher-order closure schemes. Numerical experiments are carried out to assess the performance of these invariant schemes in studies of freely decaying two-dimensional turbulence, and it is verified that the invariant parameterization schemes give, on average, better results than the standard non-invariant closure models do.