Two-dimensional turbulence and inverse cascades

For a velocity $(u(x,y),v(x,y),0)$ independent of the third coordinate, vorticity has only a perpendicular component $\omega=\partial_xv-\partial_yu$. The three-dimensional stretching term $(\boldsymbol\omega\cdot\nabla)\mathbf u$ vanishes because there is no variation along the vorticity direction. In the unforced inviscid problem, each fluid parcel therefore preserves its scalar vorticity. Advection can still create complicated spatial patterns and increasingly sharp vorticity gradients; zero stretching does not mean zero nonlinear evolution.

For a periodic domain, or appropriate boundaries that remove energy and vorticity fluxes, this material conservation also conserves enstrophy. The kinetic energy is independently conserved by the inviscid incompressible equation. Finite viscosity dissipates both quantities, but enstrophy weights smaller spatial scales much more strongly. Boundaries and forcing require explicit budget terms. A thin laboratory layer may approximate planar dynamics over part of its scale range, while still admitting three-dimensional motion at smaller scales or through interactions with its surroundings.

In Fourier space, the enstrophy carried by a planar incompressible mode is its kinetic energy multiplied by wavenumber magnitude squared. Thus moving energy to a higher wavenumber carries a disproportionately large enstrophy cost. An illustrative exchange among three wavenumber shells cannot simply move all energy from an intermediate shell to one neighboring shell while conserving both positive quadratic quantities. Some redistribution in both directions is needed for the particular transfer considered below. This is a constraint, not yet a complete theory of statistical flux.

In forced nearly ideal two-dimensional turbulence with forcing at an intermediate scale and sufficient separation, the standard dual-cascade picture has mean energy transfer toward larger scales and mean enstrophy transfer toward smaller scales. Large-scale damping or finite-domain effects are needed to control energy accumulating at the largest scales in a statistically steady experiment. The flux directions belong to this dynamical and statistical regime. A three-shell conservation exercise alone does not establish all transfer directions, locality, or a universal spectrum for any arbitrary initial condition.

An inverse energy cascade can form persistent large vortices or a domain-scale condensate, but a large coherent structure by itself does not prove a constant inverse flux. Inspect transfer budgets and the effects of forcing and friction. A forward enstrophy range can contain fine vorticity filaments even while most kinetic energy resides in large-scale motion. Energy and enstrophy therefore emphasize different aspects of the same field. Measurement resolution adequate for energy may be inadequate for vorticity gradients or enstrophy dissipation.

Oceanic and atmospheric flows can show approximate large-scale two-dimensional behavior because geometry, stratification, and rotation restrict aspects of motion. Their relevant conserved quantities may include potential vorticity and available potential energy, so the strictly planar model must not be substituted unchanged. Likewise, making a simulation plot two-dimensional does not eliminate a third velocity component or third-direction derivatives unless the equations explicitly do so. State which dimensions the dynamics actually retain before using a cascade argument to interpret a spectrum or a numerical result.