Transition from weak to strong cascade in MHD turbulence.

The transition from weak to strong turbulence when passing from large to small scales in magnetohydrodynamic (MHD) turbulence with guide field is a cornerstone of anisotropic turbulence theory. We present the first check of this transition, using the Shell-RMHD, which combines a shell model of perpendicular nonlinear coupling and linear propagation along the guide field. This model allows us to reach Reynolds numbers around 10(6). We obtain surprisingly good agreement with the theoretical predictions, with a reduced perpendicular energy spectrum scaling as k(⊥)(-2) at large scales and as k(⊥)(-5/3) at small scales, where critical balance between nonlinear and propagation time is reached. However, even in the strong regime, a high level of excitation is found in the weak coupling region of Fourier space, which is due to the rich frequency spectrum of large eddies. A corollary is that the reduced parallel spectral slope is not a definite test of the spectral anisotropy, contrary to standard belief.

Scaling and anisotropy in magnetohydrodynamic turbulence in a strong mean magnetic field.

We present an analysis of the anisotropic spectral energy distribution in incompressible magnetohydrodynamic turbulence permeated by a strong mean magnetic field. The turbulent flow is generated by high-resolution pseudospectral direct numerical simulations with large-scale isotropic forcing. Examining the radial energy distribution for various angles θ with respect to B0 reveals a specific structure which remains hidden when not taking axial symmetry with respect to B0 into account. For each direction, starting at the forced large scales, the spectrum first exhibits an amplitude drop around a wave number k0 which marks the start of a scaling range and goes on up to a dissipative wave number k(d)(θ). The three-dimensional spectrum for k≥k0 is described by a single θ-independent functional form F(k/k(d)), with the scaling law being the same in every direction. The previous properties still hold when increasing the mean field from B0=5 up to B0=10b(rms), as well as when passing from resistive to ideal flows. We conjecture that at fixed B0 the direction-independent scaling regime is reached when increasing the Reynolds number above a threshold which raises with increasing B0. Below that threshold critically balanced turbulence is expected.

Spectral energy dynamics in magnetohydrodynamic turbulence.

Spectral direct numerical simulations of incompressible MHD turbulence at a resolution of up to 1024(3) collocation points are presented for a statistically isotropic system as well as for a setup with an imposed strong mean magnetic field. The spectra of residual energy, E(R)k=|E(M)k - E(K)k|, and total energy, Ek=E(K)k+E(M)k, are observed to scale self-similarly in the inertial range as E(R)k approximately k(-7/3), E(k)approximately k(-5/3) (isotropic case) and E(R)(k(perpendicular) approximately k(-2)(perpendicular), E(k(perpendicular))approximately k(-3/2)(perpendicular) (anisotropic case, perpendicular to the mean field direction). A model of dynamic equilibrium between kinetic and magnetic energy, based on the corresponding evolution equations of the eddy-damped quasinormal Markovian closure approximation, explains the findings. The assumed interplay of turbulent dynamo and Alfvén effect yields E(R)k approximately kE2(k), which is confirmed by the simulations.

Statistical anisotropy of magnetohydrodynamic turbulence.

Direct numerical simulations of decaying and forced magnetohydrodynamic (MHD) turbulence without and with mean magnetic field are analyzed by higher-order two-point statistics. The turbulence exhibits statistical anisotropy with respect to the direction of the local magnetic field even in the case of global isotropy. A mean magnetic field reduces the parallel-field dynamics while in the perpendicular direction a gradual transition towards two-dimensional MHD turbulence is observed with k(-3/2) inertial-range scaling of the perpendicular energy spectrum. An intermittency model based on the log-Poisson approach, zeta(p)=p/g(2)+1-(1/g)(p/g), is able to describe the observed structure function scalings.