physics0506088/ms.tex
1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: 
4: \usepackage{graphicx}% Include figure files
5: \usepackage{dcolumn} % Align table columns on decimal point
6: \usepackage{bm}      % bold math
7: \usepackage{times}
8: \usepackage{color}
9: 
10: % MATH BOLDFACE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
11: \newcommand{\vB}{\mathbf{B}}
12: \newcommand{\vA}{\mathbf{A}}
13: \newcommand{\vF}{\mathbf{F}}
14: \newcommand{\vv}{\mathbf{v}}
15: \newcommand{\vu}{\mathbf{u}}
16: \newcommand{\vj}{\mathbf{j}}
17: \newcommand{\vx}{\mathbf{x}}
18: \newcommand{\vk}{\mathbf{k}}
19: \newcommand{\vomega}{\mbox{\boldmath $\omega$}}
20: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
21: 
22: % COLOR %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
23: \def\note#1{{\textcolor{green}{\bf [#1]}}} % note
24: \def\resp#1{{\textcolor{blue}{\bf [#1]}}}  % response
25: \def\add#1{{#1}}%{\textcolor{red}{#1}}}          % addition
26: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
27: 
28: %\topmargin -3pt
29: 
30: \begin{document}
31: 
32: \title{Cancellation exponent and multifractal structure in two-dimensional
33:        magnetohydrodynamics: direct numerical simulations and Lagrangian 
34:        averaged modeling}
35: 
36: \author{Jonathan Pietarila Graham, Pablo D. Mininni, and Annick Pouquet}
37: \affiliation{National Center for Atmospheric Research, P.O. Box 3000, 
38:              Boulder, Colorado 80307}
39: 
40: \date{\today}
41: 
42: \begin{abstract}
43: We present direct numerical simulations and Lagrangian averaged 
44: (also known as $\alpha$-model) simulations of forced and free 
45: decaying magnetohydrodynamic turbulence in two dimensions. 
46: The statistics of sign cancellations of the current at small 
47: scales is studied using both the cancellation exponent and the 
48: fractal dimension of the structures. The $\alpha$-model is 
49: found to have the same scaling behavior between positive and 
50: negative contributions as the direct numerical simulations. 
51: The $\alpha$ model is also able to reproduce the time 
52: evolution of these quantities in free decaying turbulence. At 
53: large Reynolds numbers, an independence of the cancellation 
54: exponent with the Reynolds numbers is observed.
55: \end{abstract}
56: 
57: \pacs{47.27.Eq; 47.27.Gs; 47.11.+j}
58: \maketitle
59: 
60: The magnetohydrodynamic (MHD) approximation is often used 
61: to model plasmas or conducting fluids in astrophysical and 
62: geophysical environments. However, given the huge amount of temporal 
63: and spatial scales involved in the dynamics of these objects, simulations 
64: are always carried out in a region of parameter space far from the observed 
65: values. Lagrangian averaged magnetohydrodynamics (LAMHD), also called 
66: the MHD alpha-model \cite{Holm02,Mininni05} (or 
67: the Camassa-Holm equations in early papers studying the 
68: hydrodynamic case \cite{LANS}), has been recently introduced as a way 
69: to reduce the number of degrees of freedom of the system, while 
70: keeping accurate evolution for the large scales. This approach (as 
71: well as large eddy simulations, or LES, for MHD; see e.g. 
72: \cite{Agullo01}) is intended to model astrophysical or geophysical 
73: flows at high Reynolds numbers using available computational resources. 
74: Several aspects of the MHD alpha-model have already been tested in two 
75: and three dimensions at moderate Reynolds numbers, against direct 
76: numerical simulations of the MHD equations \cite{Mininni05}. 
77: These studies were focused on comparisons of the evolution of global 
78: quantities and the dynamics of the large scale components of the 
79: energy spectrum \cite{Mininni05,Ponty05}.
80: 
81: All these models introduce changes in the small scales %dynamics 
82: in order to preserve the evolution of the large scales. %in the system. 
83: In several cases, it is of interest to know the statistics 
84: of the small scales. It is also important to model properly the 
85: small scales because they have an effect on large scales, as for 
86: example in the case of eddy noise: the beating of two small scales 
87: eddies produces energy at the large scale, and this may affect the 
88: global long-time evolution of the flow, an issue that arises in 
89: global climate evolution or in solar-terrestrial interactions. 
90: Moreover, plasmas and conducting fluids generate thin and intense 
91: current sheets where magnetic reconnection takes place. In these 
92: regions, the magnetic field and the current rapidly change sign, 
93: and after reconnection the magnetic energy is turned into mechanical 
94: and thermal energy. These events are known to take place in the 
95: magnetopause \cite{Sonnerup81}, the magnetotail \cite{Birn96}, the 
96: solar atmosphere \cite{Gosling95}, and the interplanetary medium 
97: \cite{Schmidt03}.
98: 
99: Current sheets are strongly localized and intermittent. To preserve 
100: reliable statistics of these events in models of MHD turbulence is 
101: of utmost importance to model some of these astrophysical and geophysical 
102: problems. In this work, we study whether the MHD alpha-model is able 
103: to reproduce the statistics and scaling observed in these phenomena.
104: 
105: In order to measure fast oscillations in sign of a field on arbitrary 
106: small scales, the cancellation exponent was introduced 
107: \cite{Ott92,Vainshtein94,Sorriso02}. The exponent is a measure 
108: of sign-singularity. We can define the signed measure for the current 
109: $j_z({\bf x})$ on a set $Q(L)$ of size $L$ as
110: \begin{equation}
111: \mu_i(l) = \int_{Q_i(l)}{d{\bf x} \, j_z({\bf x})} \, \bigg{/} 
112:     \int_{Q(L)}{d{\bf x}\, |j_z({\bf x})|}
113: \label{eq:mu}
114: \end{equation}
115: where $\{Q_i(l)\}\subset Q(L)$ is a hierarchy of disjoint subsets 
116: of size $l$ covering $Q(L)$.  The partition function $\chi$ measures 
117: the cancellations at a given lengthscale $l$,
118: \begin{equation}
119: \chi(l) = \sum_{Q_i(l)}|\mu_i(l)|.
120: \end{equation}
121: \add{Note that for noninteger $L/l$ the subsets will not cover $Q(L)$ and
122: finite size box effects must be considered in the normalization
123: of Eq. (\ref{eq:mu}).}
124: We can study the scaling behaviors of the cancellations 
125: defining the cancellation exponent $\kappa$, where
126: \begin{equation}
127: \chi(l) \sim l^{-\kappa} .
128: \label{eq:scaling}
129: \end{equation}
130: %Note that
131: Positive $\kappa$ indicates fast changes in sign on 
132: small scales (in practice, a cut-off is always present at the 
133: dissipation scale). A totally smooth field has $\kappa = 0$. This 
134: exponent can also be related with the fractal dimension $D$ of the %turbulent
135: structures \cite{Sorriso02},
136: \begin{equation}
137: \kappa = (d-D)/2 ,
138: \end{equation}
139: where $d$ is the number of spatial dimensions of the system. In some 
140: circumstances, we will also be interested on the cancellation exponent 
141: for the vorticity $\omega_z$. In that case the vorticity replaces the 
142: current in the definition of %the signed measure 
143: $\mu_i(l)$ [Eq. (\ref{eq:mu})]. 
144: 
145: Under special assumptions, relations between the cancellation exponent 
146: and scaling exponents have also been derived \cite{Vainshtein94}. 
147: Positive cancellation exponent $\kappa$ has been found in plasma 
148: experiments \cite{Ott92}, direct simulations of MHD turbulence 
149: \cite{Sorriso02}, in situ solar wind observations \cite{Carbone97}, 
150: and solar photospheric active regions \cite{Sorriso03}, where changes 
151: in the scaling were identified as preludes to flares.
152: 
153: In this work we will consider both free decaying and forced simulations 
154: of incompressible MHD and LAMHD turbulence in two dimensions (2D). 
155: The MHD equations in 2D can be written in terms of the stream function 
156: $\Psi$ and the $z$ component of the vector potential $A_z$,
157: \begin{eqnarray}
158: \partial_t \nabla^2 \Psi &=& [\Psi,\nabla^2\Psi] - [A_z,\nabla^2A_z] +
159:     \nu \nabla^4 \Psi \label{2DMHDmom} \\
160: \partial_t A_{z} &=& [\Psi,A_{z}] + \eta \nabla^2 A_z, \label{2DMHDind}
161: \end{eqnarray}
162: where the velocity and magnetic field are given by 
163: $\vv = \nabla \times (\Psi \hat{z})$ and 
164: $\vB = \nabla \times (A_z \hat{z})$ respectively, and 
165: $[F,G]= \partial_xF\partial_yG - \partial_xG\partial_yF$ is the 
166: standard Poisson bracket. The LAMHD equations are obtained by 
167: introducing a smoothing length $\alpha$, and the relation between 
168: smoothed (denoted by a subindex $s$) and unsmoothed fields is given 
169: by $\vF = (1-\alpha^2 \nabla^2) \vF_s$, for any field $\vF$. The system 
170: of LAMHD equations in this geometry \cite{Mininni05} is
171: \begin{eqnarray}
172: \partial_t \nabla^2 \Psi &=& [\Psi_s,\nabla^2\Psi] - [A_{s_z},\nabla^2A_z] +
173:     \nu \nabla^4 \Psi \label{2DLAMHDmom} \\
174: \partial_t A_{s_z} &=& [\Psi_s,A_{s_z}] + \eta \nabla^2 A_z. 
175:     \label{2DLAMHDind}
176: \end{eqnarray}
177: For both systems of equations, the current is given by 
178: $j_z = - \nabla^2 A_z$, and the vorticity by $\omega_z = - \nabla^2 \Psi$. 
179: \add{In these equations and in all the following figures, all quantities will
180: be given in familiar Alfv\'enic dimensionless units.}
181: Equations (\ref{2DMHDmom}-\ref{2DLAMHDind}) are solved in a periodic box 
182: using a pseudospectral code as described in \cite{Mininni05}. The code 
183: implements the 2/3-rule for dealiasing, and the maximum wavenumber 
184: resolved is $k_{max} = N/3$, where $N$ is the linear resolution used 
185: in the simulation. All the fields are written in dimensionless units.
186: 
187: To characterize the oscillating behavior and sign singularities in
188: the flows obtained from the MHD and LAMHD simulations, we perform a 
189: signed measure analysis and compute the cancellation exponent $\kappa$ 
190: for the current and for the vorticity. Following Eq. (\ref{eq:scaling}), 
191: its value is obtained by fitting $\chi(l) = c (l/L)^{-\kappa}$ through 
192: the inertial range, where $L=2\pi$ is the length of the box, and $c$ 
193: is a constant. The lengthscales in the inertial range used for this 
194: fit are obtained studying the scaling of the third order structure 
195: function \cite{Politano98}.
196: 
197: We first present results for a forced MHD simulation with $1024^2$ grid 
198: points, with $\eta = \nu = 1.6 \times 10^{-4}$. Both the momentum and
199: the vector potential equations were forced. The external forces had 
200: random phases in the Fourier ring between $k=1$ and $k=2$, and a 
201: correlation time of $\Delta t = 5 \times 10^{-2}$. The system was 
202: evolved in time until reaching a turbulent steady state. The amplitude 
203: of the magnetic force averaged over space was held constant to $0.2$, 
204: and the amplitude of the mechanical force to $0.45$, in order to have 
205: the system close to equipartition. %in the saturated state. 
206: Two more 
207: simulations using the LAMHD system were carried out, with the same parameters 
208: as the MHD run but with resolutions of $512^2$ grid points %and 
209: ($\alpha \approx 0.0117$), and $256^2$ grid points %and 
210: ($\alpha \approx 0.0234$) respectively \add{(the choice $\alpha = 2/k_{max}$ is conventional \cite{Mininni05,LANS})}.
211: The Kolmogorov's kinetic and 
212: magnetic dissipation wavenumbers in the MHD run are 
213: $k_\nu \approx k_\eta \approx 332$; %and
214: in all the LAMHD simulations 
215: these wavenumbers are larger than the largest resolved wavenumber 
216: $k_{max}$, by virtue of the model. Note that although it is common 
217: to reduce the spatial resolution even more in studies of the large 
218: scale components of the energy spectrum in LES of hydrodynamic 
219: turbulence, this cannot be done in this context since \add{wide energy spectra
220: and large amounts of spatial statistics}
221: are needed to properly compute the cancellation exponent 
222: (see e.g. \cite{Cerutti98} for a study of intermittency in LES).
223: 
224: Fig. \ref{fig:forced} shows $\chi(l)$ for the three simulations, 
225: averaged using 11 snapshots of the current covering a total time 
226: span of 20 turnover times in the turbulent steady state. A power 
227: law can be identified at intermediate scales, scales smaller than 
228: the forcing band but larger than the dissipation scale. Note that 
229: the two LAMHD simulations reproduce the same scaling as the MHD 
230: simulation. As a result, the sign singularity and fractal structure 
231: are both well captured in the inertial range although the alpha-model 
232: is known to give thicker structures at scales smaller than $\alpha$ 
233: due to the introduction of the smoothing length \cite{Chen99,Mininni05}. 
234: The best fit for the current $j_z$ using a power law in the inertial 
235: range gives \add{$\kappa = 0.50 \pm 0.17$ for the $1024^2$ MHD run, 
236: $\kappa = 0.55 \pm 0.19$ for the $512^2$ LAMHD simulation, 
237: and $\kappa = 0.55 \pm 0.43$ for the $256^2$ LAMHD simulation. Note 
238: that a value of $\kappa = 0.50$ in the MHD simulation gives a value 
239: of the fractal dimension $D = 1.00 \pm 0.34$, close to the codimension 
240: of 1 corresponding to current sheets in MHD turbulence. For the 
241: vorticity, the cancellation exponent is $\kappa = 0.73 \pm 0.16$ for \
242: the $1024^2$ MHD run, $\kappa = 0.74 \pm 0.32$ for the $512^2$ LAMHD 
243: simulation, and $\kappa = 0.80 \pm 0.32$ for the $256^2$ LAMHD simulation, 
244: giving a fractal dimension of $D = 0.54$} in the MHD simulation. The 
245: values obtained are compatible with the values of $\kappa = 0.43\pm0.06$ 
246: and $D = 1.14\pm0.12$ for the current, and $\kappa = 0.69\pm0.12$ 
247: and $D = 0.62\pm0.24$ for the vorticity obtained in Ref. \cite{Sorriso02} 
248: for forced direct numerical simulations of 2D MHD turbulence using a 
249: $1024^2$ spatial grid and $\eta = \nu = 8 \times 10^{-4}$. Given the 
250: good agreement between MHD and LAMHD simulations, in the following 
251: we will only refer to the cancellation exponent for the current 
252: density.
253: 
254: \begin {figure}
255: \includegraphics[width=8.5cm]{fig1.eps}
256: \caption {$\chi(l)$ averaged in time for $j_z$ in forced MHD turbulence. 
257:      The pluses correspond to the $1024^2$ MHD simulation, diamonds to 
258:      the $512^2$ LAMHD run, and triangles to the $256^2$ LAMHD run. 
259:      The dashed line indicates a slope of \add{$0.50$.  The arrows indicate
260: 	the inertial range.  Note that the slopes are of import, not the
261: 	offsets.}}
262: \label{fig:forced}
263: \end{figure}
264: 
265: Fig. \ref{fig:decay}.a shows the corresponding results for free decaying 
266: MHD turbulence. Three simulations are shown, one MHD run using 
267: $2048^2$ grid points, a $1024^2$ LAMHD run with $\alpha \approx 0.0058 $, 
268: and a $512^2$ LAMHD run with $\alpha \approx 0.0117$. The three 
269: simulations were started with the same initial conditions; initial 
270: velocity and magnetic fields with random phases between $k=1$ and 
271: $k=3$ in Fourier space, and unit {\it r.m.s.} values. The kinematic 
272: viscosity and magnetic diffusivity used were $\nu = \eta = 10^{-4}$. 
273: The three simulations were evolved in time without external forces.
274: 
275: \begin {figure}
276: \includegraphics[width=8.5cm]{fig2.eps}\\
277: \caption {(a) $\chi(l)$ at $t=4$ in the free decaying simulations, 
278:      pluses correspond to the $2048^2$ MHD simulation, diamonds to 
279:      the $1024^2$ LAMHD run, and triangles to the $512^2$ LAMHD run 
280:      (the dashed line indicates a slope of \add{$0.52$ and the arrows
281: 	indicate the inertial range}); 
282:      (b) time history of the cancellation exponent (thick lines) for 
283:      the three runs, and of $\eta \left< j_z^2 \right>$, where the 
284:      brackets denote spatial average.}
285: \label{fig:decay}
286: \end{figure}
287: 
288: The evolution of the cancellation exponent as a function of time 
289: in the free decaying simulations is shown in Fig. \ref{fig:decay}.b. 
290: For these simulations, the cancellation exponent is computed between 
291: the lengthscales \add{$L/l \approx 20$ and $L/l \approx 70$}, where 
292: a power law scaling in $\chi(l)$ can be clearly identified \add{from
293: $t=2.5$} up to $t=10$. At $t = 0$ the cancellation exponent $\kappa$ 
294: is zero, which corresponds to the smooth initial conditions. 
295: \add{A gap between $t=0$ and $t=2.5$ is present where no clear scaling is
296: observed.}
297: As time evolves, $\kappa$ grows up to $0.75$ at $t \approx 8$, as the 
298: system evolves from the initially smooth fields to a turbulent state 
299: with strong and localized current sheets. After this maximum, the 
300: exponent $\kappa$ decays slowly in time. The maximum of $\kappa$ takes 
301: place slightly later than the maximum of magnetic dissipation, as is 
302: also shown in Fig. \ref{fig:decay}.b. Note that the alpha-model also 
303: captures the time evolution of the cancellation exponent in free 
304: decaying turbulence, as well as the fractal structure of the problem 
305: as time evolves.
306: 
307: As previously noted in \cite{Mininni05}, the alpha-model slightly 
308: overestimates the magnetic dissipation. Note however that in the 
309: three simulations the peak of magnetic dissipation takes place close 
310: to $t \approx 6$, just before the peak of the cancellation exponent 
311: $\kappa$. From the maximum energy dissipation rate, the Kolmogorov's 
312: dissipation wavenumber for the kinetic and magnetic energy at $t \approx 6$ 
313: are estimated as $k_\nu \approx k_\eta \approx 470$, and this is again 
314: larger than the largest wavenumbers resolved in the two LAMHD simulations.
315: 
316: The observed slow decay of the cancellation exponent (compared 
317: with the square current) is related to the persistence of strong 
318: current sheets in the system for long times, even after the peak of 
319: magnetic dissipation. %is reached.
320: The system, instead of evolving fast 
321: to a smooth solution at every point in space, keeps dissipating energy 
322: in a few thin localized structures. The existence of these current 
323: sheets at late times can be more easily verified in simulations with 
324: smaller viscosity $\nu$ and diffusivity $\eta$. While in the peak of 
325: magnetic dissipation the system is permeated by a large number of small 
326: current sheets, at late times only a few current sheets are observed 
327: isolated by large regions where the fields are %approximately
328: smooth. 
329: 
330: \begin {figure}
331: \includegraphics[width=8.5cm]{fig3.eps}
332: \caption {Time history of $\kappa$ (solid line) and 
333:     $\eta \left< j_z^2 \right>$ (dotted line), for a free decaying 
334:     LAMHD simulation with $\eta=\nu=2 \times 10^{-5}$.}
335: \label{fig:reynl}
336: \end{figure}
337: 
338: Given the good agreement between direct numerical simulations (DNS) 
339: and LAMHD as seen in the preceding figure, we can reliably explore 
340: with the model Reynolds numbers unattainable in a reasonable time with 
341: DNS. In this context, we show that the maximum values of $\kappa$ 
342: obtained in the simulations seem to be insensitive to the Reynolds 
343: numbers within a given method (MHD or LAMHD) once a turbulent state 
344: is reached. As an example, in Fig. \ref{fig:reynl} we give the time 
345: history of the cancellation exponent and the square current for a 
346: free decaying LAMHD simulation with $\eta=\nu=2 \times 10^{-5}$ up 
347: to $t=20$. The initial conditions are the same as in the previously 
348: discussed simulations, and $\alpha \approx 0.0033$. It is worth noting 
349: that the time evolution of the magnetic dissipation in both decaying 
350: runs (Figs. \ref{fig:decay}.b and \ref{fig:reynl}) confirm previous 
351: results at lower Reynolds numbers \cite{Biskamp89,Politano89}: namely 
352: that the peak dissipation ($t \sim 7$) is lower for higher Reynolds 
353: numbers, while for later times it is quite independent of the Reynolds 
354: values.
355: 
356: Fig. \ref{fig:reync} shows $\chi(l)$ for early and late 
357: times in the same simulation. At small scales, the slope of 
358: $\chi$ always goes to zero, as can be expected since close to the 
359: dissipation lengthscale the fields are expected to be smooth. 
360: However, note that as time evolves the scaling of $\chi$ with 
361: $l$ drifts to smaller scales, and at $t=20$ a %clear
362: scaling can 
363: be observed up to $l/L \approx 0.005$.  \add{By virtue of the model the scaling
364: is wider and the slope goes to zero faster than in
365: the DNS due to the larger Reynolds number.}
366: 
367: \begin {figure}
368: \includegraphics[width=8.5cm]{fig4.eps}
369: \caption {$\chi(l)$ at \add{$t=3$} (dots), and $t=20$ (pluses), for the 
370:     free decaying LAMHD simulation with $\eta=\nu=2 \times 10^{-5}$. 
371:     The arrows indicate the inertial range.}
372: \label{fig:reync}
373: \end{figure}
374: 
375: The statistics of sign cancellation in magnetofluid turbulence are 
376: related with intermittency and anomalous scaling of structure 
377: functions \cite{Vainshtein94}, inherently associated with the 
378: dynamics of the small scales, and as a result harder to model 
379: in truncations or closures of the MHD equations. For example, 
380: two-point closures of turbulence behave smoothly (they also have 
381: no information about physical structures since they deal only with 
382: energy spectra). The intermittency of LES is an open topic, in 
383: particular because for neutral fluids the need to study the 
384: three dimensional case implied until recently that only low-resolution 
385: studies could be accomplished for which intermittent structures were 
386: barely resolved (see \cite{Kang03} for a recent study). From that 
387: point of view, the present study in two dimensions allows for higher 
388: Reynolds number studies. In MHD turbulence, the energy cascade being to 
389: small scales both in two and three space dimensions, it is hoped that the 
390: information gained here will carry on to the three-dimensional case. 
391: The new result stemming from this study is that the LAMHD alpha-model, 
392: although it alters the small scales through filtering, it nevertheless 
393: preserves some statistical information concerning the small scales. 
394: It is able to reproduce the scaling observed in forced MHD turbulence, 
395: as well as the time evolution of the cancellation exponent in free 
396: decaying simulations and as such, it represents a valuable model 
397: for studies of MHD flows for example at low magnetic Prandtl number 
398: $\nu/\eta$ as encountered in the liquid core of the Earth or in the 
399: solar convection zone (see e.g., \cite{Ponty05}).
400: 
401: \begin{acknowledgments}
402: Computer time provided by NCAR. NSF grant CMG-0327888 
403: at NCAR is gratefully acknowledged.
404: \end{acknowledgments}
405: 
406: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
407: \begin{thebibliography}{1}
408: 
409: \bibitem{Holm02}
410: D.D. Holm, Phys. D {\bf 170}, 253 (2002);
411: D.D. Holm, Chaos {\bf 12}, 518 (2002).
412: 
413: \bibitem{Mininni05}
414: P.D. Mininni, D.C. Montgomery, and A. Pouquet, Phys. Fluids {\bf 17}, 
415: 035112 (2005);
416: P.D. Mininni, D.C. Montgomery, and A. Pouquet, Phys. Rev. E {\bf 71}, 
417: 046304 (2005).
418: 
419: \bibitem{LANS}
420: D.D. Holm, J.E. Marsden and T.S. Ratiu, Adv. in Math. {\bf 137}, 1 (1998); 
421: S.Y. Chen, D.D. Holm, C. Foias, E.J. Olson, E.S. Titi, and S. Wynne, 
422: Phys. Rev. Lett. {\bf 81}, 5338 (1988); S.Y. Chen, C. Foias, D.D. Holm, 
423: E. Olson, E.S. Titi, and S. Wynne, Physica D {\bf 133} 49 (1999); 
424: S.Y. Chen, C. Foias, D.D. Holm, E.J. Olson, E.S. Titi, and S. Wynne, 
425: Phys.  Fluids {\bf 11}, 2343 (1999).
426: 
427: \bibitem{Agullo01}
428: O. Agullo, W.-C. M\"uller, B. Knaepen, and D. Carati, Phys. Plasmas 
429: {\bf 8}, 3502 (2001).
430: 
431: \bibitem{Ponty05}
432: Y. Ponty, P.D. Mininni, D.C. Montgomery, J.-F. Pinton, H. Politano, 
433: and A. Pouquet, Phys. Rev. Lett. {\bf 94}, 164502 (2005).
434: 
435: \bibitem{Sonnerup81}
436: B.U.O. Sonnerup, G. Paschmann, I. Papamastorakis, N. Sckopke, G. 
437: Haerendel, S.J. Bame, J.R. Asbridge, J.T. Gosling, and C.T. Russell, 
438: J. Geophys. Res. {\bf 86}, 10049 (1981).
439: 
440: \bibitem{Birn96}
441: J. Birn and M. Hesse, J. Geophys. Res. {\bf 101}, 15345 (1996).
442: 
443: \bibitem{Gosling95}
444: J.T. Gosling, J. Birn, M. Hesse, Geophys. Res. Lett. {\bf 22}, 869 (1995).
445: 
446: \bibitem{Schmidt03}
447: J.M. Schmidt, and P.J. Cargill, J. Geophys. Res. {\bf 108}, 1023 (2003).
448: 
449: \bibitem{Ott92}
450: E. Ott, Y. Du, K.R. Sreenivasan, A. Juneja, and A.K. Suri, 
451: Phys. Rev. Lett. {\bf 69}, 2654 (1992).
452: 
453: \bibitem{Vainshtein94}
454: S.I. Vainshtein, K.R. Sreenivasan, R.T. Pierrehumbert, V. Kashyap, and 
455: A. Juneja, Phys. Rev. E {\bf 50}, 1823 (1994).
456: 
457: \bibitem{Sorriso02}
458: L. Sorriso-Valvo, V. Carbone, A. Noullez, H. Politano, A. Pouquet, and 
459: P. Veltri, Phys. Plasmas {\bf 9}, 89 (2002).
460: 
461: \bibitem{Carbone97}
462: V. Carbone and R. Bruno, Astroph. J. {\bf 488}, 482 (1997).
463: 
464: \bibitem{Sorriso03}
465: L. Sorriso-Valvo, V. Carbone, V. Abramenko, V. Yurchysshyn,
466: A. Noullez, H. Politano, A. Pouquet, and P.L. Veltri, Planet. 
467: Space Sc. {\bf 52}, 937 (2004).
468: 
469: \bibitem{Politano98}
470: H. Politano and A. Pouquet, Phys. Rev. E {\bf 57}, R21 (1998); 
471: H. Politano and A. Pouquet, Geophys. Res. Lett. {\bf 25} 273 (1998).
472: 
473: \bibitem{Cerutti98}
474: S. Cerutti and C. Meneveau, Phys. Fluids {\bf 10}, 928 (1998).
475: 
476: \bibitem{Chen99}
477: S.Y. Chen, D.D. Holm, L.G. Margolin, and R. Zhang, Phys. D {\bf 133}, 
478: 66 (1999).
479: 
480: \bibitem{Biskamp89}
481: D. Biskamp and H.Welter, Phys. Fluids B {\bf 1}, 1964 (1989).
482: 
483: \bibitem{Politano89}
484: H. Politano, A. Pouquet, and P.L. Sulem, Phys. Fluids B {\bf 1}, 2330 (1989).
485: 
486: \bibitem{Kang03}
487: H.S. Kang, S. Chester, and C. Meneveau, J. Fluid. Mech. {\bf 480}, 
488: 129 (2003).
489: 
490: \end{thebibliography}
491: 
492: \end{document}
493: 
494: % LocalWords:  magnetohydrodynamic MHD plasmas LAMHD Camassa Holm magnetopause
495: % LocalWords:  magnetotail lengthscale noninteger Eq photospheric Alfv enic eps
496: % LocalWords:  pseudospectral dealiasing equipartition intermittency DNS NCAR
497: % LocalWords:  codimension diffusivity magnetofluid Prandtl CMG
498: