astro-ph0701872/ms.tex
1: \documentclass{emulateapj}
2: \usepackage{latexsym}
3: \usepackage{mathrsfs}
4: \usepackage{amsfonts}
5: \usepackage{amssymb}
6: \usepackage{amsmath}
7: \usepackage{natbib}
8: \usepackage{subfig}
9: 
10: 
11: \newcommand{\be}{\begin{equation}}
12: \newcommand{\ee}{\end{equation}}
13: \newcommand{\bes}{\begin{equation*}}
14: \newcommand{\ees}{\end{equation*}}
15: \newcommand{\bsy}{\boldsymbol}
16: 
17: \shorttitle{Coronal Heating}
18: \shortauthors{Rappazzo et al.}
19: 
20: \begin{document}
21: \title{Coronal Heating, Weak MHD Turbulence and Scaling Laws}
22: \author{A. F. Rappazzo \altaffilmark{1,2},
23: M. Velli \altaffilmark{2,3},
24: G.~Einaudi \altaffilmark{1} and
25: R.~B. Dahlburg \altaffilmark{4} 
26: }
27: 
28: \altaffiltext{1}{Dipartimento di Fisica ``E.~Fermi'', Universit\`a di Pisa,
29:                         56127 Pisa, Italy; rappazzo@jpl.nasa.gov}
30: \altaffiltext{2}{Jet Propulsion Laboratory, Pasadena, CA,
31:                         91109}
32: \altaffiltext{3}{Dipartimento di Astronomia e Scienza dello Spazio, 
33:                         Universit\`a di Firenze, 50125 Firenze, Italy}
34: \altaffiltext{4}{LCP\&FD, Naval Research Laboratory, 
35:                         Washington, DC 20375}
36:        
37: \begin{abstract}
38: Long-time high-resolution simulations of the dynamics of a coronal loop in 
39: cartesian geometry are carried out, within the framework of reduced 
40: magnetohydrodynamics (RMHD), to understand coronal heating driven by motion of 
41: field lines anchored in the photosphere. 
42: We unambiguously identify  MHD anisotropic turbulence as the physical mechanism
43: responsible for the transport of energy from the large scales, where energy is injected
44: by photospheric motions,  to the small scales, where it is dissipated.
45: As the loop parameters vary  different regimes of turbulence develop: 
46: strong turbulence is found for weak axial magnetic fields and long loops,
47: leading to Kolmogorov-like spectra in the perpendicular direction, while 
48: weaker and weaker regimes (steeper spectral slopes of total energy) are 
49: found for strong axial magnetic fields and short loops. 
50: As a consequence we predict that the  scaling of  the heating rate with 
51: axial magnetic field intensity $B_0$,
52: which depends on the spectral index of total energy for given loop 
53: parameters, must vary from $B_0^{3/2}$ for weak fields to 
54: $B_0^{2}$ for 
55: strong fields at a given aspect ratio. The predicted heating rate is within the 
56: lower range of observed active region and quiet Sun coronal energy losses.
57: \end{abstract}
58: \keywords{Sun: corona --- Sun: magnetic fields --- turbulence}
59: 
60: \section{INTRODUCTION}
61: 
62: In this letter we solve, within the framework of RMHD in cartesian 
63: geometry, the Parker field-line tangling (coronal heating) 
64: problem \citep{park72,park88}.
65: We do this via long simulations at high resolutions, introducing 
66: hyper-resistivity models to attain extremely large Reynolds 
67: numbers. We show how small scales form and how the coronal 
68: heating rate depends on the loop and photospheric 
69: driving parameters, and derive simple formulae which may be used in 
70: the coronal heating context for other stars. 
71: 
72: Over the years a number of numerical experiments have been carried out to 
73: investigate coronal heating, with particular emphasis on exploring  how photospheric 
74: field line tangling leads to current sheet formation.
75: 
76: \citet{mik89} and \citet{hen96} first carried out simulations 
77: of a loop driven by photospheric motions using a cartesian approximation (a 
78: straightened out loop bounded at each end by the photosphere) imposing a 
79: time-dependent alternate direction flow pattern at the boundaries.  
80: A complex coronal magnetic field results  from the photospheric field line random walk, and though the  field  does not, strictly speaking, evolve through a sequence of static force-free 
81: equilibrium states (the original Parker hypothesis), magnetic energy nonetheless
82: tends to dominate kinetic energy in the system. 
83: In this limit the field is structured by current sheets elongated along the axial 
84: direction,  separating quasi-2D flux tubes which constantly move around and interact.  
85: \citet{long94}  focused on the current sheet formation process within 
86: the RMHD approximation, also used in the simulations by
87: \citet{dmi99}. The results from these studies agreed qualitatively among 
88: themselves, in that all simulations display the development of field aligned 
89: current sheets. However, estimates of the dissipated power and its scaling 
90: characteristics differed largely, depending on the way in which
91: extrapolations from low to large values of the plasma conductivity 
92: of the properties such as  inertial range power law 
93: indices were carried out.
94: 2D numerical simulations of incompressible MHD with magnetic forcing 
95: \citep{ein96,georg98,dmi98,ein99} showed that turbulent current sheets 
96: dissipation is distributed intermittently, and that the
97: statistics of dissipation events, in terms of total energy, peak energy and event duration
98: displays power laws not unlike
99: the distribution of observed emission events in optical, ultraviolet and x-ray 
100: wavelengths of the quiet solar corona.
101: 
102: More recently full 3D sections of the solar corona with a realistic geometry have been 
103: simulated by \citet{gud05}. While this approach has 
104: advantages when investigating the coronal loop dynamics within its neighboring 
105: coronal region, modeling numerically a larger part of the solar corona drastically reduces 
106: the number of points occupied by the coronal loops. Thus, these  simulations 
107: have not been able to shed further light on the \emph{physical mechanism} 
108: responsible for the coronal heating.
109: 
110: In \S~\ref{model} we introduce the coronal loop 
111: model and the simulations we have carried out; 
112: in \S~\ref{results} we describe
113: our numerical results, and in \S~\ref{disc} we give simple scaling arguments 
114: to understand the magnetic energy spectral slopes. This will lead to a 
115: quantitative asymptotic estimate of the coronal loop heating rate, and of its 
116: scaling with the axial magnetic field, photospheric velocity amplitude 
117: and coronal loop length. 
118: \begin{figure}
119:      \includegraphics[width=0.47\textwidth]{f1.eps}
120:       \caption{High-resolution simulation with 
121:                 $v_{\mathcal A}/u_{ph} = 200$, 512x512x200 
122:                 grid points and $\mathcal{R}_1=800$.
123:                 Magnetic ($E_M$) and kinetic ($E_K$) energies as a function 
124:                 of time ($\tau_{\mathcal A}=L/v_{\mathcal A}$ is the axial Alfv\'enic 
125:                 crossing time).  
126:      \label{en}}
127: \end{figure}
128: 
129: 
130: \section{THE MODEL} \label{model}
131: 
132: A coronal loop is a closed magnetic structure threaded by a  strong axial 
133: field, with the footpoints rooted in the photosphere.
134: This makes it a strongly anisotropic system, as measured by 
135: the relative magnitude of the Alfv\'en velocity 
136: $v_{\mathcal A} \sim 1000\ \textrm{km}\, \textrm{s}^{-1}$ compared 
137: to the typical photospheric velocity
138: $u_{ph} \sim 1\ \textrm{km}\, \textrm{s}^{-1}$. 
139: This means that the relative amplitude of the
140: Alfv\'en waves that are launched into the corona is very
141: small. The loop dynamics 
142: may be studied in 
143: a simplified geometry, neglecting any curvature effect,  as a 
144: ``straightened out'' cartesian box,  
145: with an orthogonal square cross section of 
146: size $\ell_{\perp}$, and an 
147: axial length $L$ embedded in an axial homogeneous uniform magnetic field 
148: $\boldsymbol{B}_0 = B_0\ \boldsymbol{e}_z$. This system may be described 
149: by the reduced MHD (RMHD) equations 
150: \citep{kp74,stra76}: introducing the velocity and magnetic field 
151: potentials $\varphi$ and $\psi$,
152: $\boldsymbol u_\perp = \boldsymbol{\nabla} \times 
153: \left( \varphi\, \boldsymbol{e}_z \right)$,
154: $\boldsymbol b_\perp = \boldsymbol{\nabla} \times 
155: \left( \psi\, \boldsymbol{e}_z \right)$, and vorticity and current, 
156: $\omega =- \boldsymbol{\nabla}^2_\perp \varphi$,
157: $j = - \boldsymbol{\nabla}^2_\perp \psi$ the non-dimensioned RMHD system is  
158: given by
159: \begin{eqnarray}
160: & &\frac{\partial \psi}{\partial t} = v_{\mathcal A}\, 
161: \frac{\partial \varphi}{\partial z} + 
162: \left[ \varphi, \psi \right] + \frac{\left(-1\right)^{n+1}}{\mathcal{R}_n} \boldsymbol{\nabla}^{2n}_\perp 
163: \psi,  \label{pot1} \\
164: & &\frac{\partial \omega}{\partial t} = v_{\mathcal A}\, 
165: \frac{\partial j}{\partial z} + 
166: \left[ j , \psi \right] - \left[ \omega , \varphi \right] 
167: + \frac{\left(-1\right)^{n+1}}{\mathcal{R}_n} \boldsymbol{\nabla}^{2n}_\perp \omega. \label{pot2} 
168: \end{eqnarray}
169: As characteristic quantities we use the perpendicular length of the computational
170: box $\ell_{\perp}$, the typical photospheric velocity $u_{ph}$, and the related
171: crossing time $t_{\perp} = \ell_{\perp} / u_{ph}$. The equations have been 
172: rendered dimensionless using velocity units for the magnetic field (the density in the 
173: loops $\rho$ is taken to be constant) and normalizing by $u_{ph}$.
174: Then the non-dimensioned Alfv\'en speed $v_{\mathcal A}$ in eqs.~(\ref{pot1})-(\ref{pot2})
175: is given by the ratio $v_{\mathcal A}/u_{ph}$ between the dimensional velocities.
176: The Poisson bracket of two functions $g$ and $h$ is defined as
177: $\left[ g, h \right] = \partial_x g\, \partial_y h - \partial_y g\, \partial_x h$,
178: where $x,y$ are transverse coordinates across the loop while $z$ is the axial 
179: coordinate along the loop. A simplified diffusion model is assumed and
180: $\mathcal{R}_n$ is the Reynolds number, with $n$ the hyperdiffusion index
181: (\emph{dissipativity}): for $n=1$ ordinary diffusion is recovered.
182: 
183: 
184: The computational box spans $ 0 \le x, y \le 1 $ and $ 0 \le z \le L$, 
185: with $L=10$, corresponding to an aspect ratio equal to $10$.
186: As boundary conditions at the photospheric surfaces 
187: ($z=0,\ L$) we impose a velocity pattern intended to
188: mimic photospheric motions, made up of two independent large spatial 
189: scale projected convection cell flow patterns.  The wave number values 
190: $k$ excited are all those in the range $3 \le k \le 4$, and the 
191: average injection wavenumber is $k_{in} \sim 3.4$.
192: \begin{figure}
193:      \includegraphics[width=0.47\textwidth]{f2.eps}
194:       \caption{Same simulation of Figure~\ref{en}.
195:                  The integrated Poynting flux $S$ dynamically balances the 
196:                  Ohmic ($J$) and viscous ($\Omega$) dissipation.
197:                  Inset shows a magnification of total dissipation and
198:                  $S$ for $200 \le t/\tau_{\mathcal{A}} \le 300$.
199:      \label{diss}}
200: \end{figure}
201: 
202: 
203: \section{RESULTS} \label{results}
204: 
205: Plots of the rms magnetic and kinetic energies as a function of time, 
206: together with 
207: the dissipation due to currents, vorticity, as well as the integrated 
208: Poynting flux, 
209: are shown in Figures~\ref{en} and \ref{diss}. As a result of the photospheric forcing, 
210: energy in the magnetic 
211: field first grows with time, until it dominates over the kinetic energy 
212: by a large 
213: factor, before oscillating, chaotically, around a stationary state.  Fluctuating 
214: magnetic energy $E_M$ is $\sim 35$ times bigger than kinetic energy $E_K$.
215: \begin{figure} \label{fig:hypsp}
216: \includegraphics[width=0.45\textwidth]{f3.eps}
217:       \caption{Time-averaged total energy spectra
218:       for simulations with
219:       $v_{\mathcal A}/u_{ph} = 50,\ 200,\ 400,\ 1000$. Hyperdiffusion ($n=4$) 
220:       has been used  with 
221:       $\mathcal{R}_4 = 3 \cdot 10^{20}$, $10^{20}$, $10^{19}$,
222:       $10^{19}$ respectively, and a grid with
223:       $512\times512\times200$ points.}
224: \end{figure}
225: 
226: The same generic features are seen in the rms current and vorticity dissipation, 
227: where however the time dependence of the signal is more strongly oscillating. 
228: The ohmic dissipation rate $J$ is $\sim 6.5$ times viscous dissipation $\Omega$. 
229: The Poynting flux, on average, follows the current dissipation (there is no 
230: accumulation of energy in the box), however a detailed  examination shows that  
231: the dissipation time-series tends to lag the Poynting flux, with notable 
232: de-correlations around significant dissipation peaks. 
233: The spatial configuration of the currents which corresponds to a snapshot at a 
234: given time is displayed in Figure~\ref{isofig}. The currents
235: collapse into warped, torn sheets 
236: which extend almost completely along the loop. The current peaks are embedded 
237: within the 2D sheet-like structures,  corresponding to an anisotropic structure 
238: for the turbulence, in agreement with previous results. 
239: 
240: A dimensional analysis of eqs.~(\ref{pot1})-(\ref{pot2}) shows that
241: the only free nondimensional quantity is $f = \ell_{\perp} v_{\mathcal A} /L u_{ph}$. 
242: We fix
243: $L / \ell_{\perp} = 10$ and 
244: vary the ratio  of the Alfv\'en speed to photospheric 
245: convection speed $v_{\mathcal A} / u_{ph}$.
246: Both runs with standard second order dissipation ($n=1$) as well as
247: hyperdiffusion ($n=4$) have been carried out to obtain extended inertial ranges in the 
248: resulting spectra.
249: 
250: The power spectrum of total energy in the 
251: simulation box, once a statistically stationary state has been achieved, 
252: depends strongly on the ratio $v_{\mathcal A}/u_{ph}$.
253: This was first found in simulations by \citet{dmi03},
254: devoted to understanding how anisotropic regimes of MHD turbulence depend on boundary driving strength, with whom our numerical work is in broad agreement.
255: 
256: 
257: 
258: The total energy spectrum, for values of 
259: $v_{\mathcal A}/u_{ph} = 50,\ 200,\ 400,\ 1000$ is shown in 
260: Figure~\ref{fig:hypsp}, together with fits to the inertial range power law. As 
261: $v_{\mathcal A}/u_{ph}$ increases, 
262: the spectrum steepens visibly (note that the hump at the high wave-vector values 
263: for the runs with large $v_{\mathcal A}/u_{ph}$ is a feature, the bottleneck effect, which is well 
264: known and documented in spectral simulations of turbulence with the hyperdiffusion
265: used here, e.g.\ \citet{falk94}), with the slopes ranging from $-2$ to almost $-3$. At the same time
266: while total energy increases, the ratio of the mean magnetic field over the
267: axial Alfv\'en velocity decreases, in good accordance with the theory.
268: This steepening, which may be 
269: interpreted both as the effect of inertial line-tying of the 
270: coronal magnetic field and the progressive weakening of non-linear interactions 
271: as the magnetic field is increased, has a strong and direct bearing on the coronal heating 
272: scaling laws.
273: 
274: 
275: \section{DISCUSSION} \label{disc}
276: 
277: A characteristic of anisotropic MHD turbulence is that
278: the cascade takes place mainly in the plane orthogonal
279: to the DC magnetic guide field \citep{sheb83}.
280: Consider then the anisotropic version of the 
281: Iroshnikov-Kraichnan (IK) theory \citep{gold94,gold97}. 
282: Dimensionally the energy cascade rate 
283: may be written as
284: $\rho\, {\delta z_{\lambda}}^2 / T_{\lambda}$,
285: where $\delta z_{\lambda}$ is the rms value of the Els\"asser fields
286: $\bsy{z}^{\pm} = \bsy{u}_{\perp} \pm \bsy{b}_{\perp}$ at the 
287: perpendicular scale $\lambda$, where because the system is 
288: magnetically dominated $\delta z^+_{\lambda} \sim \delta z^-_{\lambda}$. 
289: $\rho$ is the average density and 
290: $T_{\lambda}$ is the energy transfer time
291: at the scale $\lambda$, which is greater than the eddy turnover time 
292: $\tau_{\lambda} \sim \lambda / \delta z_{\lambda}$ because of the
293: Alfv\'en effect \citep{iro64,kra65}. 
294: 
295: In the classical IK case,  
296: $T_{\lambda}\sim \tau_{\mathcal A} \bigl(\tau_{\lambda}/\tau_{\mathcal A}\bigr)^2$. 
297: This corresponds to the fact that wave-packets interact over an 
298: Alfv\'en crossing time (with $\tau_{\lambda} > \tau_{\mathcal A}$), 
299: and the collisions follow a standard random walk in energy exchange. 
300: In terms of the number of collisions $N_{\lambda}$ that a wave packet must 
301: suffer for the perturbation to build up to order unity, for IK
302: $N_{\lambda} \sim (\tau_{\lambda} / \tau_{\mathcal A})^2$. 
303:  
304: More generally, however, as the Alfv\'en speed is increased the interaction
305: time becomes smaller, so that  turbulence becomes weaker and the number 
306: of collisions required for efficient energy transfer scales as
307: \be \label{eq:atnc}
308: N_{\lambda} = \left( \frac{\tau_{\lambda}}{\tau_{\mathcal A}} \right)^{\alpha}
309: \qquad \mathrm{with} \qquad \alpha > 2,
310: \ee
311: where $\alpha$ is the scaling index
312: (note that $\alpha =1$ corresponds to standard hydrodynamic turbulence),
313: so that 
314: \be \label{eq:btnc}
315: T_{\lambda} \sim N_{\lambda}\, \tau_{\mathcal A} \sim
316: \left( \frac{v_{\mathcal A}}{L} \right)^{\alpha -1} 
317: \left( \frac{\lambda}{\delta z_{\lambda}} \right)^{\alpha}.
318: \ee
319: Integrating over the whole volume, the energy transfer rate becomes
320: \be \label{eq:sbe1}
321: \epsilon \sim \ell_{\perp}^2 L\cdot \rho\, \frac{\delta z_{\lambda}^2}{T_{\lambda}} 
322: \sim \ell_{\perp}^2 L\cdot \rho\, \left( \frac{L}{v_{\mathcal A}} \right)^{\alpha - 1} \, 
323: \frac{\delta z_{\lambda}^{\alpha + 2}}{\lambda^{\alpha}}.
324: \ee
325: Considering the injection scale $\lambda \sim \ell_{\perp}$,
326: eq.~(\ref{eq:sbe1}) becomes
327: \be \label{eq:sbe2}
328: \epsilon 
329: \sim \ell_{\perp}^2 L\cdot \rho\, \frac{\delta z_{\ell_{\perp}}^2}{T_{\ell_{\perp}}} 
330: \sim \frac{\rho \ell_{\perp}^2 L^{\alpha}}{\ell_{\perp}^{\alpha} \, v_{\mathcal A}^{\alpha - 1}} \, 
331: \delta z_{\ell_{\perp}}^{\alpha + 2}.
332: \ee
333: On the other hand the energy injection rate is given by the 
334: Poynting flux integrated across the photospheric boundaries: 
335: $\epsilon_{in} = \rho\, v_{\mathcal A} 
336: \int \! \mathrm{d} a\, \bsy{u}_{ph} \cdot \bsy{b}_{\perp}$.
337: Considering that  this integral is dominated by energy at the
338: large scales, due to the characteristics of the forcing function, we can approximate it with 
339: \be \label{eq:pf}
340: \epsilon_{in} \sim \rho\, \ell_{\perp}^2 v_{\mathcal A} u_{ph} \delta z_{\ell_{\perp}},
341: \ee
342: where the large scale component of the magnetic
343: field can be replaced with $\delta z_{\ell_{\perp}}$ because the system is magnetically dominated.
344: 
345: The last two equations show that the system is self-organized because 
346: both $\epsilon$ and $\epsilon_{in}$ depend on $\delta z_{\ell_{\perp}}$, 
347: the rms values of the fields
348: $\boldsymbol{z}^{\pm}$ at the scale $\ell_{\perp}$:
349: the internal dynamics depends
350: on the injection of energy and the injection of energy itself depends 
351: on the internal dynamics via the boundary forcing. 
352: Another aspect of self-organization results from our simulations: 
353: the perpendicular magnetic field develops few spatial structures
354: along the axial direction $z$, and in the nonlinear stage 
355: its topology substantially departs from the mapping of
356: the boundary velocity pattern which characterizes its evolution
357: during the linear stage. These and other features will be 
358: discussed more in depth in \citet{rapprep}.
359: 
360: In a stationary cascade the injection rate (\ref{eq:pf}) is equal to 
361: the transport rate (\ref{eq:sbe2}). Equating the two yields for 
362: the amplitude at the scale $\ell_{\perp}$:
363: \be \label{eq:amp}
364: \frac{\delta z_{\ell_{\perp}}^{\ast}}{u_{ph}} 
365: \sim \left( \frac{\ell_{\perp} v_{\mathcal A}}{L u_{ph}} \right)^{\frac{\alpha}{\alpha + 1}}
366: \ee
367: Substituting this value in (\ref{eq:sbe2}) or (\ref{eq:pf}) we obtain
368: for the energy flux
369: \be \label{eq:chs}
370: \epsilon^{\ast} 
371: \sim \ell_{\perp}^2 \, \rho \, v_{\mathcal A} u_{ph}^2 
372: \left( \frac{\ell_{\perp} v_{\mathcal A}}{L u_{ph}} \right)^{\frac{\alpha}{\alpha+1}},
373: \ee
374: where $v_{\mathcal A} = B_0 / \sqrt{4\pi\rho}$.
375: This is also the dissipation rate, and hence the \emph{coronal heating scaling}.
376: The crucial parameter here is $f = \ell_{\perp} v_{\mathcal A}/ L u_{ph}$ 
377: because the scaling index $\alpha$~(\ref{eq:atnc}), upon which the strength of 
378: the stationary turbulent 
379: regime depends, must be a function of $f$ itself. The relative amplitude of 
380: the turbulence $\delta z_{\ell_{\perp}}^{\ast} / v_{\mathcal A}$, is a function of $f$, 
381: and as
382: $f$ increases the effect of \emph{line-tying} becomes stronger, decreasing the 
383: strength of turbulent interactions (wave-packet collision efficiency becomes 
384: sub-diffusive) so that $\alpha$ increases above $2$. 
385: The ratio $\delta z_{\ell_{\perp}}^{\ast} / v_{\mathcal A}$
386: can also be interpreted as the rms value of the
387: Parker angle $\Theta_P$,
388: and is given by
389: \be \label{pan}
390: < \Theta_P > 
391: \sim \frac{\delta z_{\ell_{\perp}}^{\ast}}{v_{\mathcal A}} 
392: \sim \left( \frac{\ell_{\perp}}{L} \right)^{\frac{\alpha}{\alpha + 1}}
393: \left( \frac{u_{ph}}{v_{\mathcal A}} \right)^{\frac{1}{\alpha + 1}}.
394: \ee
395: This is actually an estimate of the average inclination of the magnetic field
396: lines,  while the rms value of the shear  angle between 
397: neighboring field lines is at least twice that given by eq.~(\ref{pan}),
398: not considering that close to a current sheet an enhancement of 
399: the orthogonal magnetic field is observed (which leads to a 
400: higher value for the angle).
401: 
402: Numerical simulations determine the remaining unknown nondimensional 
403: dependence of the scaling index $\alpha$ on $f$. The power law slopes of the total energy
404: spectra shown in Figure~\ref{fig:hypsp} are used to determine  $\alpha$.
405: Identifying, as usual, the eddy energy with the band-integrated
406: Fourier spectrum $\delta z_{\lambda}^2 \sim k_{\perp}\, E_{k_{\perp}}$,
407: where $k_{\perp} \sim \ell_{\perp}/\lambda$, from eq.~(\ref{eq:sbe1}) we obtain
408: \begin{equation}
409: E_{k_{\perp}} \propto k_{\perp}^{-\frac{3\alpha+2}{\alpha+2}},
410: \end{equation}
411: where for $\alpha = 1$ the $-5/3$ slope for the ``anisotropic Kolmogorov''
412: spectrum is recovered, and for $\alpha = 2$ the $-2$ slope for the
413: anisotropic IK case. At higher values of $\alpha$ correspond 
414: steeper spectral slopes up to the asymptotic value of $-3$.
415: 
416: \begin{figure}
417:      \includegraphics[width=0.47\textwidth]{f4.eps}
418:           \caption{The solid line shows the exponent $\alpha/(\alpha+1)$ as a function
419:           of $\alpha$. Symbols  show values of $\alpha$ corresponding
420:           to different values of  $v_{\mathcal A}/u_{ph}$,
421:           at fixed $L/\ell_{\perp}=10$. \label{fig:spzl}}
422: \end{figure}
423: 
424: In Figure~\ref{fig:spzl} we plot the values of $\alpha$ determined in
425: this way,  together 
426: with the resulting power dependence $\alpha/(\alpha+1)$ of the 
427: amplitude~(\ref{eq:amp}) and of the energy flux~(\ref{eq:chs})  
428: on the parameter $f$.
429: The other power dependences are easily obtained from this last one,
430: e.g.\  for the energy flux~(\ref{eq:chs}) the power of the axial 
431: Alfv\'en speed $v_{\mathcal A}$ is given by $1+ \alpha/(\alpha+1)$,
432: so that in terms of the magnetic field $B_0$ it scales as $B_0^{3/2}$ for weak 
433: fields and/or long loops, to $B_0^2$ for strong fields 
434: and short loops. 
435: 
436: 
437: Dividing eq.~(\ref{eq:chs}) by the surface $\ell_{\perp}^2$ we obtain the energy flux 
438: per unit area $F=\epsilon^{\ast}/\ell_{\perp}^2$.
439: Taking for example a coronal loop $40,000\ \textrm{km}$ long, 
440: with a  number density of 
441: $10^{10}\ \textrm{cm}^{-3}$, $v_{\mathcal A} = 2,000\ \textrm{km}\, \textrm{s}^{-1}$ and  
442: $u_{ph} = 1\ \textrm{km}\, \textrm{s}^{-1}$, 
443: (for these parameters we can estimate a value of $\alpha/(\alpha+1) \sim 0.95$),
444: which models an active region loop, we obtain 
445: $F \sim 5 \cdot {10}^6\ \textrm{erg}\, \textrm{cm}^{-2}\, \textrm{s}^{-1}$ 
446: and a Parker angle~(\ref{pan}) 
447: $<\Theta_P> \sim 4^{\circ}$. On the other hand, for
448: a coronal loop typical of a quiet Sun region, with a length of
449:  $100,000\ \textrm{km}$, a  number density of 
450: $10^{10}\ cm^{-3}$, $v_{\mathcal A} = 500\ \textrm{km}\, \textrm{s}^{-1}$ and  
451: $u_{ph} = 1\ \textrm{km}\, \textrm{s}^{-1}$, 
452: (for these parameters we can estimate a value of $\alpha/(\alpha+1) \sim 0.7$) 
453: we obtain $F \sim 7\cdot {10}^4\ \textrm{erg}\, \textrm{cm}^{-2}\, \textrm{s}^{-1}$ and 
454: $<\Theta_P> \sim 0.9^{\circ}$. 
455: 
456: In summary, with this paper we have shown how coronal heating rates in the 
457: Parker scenario scale with coronal loop and photospheric driving parameters,
458: demonstrating that field line tangling can supply the coronal heating energy 
459: requirement. We also predict that there is no universal scaling with axial 
460: magnetic field intensity, a feature which can be tested 
461: by observing weak field regions on the Sun, or the atmospheres 
462: of other stars with differing levels of magnetic activity.
463: 
464: \acknowledgements
465: M.V. thanks W.H.~Matthaeus for useful 
466: discussions. R.B.D. is supported by NASA SPTP.  \\
467: 
468: \begin{thebibliography}{100}
469: \bibitem[Dmitruk et al.(1998)]{dmi98}Dmitruk, P., 
470: G\'omez, D.~O., \& DeLuca, D.~D. 1998, \apj, 505, 974
471: 
472: \bibitem[Dmitruk \& G\'omez(1999)]{dmi99}Dmitruk, P., \&
473: G\'omez, D.~O. 1999, \apj, 527, L63
474: 
475: \bibitem[Dmitruk et al.(2003)]{dmi03}Dmitruk, P., 
476: G\'omez, D.~O., \& Matthaeus, W.~H. 2003, Phys. Plasmas, 10, 3584
477: 
478: \bibitem[Einaudi et al.(1996)]{ein96}Einaudi, G., 
479: Velli, M., Politano, H., \& Pouquet, A. 1996, \apj, 457, L113
480: 
481: \bibitem[Einaudi \& Velli(1999)]{ein99}Einaudi, G., \&
482: Velli, M. 1999, Phys. Plasmas, 6, 4146
483: 
484: \bibitem[Falkovich(1994)]{falk94}Falkovich, G. 1994, Phys. Fluids, 6, 1411
485: 
486: \bibitem[Georgoulis et al.(1998)]{georg98} Georgoulis, M.~K., 
487: Velli, M., \& Einaudi, G., 1998, \apj, 497, 957
488: 
489: \bibitem[Goldreich \& Sridhar(1997)]{gold97}Goldreich, P.,  \& Sridhar, S.
490: 1997, \apj, 485, 680
491: 
492: \bibitem[Gudiksen \& Nordlund(2005)]{gud05}Gudiksen, B.~V., \&
493: Nordlund, \AA. 2005, \apj, 618, 1020
494: 
495: \bibitem[Hendrix \& Van~Hoven(1996)]{hen96}Hendrix, D.~L., \&
496: Van~Hoven, G. 1996, \apj, 467, 887
497: 
498: \bibitem[Iroshnikov(1964)]{iro64}Iroshnikov, P.~S. 
499: 1964, Sov. Astron., 7, 566
500: 
501: \bibitem[Kadomtsev \& Pogutse(1974)]{kp74}Kadomtsev, B.~B., 
502: \& Pogutse, O.~P. 1974, Sov. J. Plasma Phys., 1, 389
503: 
504: \bibitem[Kraichnan(1965)]{kra65}Kraichnan, R.~H. 
505: 1965, Phys. Fluids, 8, 1385
506: 
507: \bibitem[Longcope \& Sudan(1994)]{long94}Longcope, D.~W., \&
508: Sudan, R.~N. 1994, \apj, 437, 491
509: 
510: \bibitem[Mikic et al.(1989)]{mik89}Mikic, Z., Schnack, D.~D., \&
511: Van~Hoven, G. 1989, \apj, 338, 1148
512: 
513: \bibitem[Parker(1972)]{park72}Parker, E.~N. 1972, \apj, 174, 499
514: 
515: \bibitem[Parker(1988)]{park88}Parker, E.~N. 1988, \apj, 330, 474
516: 
517: \bibitem[Rappazzo et al.(2007)]{rapprep}Rappazzo, A.~F., Velli, M.,
518: Einaudi, G., \& Dahlburg, R.~B., \apj, \emph{in preparation}
519: 
520: \bibitem[Shebalin et al.(1983)]{sheb83} Shebalin, J.~V., Matthaeus, W.~H.,
521: \& Montgomery, D. 1983, J. Plasma Phys., 29, 525
522: 
523: \bibitem[Sridhar \& Goldreich(1994)]{gold94} Sridhar, S. \& Goldreich, P.
524: 1994, \apj, 432, 612
525: 
526: \bibitem[Strauss(1976)]{stra76}Strauss, H.~R. 1976, 
527: Phys. Fluids, 19, 134
528: \end{thebibliography}
529: 
530: 
531: \begin{figure}[p]
532:       \centering
533:       \subfloat{
534:                \includegraphics[height=0.56\linewidth]{f5a.eps}}\\[20pt]
535:       \subfloat{
536:                \includegraphics[height=0.56\linewidth]{f5b.eps}}
537:       \caption{\emph{Top}: side view of 
538:                two isosurfaces of the squared current at a selected time
539:                for a numerical simulation with $v_{\mathcal A}/u_{ph} = 200$,
540:                 512x512x200 grid points and a Reynolds
541:                number $\mathcal R_1 =800$.
542:                The isosurface at the value $j^2 = 2.8 \cdot 10^5$ is represented in 
543:                partially
544:                transparent yellow,  while  red displays
545:                the isosurface with  $j^2 = 8 \cdot 10^5$, well below the value of the
546:                maximum of the squared current that at this time is 
547:                $j^2 = 8.4 \cdot 10^6$.
548:                N.B.: The red isosurface is always nested inside the yellow one, and
549:                appears pink in the figure. The computational box has been rescaled
550:                for an improved viewing, but the aspect ratio of the box is $10$,
551:                i.e.\ the axial length of the box is ten times bigger than
552:                its orthogonal length.
553:                \emph{Bottom}: top view  of the same two isosurfaces
554:                 using the same color display.
555:                 The isosurfaces are extended along the axial direction, and the
556:                 corresponding filling factor is small.}
557:                 \label{isofig}
558: \end{figure}
559: 
560: \end{document}
561: 
562: