1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[preprint2]{aastex}
3: %\documentclass[manuscript]{aastex}
4: \shorttitle{Catastrophic flux rope eruption}
5:
6: \shortauthors{Chen, Hu {\&} Sun}
7:
8: \begin{document}
9:
10: \title{Catastrophic eruption of magnetic flux rope in the corona and
11: solar wind with and without magnetic reconnection}
12:
13: \author{Y. Chen\altaffilmark{1,2}, Y. Q. Hu\altaffilmark{2}, AND S. J. SUN\altaffilmark{2}}
14:
15: \altaffiltext{1}{Institute for Space Sciences, Shandong
16: University; Department of Space Science and Applied Physics,
17: Shandong University at Weihai, Weihai Shandong, 264209 China;
18: yaochen@sdu.edu.cn} \altaffiltext{2}{School of Earth and Space
19: Sciences, University of Science and Technology of China, Hefei
20: Anhui 230026, China}
21:
22: \begin{abstract}
23: It is generally believed that the magnetic free energy accumulated
24: in the corona serves as a main energy source for solar explosions
25: such as coronal mass ejections (CMEs). In the framework of the
26: flux rope catastrophe model for CMEs, the energy may be abruptly
27: released either by an ideal magnetohydrodynamic (MHD) catastrophe,
28: which belongs to a global magnetic topological instability of the
29: system, or by a fast magnetic reconnection across preexisting or
30: rapidly-developing electric current sheets. Both ways of magnetic
31: energy release are thought to be important to CME dynamics. To
32: disentangle their contributions, we construct a flux rope
33: catastrophe model in the corona and solar wind and compare
34: different cases in which we either prohibit or allow magnetic
35: reconnection to take place across rapidly-growing current sheets
36: during the eruption. It is demonstrated that CMEs, even fast ones,
37: can be produced taking the ideal MHD catastrophe as the only
38: process of magnetic energy release. Nevertheless, the eruptive
39: speed can be significantly enhanced after magnetic reconnection
40: sets in. In addition, a smooth transition from slow to fast
41: eruptions is observed when increasing the strength of the
42: background magnetic field, simply because in a stronger field
43: there is more free magnetic energy at the catastrophic point
44: available to be released during an eruption. This suggests that
45: fast and slow CMEs may have an identical driving mechanism.
46: \end{abstract}
47:
48: \keywords{Sun: corona $-$ Sun: magnetic fields $-$ Sun: coronal
49: mass ejections(CMEs)}
50:
51: \section{INTRODUCTION}
52:
53: It is generally believed that the accumulated magnetic free energy
54: serves as a main energy source for the spectacular solar eruptive
55: phenomena such as coronal mass ejections (CMEs) (see reviews by
56: Forbes (2000) and Low (2001)), but it remains open how the
57: magnetic energy is released. Among various scenarios, the flux
58: rope catastrophe mechanism is a very promising one (reviewed
59: recently by Lin et al. (2003) and Hu (2005)). The catastrophe is
60: an ideal magnetohydrodynamic (MHD) process belonging to a global
61: magnetic topological instability of the system. It releases energy
62: without ohmic heating, especially suitable for CMEs without
63: associated flares (e.g., Forbes {\&} Isenberg, 1991; Isenberg et
64: al., 1993; Forbes {\&} Priest, 1995; Hu et al., 2003b). A
65: by-product of the catastrophe is the formation of one or more
66: electric current sheets, which grows at the Alfv\'enic time scale
67: and provides proper sites for fast magnetic reconnection. Such a
68: reconnection further releases the stored magnetic energy and
69: should be responsible for a solar flare associated with a CME
70: event (e.g., Forbes {\&} Lin, 2000; Lin {\&} Forbes, 2000). Thus,
71: in terms of this scenario, there are two means of magnetic energy
72: release process involved in solar eruptions, which are the ideal
73: MHD catastrophe and resistive magnetic reconnection. Both
74: processes are thought to be important to the CME initiation and
75: acceleration. It is therefore necessary and important to
76: disentangle their contributions to the CME dynamics and
77: energetics. This article serves as a first step to solve this
78: problem under the framework of the specific catastrophe model for
79: CMEs.
80:
81: The eruptive flux rope after catastrophe is accelerated mainly by
82: the net or unbalanced Lorentz force, which is contributed by
83: various source currents both inside and outside the flux rope,
84: including that along each electric current sheet and the
85: corresponding image current, and the background potential field
86: which is determined by the normal component of the background
87: field at the solar surface. To reveal the roles of these various
88: components of magnetic forces in sustaining the flux rope in
89: equilibrium and causing the eruption, Chen et al. (2006b)
90: conducted detailed force balance analyses. It was found that the
91: primary lifting force is provided by the azimuthal current inside
92: the rope and its image below the photosphere, which is mainly
93: balanced by the pulling force produced by the background potential
94: field when the rope is in equilibrium. During an eruption of the
95: rope caused by the catastrophe, the force associated with the
96: rapidly-developing current sheet(s) constitutes an additional and
97: significant restoring force that decelerates the rope. This force
98: will be greatly reduced or even eliminated once magnetic
99: reconnection sets in across the current sheet(s), as a consequence
100: the magnetic reconnection may cause a further acceleration of the
101: flux rope. This will be confirmed and quantified by the
102: calculations presented in this article with an axisymmetric flux
103: rope catastrophe model. The effect of the solar wind plasma with
104: an equatorial current sheet above the streamer cusp point
105: extending to infinity is considered so as to get a more realistic
106: description of the CME acceleration and propagation. Note that in
107: most previous studies on the flux rope catastrophe this effect was
108: not included. Sun {\&} Hu (2005) did consider the effect of the
109: solar wind on the rope catastrophe and the consequent eruption.
110: They suggested that the flux rope catastrophe can serve as a
111: mechanism for slow CMEs. Yet, they investigated only the cases
112: corresponding to a relatively weak background field in terms of
113: ideal MHD. Based on their work, we take the effect of both the
114: solar wind and magnetic reconnection into account in the present
115: study, which may be regarded as a starting point to disentangle
116: the impacts of MHD catastrophe and magnetic reconnection on the
117: CME dynamics in the solar wind. We put our focus on the
118: differences between the flux rope dynamics in two situations in
119: which we either prohibit or allow magnetic reconnection to take
120: place across the rapidly-growing current sheets. We introduce the
121: background corona and solar wind model and the flux rope
122: catastrophe model in the following section. In $\S$ 3.1 we compare
123: the solutions with and without magnetic reconnection to illustrate
124: its impact on the CME dynamics. We emphasize the effect of the
125: background magnetic field strength in $\S$ 3.2 and provide a brief
126: summary of this article with discussions in the last section of
127: this article.
128:
129: \section{FLUX ROPE CATASTROPHE MODEL IN THE CORONA AND SOLAR WIND}
130: Coronal magnetic flux ropes are in essence three-dimensional
131: structures with two ends anchored to the photosphere. Considering
132: that the length of a magnetic flux rope is much larger than its
133: diameter, the axisymmetrical simplification (also called 2.5-D
134: models) can be used to approximate the realistic flux rope system.
135: With this approximation, the magnetic field $\mathbf{B}$ can be
136: expressed with a magnetic flux function $\psi(t,r,\theta)$ in
137: spherical coordinates ($r$, $\theta$, $\varphi$),
138: \begin{equation}
139: \emph{\textbf{B}}= \nabla\times({\psi \over
140: r\sin\theta}\mathbf{\hat{\varphi}})
141: +B_\varphi\mathbf{\hat{\varphi}}. %\eqno{(1)}
142: \end{equation}
143: The derived resistive MHD equations in spherical coordinates are
144: written into the following form (see also Ding et al., 2006):
145: \begin{equation}
146: {\partial \rho \over \partial
147: t}+\nabla\cdot(\rho\emph{\textbf{v}})=0, %\eqno{(2)}
148: \end{equation}
149: \begin{equation}
150: \begin{array}{c}
151: {\partial \emph{\textbf{v}} \over \partial t} +
152: \emph{\textbf{v}}\cdot\nabla\emph{\textbf{v}}+{1\over \rho}\nabla
153: p+{1\over \mu\rho}[L\psi\nabla\psi+
154: \emph{\textbf{B}}_\varphi\times(\nabla\times\emph{\textbf{B}}_\varphi)]\\[3mm]
155: +{1\over\mu\rho
156: r\sin{\theta}}\nabla\psi\cdot(\nabla\times\emph{\textbf{B}}_\varphi)\hat{\varphi}+{GM_\odot\over
157: r^2}\hat{r}=0, %\eqno{(3)}
158: \end{array}
159: \end{equation}
160: \begin{equation}
161: {\partial \psi\over\partial
162: t}+\emph{\textbf{v}}\cdot\nabla\psi-{1\over\mu}\eta
163: r^2\sin^2\theta L\psi=0, %\eqno{(4)}
164: \end{equation}
165: \begin{equation}
166: \begin{array}{c}
167: {\partial B_\varphi \over \partial t}
168: +r\sin\theta\nabla\cdot({B_\varphi\emph{\textbf{v}}\over
169: r\sin\theta})+[\nabla\psi\times\nabla({v_\varphi\over r
170: \sin\theta})]_\varphi\\[3mm]
171: -{1\over r\sin\theta}\nabla\eta \cdot\nabla(\mu r\sin\theta
172: B_\varphi)-{1\over\mu}\eta r \sin\theta L(r
173: B_\varphi\sin\theta)=0,
174: % \eqno{(5)}
175: \end{array}
176: \end{equation}
177: \begin{equation}
178: {\partial T\over \partial t}+\emph{\textbf{v}}\cdot\nabla
179: T+(\gamma-1)T\nabla\cdot \emph{\textbf{v}}-{{\gamma-1}\over \rho
180: R}\eta \emph{\textbf{j}}^2=0, %\eqno{(6)}
181: \end{equation}
182: where
183: \begin{equation}
184: L\equiv{1\over r^2\sin^2\theta}({\partial^2 \over
185: \partial r^2}+{1\over r^2}{\partial^2 \over \partial \theta^2}
186: -{\cot\theta \over r^2} {\partial \over \partial \theta}),
187: %\eqno{(7)}
188: \end{equation}
189: \begin{equation}
190: \emph{\textbf{j}}={1\over \mu}\nabla\times
191: \emph{\textbf{B}}=-{1\over \mu}r\sin\theta L\psi\hat\varphi +
192: {1\over\mu}\nabla\times(B_\varphi\hat\varphi). %\eqno{(8)}
193: \end{equation}
194: The symbols $\rho$, $\emph{\textbf{v}}$, $T$, and
195: $\emph{\textbf{j}}$ represent the density, flow velocity,
196: temperature and the current density, respectively. $\mu$ is the
197: vacuum magnetic permeability, $R$ the gas constant, $G$ the
198: gravitational constant, $M_\odot$ the solar mass, and $\gamma$ the
199: polytropic index which is taken to be 1.05 so as to obtain a
200: reasonable solution of the background steady-state solar wind. The
201: temperature and density at the coronal base are taken to be
202: $T_0=2\times10^6$ K and $\rho_0=1.67\times10^{-13}$ kg m$^{-3}$,
203: respectively. The magnetic flux function at the coronal base is
204: taken to be the same as a dipolar field given by
205: $\psi(r=R_\odot)=\psi_0 \sin^2\theta/R_\odot$ where $\psi_0=B_0
206: {R_\odot}^2$. $B_0$ represents half of the magnetic field strength
207: at the polar hole ($\theta=0$) on the solar surface, which can be
208: used to represent the strength of the background magnetic field of
209: the system. Note that in this study the magnetic flux function at
210: the solar surface is fixed to its initial distribution which is
211: determined by $B_0$. For the effect of the photospheric magnetic
212: flux distribution on the coronal flux rope catastrophe, please
213: check a recent paper written by Sun et al. (2007). The parameter
214: $B_0$ will be freely adjusted. For example, $\psi_0=9.7 \times
215: 10^{13}$ Wb with $B_0$ taken to be 2 G giving the case
216: investigated by Sun {\&} Hu (2005). It is defined that the ideal
217: MHD situation corresponds to a zero magnetic resistivity $\eta$.
218: In the resistive situation, an anomalous homogeneous resistivity
219: is used with $\eta$ given by $\eta=\eta_0 \mu v_{s0}R_\odot$ where
220: $\eta_0=0.1$ and $v_{s0}=\sqrt{2RT_0}=181.8$ km s$^{-1}$. The
221: steady-state polytropic solar wind solution is obtained by solving
222: the above MHD equations with $\eta=0$. The magnetic topology
223: (white lines) and the velocity color contour map for $B_0=6$ G
224: from 1 $R_\odot$ to 20 $R_\odot$ are illustrated in Figure 1a. For
225: other values of $B_0$, the field topology and velocity
226: distribution are basically similar to that shown in this figure.
227: It can be seen that the solution is characterized by a typical
228: streamer-current sheet-solar wind configuration. The cusp point is
229: located at about 3 $R_\odot$, and the flow velocity reaches up to
230: 400 km s$^{-1}$ at about 10 $R_\odot$ along the equator.
231:
232: Based on the obtained corona and solar wind solution, we let a
233: flux rope with prescribed mass, toroidal and poloidal magnetic
234: fluxes ($\Psi_p$ and $\Psi_\varphi$) emerge from the equator at
235: the coronal base. The detailed emerging process has been described
236: previously by Hu et al. (2003b) and Chen et al. (2006a) and will
237: not be repeated here. Special numerical measures are taken to
238: maintain $\Psi_p$ and $\Psi_\varphi$ invariant and equal to their
239: initial given values during the simulation (see Hu et al., 2003b).
240: Figure 1b exemplifies the magnetic configuration and velocity
241: contour map of a flux-rope system with the solar wind, where the
242: border of the original flux rope is depicted with a green circle.
243: It can be seen that the closed field region of the streamer
244: expands apparently with the emergence of the flux rope. Such a
245: swelling of a coronal streamer is often observed with the
246: white-light coronagraphs before CMEs (e.g., Howard et al., 1985).
247: The mass contained by the flux rope per radian in the azimuthal
248: direction is set to be ${0.5 \over 2 \pi} M_0$ where the unit of
249: mass $M_0=\rho_o R_\odot^3=5.643\times10^{13}$ kg, the poloidal
250: flux $\Psi_p$ is taken to be 0.3 in units of $\psi_0$ while the
251: toroidal flux $\Psi_\varphi$ is changeable. Thus one may find MHD
252: solutions with different values of $\Psi_\varphi$ to examine
253: whether a catastrophe occurs, and find out the meta-stable state
254: of the system characterized by $\Psi_p=0.3$ and a specific value
255: of $\Psi_\varphi$ which depends on $B_0$. Starting from this state
256: any slight increase of $\Psi_p$ or $\Psi_\varphi$ may excite the
257: catastrophe. Therefore, the state is taken as the initial state
258: for our simulation of the flux rope eruption. We choose to
259: increase $\Psi_p$ from $0.3$ to $0.305$ at $t=0$ so as to trigger
260: the catastrophe. Physically speaking, the increase of the poloidal
261: flux can be achieved by a twist of a long three-dimensional flux
262: rope anchored to the photosphere. Besides the mass, magnetic
263: fluxes of the flux rope, the helicity of the flux rope is also of
264: interest to the study on CMEs. In an axisymmetric system like that
265: investigated in this article, the two-dimensional magnetic
266: helicity can be calculated by the following integral according to
267: Hu et al. (1997),
268: $$
269: H_T=2\pi\int\int \psi B_\varphi r dr d\theta, \eqno{(9)}$$ where
270: the factor $2\pi$ comes from the integral over the azimuthal
271: direction, which should be removed if one wants to evaluate the
272: magnetic helicity per radian in the azimuthal direction.
273:
274: When the flux rope breaks away from the surface and erupts
275: upwards, a current sheet may develop below the flux rope. It is
276: well known that numerical pseudoreconnection takes place across
277: the current sheet in most numerical simulations, which causes a
278: false transfer of poloidal flux from the background to the flux
279: rope and results in a topological change. In this work, we take
280: special measure to prohibit such numerical reconnection in order
281: to investigate the flux rope dynamics in the framework of ideal
282: MHD. The magnetic flux function $\psi$ along the current sheet is
283: invariant, which is known a priori, and any reconnection across
284: the sheet reduces it in the present simulation. We therefore
285: reassign $\psi$ along the current sheet to the known constant
286: value at each time step. This technique, first proposed by Hu et
287: al. (2003b), effectively eliminates numerical reconnection across
288: the equatorial current sheet. Note that this special measure is
289: not employed in our calculations in terms of resistive MHD with a
290: non-zero $\eta$.
291:
292: The calculations are carried out in a domain of $R_\odot \le r \le
293: 30 R_\odot$ and $0 \le \theta \le \pi/2$, which is discretized
294: into 150 $\times$ 90 grid points. The grid spacing increases
295: according to a geometric series of a common ratio 1.024 along the
296: radial direction from 0.02 at the solar surface to 0.71 at the top
297: boundary. And a uniform mesh is adopted in the $\theta$ direction.
298: The multistep implicit scheme developed by Hu (1989) is used to
299: solve the MHD equations. For the eruptive solutions, the
300: calculations are terminated once the top part of the ejecta
301: reaches the upper boundary.
302:
303: \section{Numerical results}
304: In this section, we first present and compare the solutions given
305: by the ideal and resistive MHD calculations for the case with
306: $B_0=6$ G as a first step to disentangle the impacts of MHD
307: catastrophe and magnetic reconnection on the CME dynamics in the
308: solar wind. Then, we investigate the effect of the background
309: magnetic field strength by comparing results with different values
310: of $B_0$.
311:
312: \subsection{Impact of magnetic reconnection on flux rope dynamics}
313: As mentioned previously, our simulations on the flux rope
314: catastrophe and eruption start from an equilibrium state which is
315: a meta-stable flux rope system in the corona and solar wind
316: background. The catastrophe is triggered by a slight increase of
317: the poloidal flux $\Psi_p$ of the flux rope from 0.3 to 0.305 (or
318: 8.7 - 8.85 $\times 10^{13}$ Wb in physical units) at $t=0$ with
319: the critical axial flux in the rope $\Psi_\varphi=0.209$ (6.06
320: $\times 10^{13}$ Wb) for $B_0$=6 G. After that, the flux rope
321: starts to break away from the photosphere and erupts upwards.
322: Figures 1c and 1d show the magnetic topology and velocity color
323: contours at the same instant ($t=280$ minutes) for the two
324: solutions with and without magnetic reconnection (i.e., the cases
325: with $\eta \ne 0$ and $\eta = 0$). An apparent difference between
326: the two solutions lies in whether a current sheet develops below
327: the flux rope. The sheet forms and grows with the rope eruption in
328: the ideal MHD case, while it is eroded by magnetic reconnection in
329: the resistive MHD calculation. It is also apparent that a
330: significant part of the poloidal flux has been transferred from
331: the background to the flux rope, and a new streamer appears as the
332: aftermath of magnetic reconnection in the resistive calculation.
333: The rate of magnetic flux transfer is mainly determined by the
334: effective resistivity consisting of the anomalous resistivity and
335: the numerical resistivity involved in the pseudoreconnection.
336: Unfortunately, at this time, it is not possible to eliminate the
337: pseudoreconnection in our calculations in terms of resistive MHD.
338: Therefore, it is difficult to control the flux transfer rate by
339: simply adjusting the magnetic resistivity in this case. Further
340: discussion regarding this issue will be given in our discussion
341: section. Another major difference is the color distribution of the
342: velocity contour maps which indicates how fast the flux rope
343: ejecta is. It can be seen that the rope erupts faster in the
344: solution with magnetic reconnection, as will be quantitatively
345: revealed in Figure 2.
346:
347: In Figure 2, we plot the profiles of the heliocentric distance,
348: velocity, and acceleration of different parts of the flux rope
349: ejecta, including the cusp point (in dotted), the rope top (in
350: dashed), the rope axis (in solid), and the rope bottom (in
351: dot-dashed), left panels for the case without reconnection and
352: right for the reconnection case. It can be seen that in both cases
353: the flux rope starts to take off at $t \approx 70$ minutes. There
354: is an apparent delay of about 50 minutes of the time when the cusp
355: point starts to move upwards rapidly. The delay reflects the time
356: taken for the eruptive flux rope to propagate from the coronal
357: base to the initial cusp point location in the corona. We can see
358: that the cusp point undergoes the most dramatic acceleration in
359: both cases: in about 70 minutes the velocity of the cusp point
360: reaches up to 800 km s$^{-1}$ with the maximum acceleration being
361: 300 m s$^{-2}$ in the ideal case, and 1200 km s$^{-1}$ and 500 m
362: s$^{-2}$ for the resistive one. On the other hand, it takes 2 to 3
363: hours for the flux rope to be accelerated to the maximum speed.
364: After the maximum, the velocities become more or less constant.
365: The velocities of different parts of the ejecta vary significantly
366: from the cusp point to the rope bottom. For example, at $t=280$
367: minutes the exact moment at which the snapshot is taken for Figure
368: 1, the velocity decreases monotonically from about 800 km s$^{-1}$
369: at the cusp point to 350 km s$^{-1}$ at the rope bottom in the
370: ideal case, and from 1100 km s$^{-1}$ to 750 km s$^{-1}$ in the
371: resistive case. This monotonic decrease of velocities from the
372: leading to trailing edges has been often observed by measurements
373: of flux-rope like structures in the interplanetary space (e.g.,
374: Gosling et al., 1998), which simply indicates that the rope
375: undergoes a rapid expansion during its eruption. The velocity
376: profiles given by both solutions in Figure 2 as well as that in
377: the following figure are in a good agreement with a recent
378: statistical study on the CME accelerations, which indicates that a
379: CME usually undergoes multiphased kinematic evolution including an
380: initial slow rise phase and a main rapid acceleration phase in the
381: inner corona, and a relatively smooth propagation phase in the
382: outer corona (Zhang {\&} Dere, 2006). The magnitude and duration
383: of the main acceleration given by our calculations are also in
384: line with their statistical results. Taking the resistive case
385: shown in Figure 2 as an example, the main acceleration phase
386: starts from $t\approx 70$ minutes and ends at about $t=190$
387: minutes lasting for nearly 2 hours. Comparing the solutions with
388: and without magnetic reconnection, it is apparent that the speeds
389: and accelerations of the flux rope are significantly enhanced in
390: the case involved with magnetic reconnection. Further discussions
391: regarding the roles of magnetic reconnection in the CME dynamics
392: will be given in our discussion section.
393:
394: \subsection{Effect of background field strength}
395: In this subsection, we present numerical results given by
396: calculations with different values of the parameter $B_0$, which
397: represents the strength of the background field and directly
398: relates to the amounts of magnetic energy that can be stored and
399: released in the system. In Figure 3, we plot radial profiles of
400: the velocity and acceleration of different parts of the system
401: including the cusp point (in dotted), and the rope top (in
402: dashed), axis (in solid), and bottom (in dot-dashed) for the cases
403: with (thick lines) and without (thin lines) reconnection, the left
404: panels are for the solution with $B_0=2$ G and the right panels
405: with $B_0=10$ G. Note that for clearness only the accelerations of
406: the cusp point and rope axis are plotted in the lower panels.
407: Similar to the above calculation the catastrophe is triggered by a
408: slight increase of the dimensionless poloidal flux inside the rope
409: $\Psi_p$ from 0.3 to 0.305 with the corresponding critical
410: dimensionless axial flux $\Psi_\varphi$ equal to 0.129 for the
411: case with $B_0=2$ G and 0.244 for $B_0=10$ G. Note that the
412: magnetic fluxes in physical units are listed in Table 1. Since the
413: relative magnitude of the rope poloidal flux does not change with
414: varying $B_0$, the size of the flux rope does not change
415: apparently either. To be quantitatively, we checked the
416: heliocentric distance of the rope axis $r_{a}$ which can be used
417: to represent the size of the flux rope. It was found that
418: $r_a$=1.42, 1.52, and 1.54 solar radii for the cases with $B_0$=2,
419: 6, and 10 G, respectively. As a result, the plasma density inside
420: the flux rope gets slightly different in different cases since the
421: total mass contained by the rope does not change with $B_0$. In
422: Table 1, we also list the two-dimensional magnetic helicity per
423: radian in the azimuthal direction $H_T$ calculated with Equation
424: (9). Similar as the rope fluxes, $H_T$ also increases dramatically
425: with increasing $B_0$. We see that the physical properties of the
426: flux rope differ significantly from case to case in our
427: calculations, which naturally have impacts on the flux rope
428: dynamics. Further discussion along this direction will be
429: presented at the end of the following paragraph.
430:
431: It can be seen from Figure 3 that the most obvious difference
432: between the two sets of solution is the magnitude of velocity and
433: acceleration at different points of the ejecta. In the resistive
434: calculation with $B_0=2$ G, the main acceleration phase of the
435: flux rope lasts for about 4 hours from $t\approx 100$ minutes to
436: $t\approx 350$ minutes with speeds rising up to 670 km s$^{-1}$ at
437: the cusp point and to 460 km s$^{-1}$ at the rope bottom, which is
438: followed by the so-called propagation phase with a nearly constant
439: speed (Zhang {\&} Dere, 2006). In the ideal calculation with the
440: same value of $B_0$, the velocity keeps increasing till $t=600$
441: minutes, yet the corresponding acceleration gets smaller than 5 m
442: s$^{-2}$. In the resistive case with $B_0=10$ G, the main
443: acceleration phase lasts for about 2 hours starting from $t\approx
444: 40$ minutes and ending at about $t=150$ minutes with the maximum
445: acceleration rising up to 750 m s$^{-2}$. After the maximum of
446: velocity is reached, which is about 1600 km s$^{-1}$ for the cusp
447: point and 1150 km s$^{-1}$ for the rope bottom, the flux rope gets
448: decelerated gradually to a velocity of 1300 km s$^{-1}$ at the
449: cusp point and 900 km s$^{-1}$ at the rope bottom. In the ideal
450: case with $B_0=10$ G, the main acceleration phase also lasts for
451: about 2 hours with the maximum acceleration being about 430 m
452: s$^{-2}$ for the cusp point and 220 m s$^{-2}$ for the rope axis.
453: In the propagation phase following the velocity maximum, the
454: velocities decrease slightly. It can be seen that CMEs, even fast
455: ones, can be produced taking the ideal MHD catastrophe as the only
456: process of magnetic energy release. It is also true, again, that
457: the eruptive speeds are significantly enhanced after magnetic
458: reconnection sets in. We point it out in passing that a smooth
459: transition of eruptions from slow to fast can be obtained when
460: varying $B_0$ continuously with a stronger background field
461: corresponding to a faster eruption, in line with the very recent
462: study on the effect of photospheric flux distribution on the flux
463: rope dynamics by Sun et al. (2007). The physical cause of such a
464: behavior can be easily understood from the following simple energy
465: analysis. Since the pattern of the magnetic flux distribution at
466: the coronal base remains the same for all cases we have discussed,
467: the associated open field energy must be proportional to the
468: square of $B_0$. On the other hand, the percentage by which the
469: catastrophic energy threshold exceeds the open field energy varies
470: in a much smaller range. It reads 6.1{\%}, 8.8\% and 9.6\% for
471: $B_0$ = 2 G, 6 G and 10 G, respectively. Such a result is
472: consistent with previous similar calculations (e.g., Hu et al.,
473: 2003b, Li {\&} Hu, 2003; Chen et al., 2006a). Therefore, the total
474: ammount of magnetic free energy of the system at the catastrophic
475: point is mainly determined by the overall strength of the
476: background field in spite of the dramatic differences in the flux
477: rope properties at the catastrophic point as listed in Table 1. In
478: summary, we argue that the stronger the background field is, the
479: more magnetic free energy is available for the flux rope system at
480: the catastrophic point, which leads to a faster eruption of the
481: flux rope.
482:
483: To shed more light on the effect of magnetic reconnection on the
484: CME dynamics, we calculate the total increase in kinetic energy of
485: the system compared with that of the pre-eruption state (i.e., the
486: state at $t=0$), represented by $\Delta E_k$. Figure 4 shows the
487: temporal profiles of $\Delta E_k$ per radian in the azimuthal
488: direction in units of $5.38\times 10^{31}$ ergs for the three sets
489: of solutions with $B_0$=2 G (in solid), 6 G (in dotted), and 10 G
490: (in dashed), where thick and thin lines represent the results with
491: and without magnetic reconnection. The velocity and acceleration
492: for these solutions have been illustrated in Figures 2 and 3. We
493: can see that $\Delta E_k$ tends to reach an asymptotic value in
494: all cases. For each set of solutions, the asymptotic value of
495: $\Delta E_k$ in the resistive case is about 2 to 3 times of that
496: in the ideal case indicating that the MHD catastrophe and magnetic
497: reconnection, the two means of magnetic energy release process,
498: are of comparable importance on the CME acceleration for the
499: resistive MHD situation.
500:
501: \section{Conclusions and discussion}
502: In terms of the catastrophe theory, there are two main processes
503: energizing the solar eruptions: MHD catastrophe and magnetic
504: reconnection. This article serves as a first step to disentangle
505: their contributions to the CME dynamics. To do this, we construct
506: a flux rope catastrophe model in the corona and solar wind and
507: compare different cases in which we either prohibit or allow
508: magnetic reconnection to take place across rapidly-growing current
509: sheets during the eruption. For simplicity, a polytropic process
510: with the polytropic index $\gamma=$1.05 is used to produce the
511: background solar wind solution. The catastrophe and the consequent
512: eruption is triggered by a tiny increase of the rope poloidal
513: flux, which reflects a slight twist of the ends of a long
514: three-dimensional realistic flux rope anchored to the photosphere.
515: It is demonstrated that CMEs, even fast ones, can be produced
516: taking the ideal MHD catastrophe as the only process of magnetic
517: energy release. Nevertheless, the eruptive speed can be
518: significantly enhanced after magnetic reconnection sets in. In
519: addition, a smooth transition from slow to fast eruptions is
520: yielded when increasing the strength of the background magnetic
521: field, i.e., a stronger field, in which more free magnetic energy
522: gets available at the catastrophic point, enables a faster
523: eruption. This suggests that fast and slow CMEs may have an
524: identical driving mechanism.
525:
526: Based on previous and present studies taking catastrophe as the
527: principle driving mechanism of CMEs, we argue that the MHD
528: catastrophe is probably the main means of energy release for CMEs
529: at least in the initial phase. It releases energy without ohmic
530: heating and provide accelerations with the Lorentz force,
531: especially suitable for non-flare associated CMEs. A by-product of
532: the catastrophe is the formation of one or more electric current
533: sheets, which proceeds at the Alfv\'enic time scale and produces
534: conditions favoring fast magnetic reconnection. Such a
535: reconnection, if takes place, further releases the magnetic energy
536: through the following two aspects. Firstly, the magnetic energy is
537: converted into thermal and kinetic energies of plasma particles at
538: the reconnection site. This process is believed to account for a
539: solar flare associated with a CME event. Secondly, the restoring
540: force contributed by the current in the current sheet is
541: significantly reduced or even eliminated, and the magnetic
542: topology changes with the magnetic reconnection. This also
543: produces a significant acceleration of the flux rope in addition
544: to that caused by catastrophe. It can be seen from our
545: quantitative calculation that the MHD catastrophe and magnetic
546: reconnection, the two magnetic energy release processes, may have
547: comparable impacts on the CME dynamics during the main
548: acceleration phase of CMEs.
549:
550: A major subject of this article is to estimate the impact of
551: magnetic reconnection on the flux rope dynamics. For this purpose,
552: we compared solutions given by calculations with and without
553: magnetic reconnection. However, this work suffers from the facts
554: that the anomalous magnetic resistivity is artificially given and
555: the numerical pseudoreconnection is unavoidable in the resistive
556: calculations. Since the effective resistivity including both the
557: anomalous and the numerical ones is believed to be much larger
558: than the realistic value in the corona and solar wind, the
559: velocity profiles given by our study for the case with magnetic
560: reconnection should be taken as the upper bound for the realistic
561: situation. As mentioned in the text, the effective resistivity is
562: a crucial factor determining the transfer rate of magnetic flux
563: from the background to the flux rope. Observationally, this
564: transfer rate can be evaluated by extrapolating the photospheric
565: field to the corona and counting the change of the total magnetic
566: flux in the aftermath of a CME event, e.g., in the coronal dimming
567: region (see, e.g., Jing et al., 2005 and Qiu {\&} Yurchyshyn,
568: 2005). Further theoretical endeavor should utilize these relevant
569: observational constraints on the flux transfer rate in the
570: modelling of a specific event.
571:
572: Our study on the effect of magnetic field strength reveals a
573: smooth transition from slow to fast eruptions when increasing the
574: background field strength. This is in support of the argument that
575: slow and fast CMEs may be driven by a single identical mechanism,
576: also in line with recent statistical results contradicting with
577: the traditional bimodal classification of slow (gradual) and fast
578: (impulsive) CMEs (e.g., Sheeley et al., 1999; Andrews {\&} Howard,
579: 2001). For instance, it is shown by Yurchyshyn et al. (2005) and
580: Zhang {\&} Dere (2006) that the velocity and acceleration of a
581: large amount of CME events have a continuous distribution instead
582: of a bimodal one, and by Vr$\breve{s}$nak et al. (2005) that
583: flare- and nonflare- associated CMEs have quite similar
584: characteristics in the LASCO C2 and C3 fields of view. The
585: velocity profiles for slow CMEs given by our model, say, the
586: solution corresponding to a weak background field, show a gradual
587: acceleration, while that for fast CMEs present a rapid
588: acceleration and a discernable deceleration following the main
589: acceleration phase. This behavior is probably a result of the
590: coupling process with the background solar wind plasma according
591: to our preliminary analysis. When the magnetic energy released in
592: an eruption is not enough to accelerate the flux rope ejecta to
593: the speed of the background plasmas, the ejecta may get gradually
594: accelerated and gain more energy through the coupling to the solar
595: wind. On the other hand, the ejecta may get decelerated and lose
596: energy through similar coupling process with the solar wind while
597: the released magnetic energy is enough to push the flux rope
598: outwards with a velocity faster than that of the background.
599:
600: There exist contradicting discrepancies between the present
601: thick-rope model in axisymmetrical spherical geometry and that
602: published in the literature in terms of thin-rope models in 2-D
603: cartesian geometry. Firstly, an infinite amount of energy is
604: required to open up a closed magnetic field in 2-D Cartesian
605: geometry (Hu et al., 2003b), therefore, it is energetically
606: impossible to open the overlying field and to let the flux rope
607: escape to infinity without magnetic reconnection, as demonstrated
608: by the catastrophe models assuming Cartesian geometry (e.g., Lin
609: {\&} Forbes, 2000). On the other hand, in the spherical geometry
610: the corresponding open-field energy is finite and it can be
611: exceeded by the flux rope system as already shown by many
612: calculations (e.g., Choe {\&} Cheng, 2002; Hu et al., 2003b; Li
613: {\&} Hu, 2003; Flyer et al., 2004; Sun {\&} Hu, 2005; Zhang et
614: al., 2005; Peng {\&} Hu, 2005; Ding {\&} Hu, 2006; Chen et al.,
615: 2006a). Thus, magnetic reconnection may not be necessarily
616: required for the flux rope to get escaped from the Sun in the
617: spherical model. Secondly, as pointed out previously, the
618: self-interaction of the azimuthal current inside the flux rope by
619: itself results in an outward radial force on the rope, which comes
620: from the curvature of the rope surrounding the Sun (Chen, 1989;
621: Lin et al., 1998; Krall et al., 2000; Chen et al., 2006b). This
622: self-force, together with that contributed by its image current
623: below the photosphere, serves as a dominant driving force for the
624: rope eruption. However, in the 2-D Cartesian models this
625: self-force is trivially zero by the symmetry of an infinitely long
626: straight current, this gives another basic difference between the
627: 2-D Cartesian and spherical models.
628:
629: No matter what geometry is used, so far most flux rope models have
630: been limited to 2-D analyses, as a necessary simplification for
631: practical tractability. Yet, the two ends of a 3-D flux rope are
632: believed to be anchored to the solar surface. It remains open
633: regarding how the catastrophic behavior of the flux rope may
634: change under this situation. Finally, since how the corona and
635: solar wind plasma is heated and accelerated is still a big issue
636: to be resolved, the polytropic process is assumed conveniently to
637: obtain the reasonable background solar wind solution in this work.
638: However, it should be noted that the polytropic solar wind
639: solution is too simple to account for some realistic properties of
640: the solar wind. For example, the effect of the fast solar wind is
641: not included in this study. This will certainly affect the
642: propagation of the ejecta at high latitudes, yet may not be very
643: important to the study on the CME propagation along the equatorial
644: plane. In future we consider to employ sophisticated heating
645: functions to produce a more realistic solar wind background (see,
646: e.g., Chen {\&}Hu, 2001; Hu et al., 2003a) for a more elaborated
647: study on the CME propagation in the meridional plane.
648:
649: \acknowledgements
650: This work was supported by grants NNSFC 40404013, NSBRSF
651: G2006CB806304, and NNSFC 10233050 in China.
652:
653: \begin{thebibliography}{}
654: \bibitem[Andrews 2001]{and01}
655: Andrews, M. D., \& Howard, R. S. 2001, Space Sci. Rev., 95, 147
656:
657: \bibitem[Chen (1989)]{Chen89}
658: Chen, J., 1989, ApJ, 338, 453
659:
660: \bibitem[Chen et al. 2006a]{che06_01}
661: Chen, Y., Chen, X. H., \& Hu, Y. Q. 2006a, ApJ, 644, 587
662:
663: \bibitem[Chenhu 2001]{chen01}
664: Chen, Y., {\&} Hu, Y. Q. 2001, Sol. Phys., 199, 371
665:
666: \bibitem[Chen et al. 2006b]{che06_02}
667: Chen, Y., Li, G. Q., \& Hu, Y. Q. 2006b, ApJ, 649, 1093
668:
669: \bibitem[choe \& Cheng (2002)]{choe02}
670: Choe, G. S., \& Cheng C. Z., 2002, ApJ, 574, L179
671:
672: \bibitem[Ding \& Hu 2006]{din06}
673: Ding, J. Y., \& Hu, Y. Q., 2006, Sol. Phys., 199, 371
674:
675: \bibitem[Ding \& Hu 2006]{ding06}
676: Ding, J. Y., Hu, Y. Q., \& Wang, J. X., 2006, Sol. Phys., 235, 223
677:
678: \bibitem[Flyer, Fornberg, Thomas, \& Low (2004)]{Flyer04}
679: Flyer, N., Fornberg, B., Thomas, S., \& Low, B. C., 2004, ApJ,
680: 606, 1210
681:
682: \bibitem[Forbes (2000)]{Forbes00_01}
683: Forbes, T. G., 2000, J. Geophys. Res., 105, 23153
684:
685: \bibitem[Forbes \& Isenberg (1991)]{Forbes91}
686: Forbes, T. G., \& Isenberg, P. A., 1991, ApJ, 373, 294
687:
688: \bibitem[Forbes \& Lin(2000)]{Forbes00_02}
689: Forbes, T. G., \& Lin, J., 2000, J. Atmos. Sol.-Terr. Phys., 62,
690: 1499
691:
692: \bibitem[Forbes \& Priest (1995)]{Forbes95}
693: Forbes, T. G., \& Priest, E. R., 1995, ApJ, 446, 377
694:
695: \bibitem[Gosling 1998]{gos98}
696: Gosling, J. T., Riley, P., McComas, D. J., {\&} Pizzo, V. J.,
697: 1998, J. Geophys. Res., 103, 1941
698:
699: \bibitem[howard 1985]{how85}
700: Howard, R. A., Sheely, N. R., Jr., Koomen, M. J., \& Michels, D.
701: J. 1985, J. Geophys. Res., 90, 8173
702:
703: \bibitem[Hu 1989]{hu89}
704: Hu, Y. Q. 1989, J. Comput. Phys., 84, 441
705:
706: \bibitem[Hu 2005]{hu05}
707: Hu, Y. Q. 2005, in IAU Symp. 266, Coronal and Stellar Mass
708: Ejections, ed. K. Dere, J. Wang, \& Y. Yan (Cambridge: Cambridge
709: Univ. press), 263
710:
711: \bibitem[hu 2003a]{hu03}
712: Hu, Y. Q., Habbal, S. R., Chen, Y., {\&} Li, X., 2003a, 108(A10),
713: 1377, doi:10.1029/2002JA009776
714:
715: \bibitem[Hu et al. 2003b]{hue03}
716: Hu, Y. Q., Li, G. Q., \& Xing X. Y. 2003b, J. Geophys. Res.,
717: 108(A2), 1072, doi:10.1029/2002JA009419
718:
719: \bibitem[Hu et al. 1997]{hux97}
720: Hu Y. Q., Xia L. D., Li X, Wang J. X., \& Ai G. X., 1997, Sol.
721: Phys., 170, 283
722:
723: \bibitem[Isenberg (1993)]{isenberg93}
724: Isenberg, P. A., Forbes, T. G., \& Demoulin, P., 1993, ApJ, 417,
725: 368
726:
727: \bibitem[Jing (2005)]{jing05}
728: Jing, J., Qiu, J., Lin, J., Qu, M., Xu, Y., \& Wang, H., 2005,
729: ApJ, 620, 1085
730:
731: \bibitem[Krall \& Chen (2000)]{krall00}
732: Krall, J., Chen, J., \& Santoro, R., 2000, ApJ, 539, 964
733:
734: \bibitem[Li \& Hu 2003]{li03}
735: Li, G. Q., \& Hu, Y. Q. 2003, Chinese J. Astron. Astrophys., 3, 555
736:
737: \bibitem[Lin \& Forbes 2000]{lin00}
738: Lin, J., \& Forbes, T. G. 2000, J. Geophys. Res., 105(A2), 2375
739:
740: \bibitem[Lin (1998)]{lin98}
741: Lin, J. Forbes, T. G., Isenberg, P. A., \& Demoulin, P., 1998,
742: ApJ, 504, 1006
743:
744: \bibitem[Lin et al. 2003]{lin03}
745: Lin, J., Soon, W., \& Baliunas, S. L. 2003, NewA Rev., 47, 53
746:
747: \bibitem[Low (2001)]{loe01}
748: Low, B. C., 2001, J. Geophys. Res., 106, 25141
749:
750: \bibitem[Peng \& Hu 2005]{pen05}
751: Peng, ZH., \& Hu, Y. Q. 2005, Chinese J. Space Sci., 25, 81
752:
753: \bibitem[Qiu (2005)]{Qiu05}
754: Qiu, J., \& Yurchyshyn, V. B., 2005, ApJ, 634, L121
755:
756: \bibitem[Sheeley 1999]{She99}
757: Sheeley, N. R., Jr., Walters, H., Wang, Y.-M., \& Howard, R. A.
758: 1999, J. Geophys. Res., 104, 24739
759:
760: \bibitem[Sun \& Hu 2000]{sun05}
761: Sun, S. J., \& Hu, Y. Q. 2005, J. Geophys. Res., 110, A05102,
762: doi:10.1029/2004JA010905
763:
764: \bibitem[Sunhu 2007]{sun07}
765: Sun, S. J., Hu, Y. Q., \& Chen, Y., 2007, ApJ, 654, L167
766:
767: \bibitem[vrs 2005]{vrs05}
768: Vr$\check{s}$nak, B., Sudar, D., \& Ruzdjak, D. 2005, A\&A, 435,
769: 1149
770:
771: \bibitem[yurchyshyn 2005]{yur05}
772: Yurchyshyn, V., Yashiro, S., Abramenko, V., Wang, H., \&
773: Gopalswamy, N. 2005, ApJ, 619, 599
774:
775: \bibitem[Zhang \& Dere 2006]{zha04}
776: Zhang, J., \& Dere, K. P. 2006, ApJ, 649, 1100
777:
778: \bibitem[Zhang et al. 2005]{zha05}
779: Zhang, Y. Z., Hu, Y. Q., \& Wang, J. X. 2005, ApJ, 626, 1096
780:
781: \end{thebibliography}
782:
783: Table 1.\hspace{1em}The poloidal and toroidal fluxes ($\Psi_{pc}$
784: and $\Psi_{\varphi c}$), and the two-dimensional magnetic helicity
785: $H_T$ (per radian in the azimuthal direction) of the flux rope
786: system at the catastrophic point in different cases. \vskip 10pt
787:
788: \begin{center}
789: \begin{tabular}{lccc}
790: $B_0$ (G) & 2 & 6 & 10 \\
791: \hline\hline
792: $\Psi_{pc}$ (Wb) & 2.91 $\times 10^{13}$ & 8.70 $\times 10^{13}$ & 1.45 $\times 10^{14}$\\
793: $\Psi_{\varphi c}$ (Wb) & 1.25 $\times 10^{13}$ & 6.06 $\times 10^{13}$ & 1.08 $\times 10^{14}$\\
794: $H_T$ (Wb$^2$) & 1.45 $\times 10^{27}$ & 2.12 $\times 10^{28}$ & 6.32 $\times 10^{28}$ \\
795: \hline
796: \end{tabular}
797: \end{center}
798:
799: \begin{figure}
800: \epsscale{1.} \plotone{f1.eps} \caption{The magnetic topology
801: (white lines) and velocity color contour map for $B_0=6$ G from 1
802: $R_\odot$ to 20 $R_\odot$ for: (a) the background corona and solar
803: wind solution before the emergence of the flux rope, (b) the
804: pre-eruption state of the flux-rope system in the solar wind
805: background, (c) the solution with an erupting flux rope at $t=280$
806: minutes without magnetic reconnection, and (d) the eruptive
807: solution with magnetic reconnection at the same instant as panel
808: (c). The outer boundary of the original flux rope is depicted with
809: a green circle. \label{fig1}}
810: \end{figure}
811: \begin{figure}
812: \epsscale{1.} \plotone{f2.eps} \caption{The temporal profiles of
813: the heliocentric distance, velocity, and acceleration of different
814: parts of the flux rope system, including the cusp point (in
815: dotted), the rope top (in dashed), the rope axis (in solid), and
816: the rope bottom (in dot-dashed). Left panels are for the case
817: without reconnection and right for the reconnection case.
818: \label{fig2}}
819: \end{figure}
820: \begin{figure}
821: \epsscale{1.} \plotone{f3.eps} \caption{The radial profiles of the
822: velocity and acceleration of different parts of the system
823: including the cusp point (in dotted), and the rope top (in
824: dashed), axis (in solid), and bottom (in dot-dashed), the left
825: panels are for the case with $B_0=2$ G and the right panels with
826: $B_0=10$ G for the cases with (thick lines) and without (thin
827: lines) reconnection. \label{fig3}}
828: \end{figure}
829: \begin{figure}
830: \epsscale{1.} \plotone{f4.eps} \caption{The temporal profiles of
831: the total increase in kinetic energy of the system $\Delta E_k$ in
832: units of $5.38\times 10^{31}$ ergs (per radian in the azimuthal
833: direction) for three sets of solutions ($B_0$=2 G (in solid), 6 G
834: (in dotted), and 10 G (in dashed)) with (thick lines) and without
835: (thin lines) magnetic reconnection. \label{fig4}}
836: \end{figure}
837:
838: \end{document}
839: