1: \documentclass[12pt,preprint]{aastex}
2:
3: \shorttitle{Are ``EIT Waves'' Fast-Mode MHD Waves?}
4: \shortauthors{Wills-Davey et al.}
5:
6: \begin{document}
7:
8: \title{Are ``EIT Waves'' Fast-Mode MHD Waves?}
9:
10: \author{M. J. Wills-Davey, C. E. DeForest, and J. O.
11: Stenflo\altaffilmark{1}}
12: \affil{Department of Space Studies, Southwest Research Institute,
13: Boulder, CO 80302}
14:
15: \altaffiltext{1}{On leave from the Institute of Astronomy, ETH
16: Zurich}
17:
18:
19: \begin{abstract}
20: We examine the nature of large-scale, coronal,
21: propagating wave fronts (``EIT waves'') and find they are incongruous with
22: solutions using fast-mode MHD plane-wave theory. Specifically, we consider
23: the following properties: non-dispersive single pulse manifestions, observed
24: velocities below the local Alfv\'en speed,
25: and different pulses which travel at any number of constant velocities,
26: rather than at the ``predicted'' fast-mode speed. We
27: discuss the possibility of a soliton-like explanation for these phenomena,
28: and show how it is consistent with the above-mentioned aspects.
29: \end{abstract}
30:
31: \keywords{waves---MHD---Sun: corona, CMEs}
32:
33: \clearpage
34:
35: \section{Introduction}
36: \label{sec:introduction}
37:
38: Long before the availability of direct observations
39: in 1997 \citep{Thompson1998}, attempts were
40: made to explain the physics of large-scale coronal pulse waves.
41: The original evidence of these wave fronts appeared in chromospheric
42: hydrogen-$\alpha$ observations of ``Moreton waves''---semi-circular
43: propagating depressions, which traveled away from flaring regions at
44: speeds orders of magnitude above the
45: chromospheric sound speed \citep{Athay_Moreton1961}. \citet{Uchida1968}
46: theorized that Moreton waves were a secondary effect caused by the
47: ``skirt'' of a coronal fast-mode magnetoacoustic
48: shock wave extending down into the chromosphere. They manifest themselves
49: in running difference images as dark fronts followed by light fronts, as shown
50: in Figure~\ref{fig:moreton}.
51:
52: The advent of continuous soft x-ray and EUV observation---instruments
53: such as {\it Yohkoh}-Soft X-ray Telescope (SXT) and the {\it SOHO} Extreme
54: Ultraviolet Imaging Telescope (EIT)---made it possible to test
55: the \citet{Uchida1968}
56: theory, and indeed all manner of large-scale
57: coronal pulse waves have been observed. Wave fronts have been recorded
58: in soft x-ray \citep{Hudson2003,Warmuth2005,Khan_Aurass2002},
59: EUV \citep{Thompson1998,WD_Thompson,Biesecker},
60: and even as a secondary response in
61: He~10830\AA\ \citep{Gilbert_etal}.
62: Moreton waves have some
63: cospatiality with EUV waves \citep{Eto,Okamoto_etal}
64: and have been particularly
65: well-correlated with soft x-ray observations
66: \citep{Narukage2002,Narukage2004},
67: lending credence to
68: the original \citet{Uchida1968} postulation.
69:
70: However, Moreton waves are observed in conjunction with only a tiny
71: fraction of coronal observations. The large majority
72: of single-pulse wave fronts
73: are seen only by EUV instruments, with no apparent chromospheric or
74: soft x-ray counterpart. These EUV fronts (often called ``EIT waves'')
75: have some of the same general characteristics as Moreton
76: waves, but in many respects they are quite different. While it would
77: appear that Moreton waves may fit the \citet{Uchida1968} fast-mode
78: MHD shock model, we postulate that existing MHD models of EIT waves
79: are not consistent with aspects of available data, and suggest that mechanisms
80: which encompass nonlinear wave pulse propagation appear more promising to
81: explain the breadth of observed EIT wave phenomena.
82:
83: \subsection{Properties of Moreton and EIT Waves}
84: \label{sec:properties}
85:
86: Moreton waves and EIT waves can be described as ``single-pulse'' phenomena.
87: Figure~\ref{fig:waves} shows examples of two different EIT wave events---
88: one observed by {\it SOHO}-EIT, and one by the Transition Region and
89: Coronal Explorer ({\it TRACE}).
90: These waves are associated with impulsive events,
91: and although actual causality has still not been determined,
92: \citet{Biesecker} and \citet{Cliver2005} find a strong correlation
93: with CME initiation.
94: There is also evidence that both EIT waves and Moreton waves displace
95: large magnetic structures: EIT waves have been observed
96: directly instigating loop ocillations
97: \citep{WD_Thompson},
98: and Moreton waves have been
99: associated with ``winking filaments''
100: \citep{Okamoto_etal}.
101:
102: However, other aspects of EIT and Moreton waves are sufficiently different
103: that some have theorized they are two entirely different populations, which
104: originate from different instigators \citep{Chen_etal,Eto,Chen_etal2005}.
105: Although both are single pulses, Moreton waves are strongly-defined,
106: narrow, semi-circular fronts, while EIT waves are broad ($\sim 100$~Mm),
107: extremely diffuse, and (when unimpeded) produce circular wave fronts.
108: Moreton waves
109: have relatively short lifetimes (usually $< 10$ minutes), and have shown
110: cospatial observational signatures between the chromosphere and the
111: soft x-ray corona \citep{Khan_Aurass2002,Narukage2002}.
112: EIT waves are primarily visible in the lower corona (at 1-2~MK),
113: but typically have lifetimes of over an hour and can travel the entire diameter
114: of the Sun while remaining coherent \citep{Myers}.
115: Moreton waves typically travel at speeds of $\sim 400$-2000~km/s
116: \citep{Becker,Smith_Harvey1971}; such
117: velocities are thought to be comparable to or much larger than the local
118: Alfv\'en speed. In recent work,
119: \citet{Narukage2004}---having
120: calculated local fast magnetoacoustic speeds
121: of 700-1000~km/s---find that Moreton waves occur at speeds of
122: $M > 1$, and disappear as they slow to $M=1$.
123: EIT waves, on the other hand, travel much more slowly, at
124: average velocities ranging
125: 25-450~km/s \citep{Myers}, which correspond to $0.03 < M < 0.53$.
126: Although there is some evidence of Moreton and EIT waves
127: traveling cospatially \citep{Thompson2000,Okamoto_etal}, most
128: studies conclude that, while they appear to originate together,
129: the two must be inherently
130: different \citep{Chen_etal,Eto,Chen_etal2005}.
131:
132: The work of \citet{Uchida1968} and \citet{Narukage2004}
133: would appear to
134: explain the nature of Moreton waves---they exist as a result of coronal
135: shock fronts. EIT waves, however, have proved much more difficult to
136: comprehend. Though Moreton waves are always viewed in conjunction with
137: EIT waves, the converse is not true, even in high-cadence data.
138: \nocite{Wills-Davey2006, Wills-Davey2002} Wills-Davey(2002,2006)
139: present quantitative analysis of a {\it TRACE}-observed
140: EIT wave from its inception, and no corresponding Moreton wave is
141: observed. \footnote{This contradicts the findings of
142: \citet{Harra_Sterling2003},
143: but their conclusions about the same
144: wave front are the result of visual inspection,
145: whereas the work of Wills-Davey(2002,2006) is quantitative.}
146:
147: Any complete theory of EIT waves must explain:
148: \begin{itemize}
149: \item why EIT waves are observed as single pulses
150: \item how most EIT waves are manifested in the absence of
151: Moreton waves,
152: \item why many EIT wave velocities are slower than predicted
153: Alfv\'en speeds,
154: \item why individual EIT waves travel at approximately constant speed,
155: but that speed varies greatly between EIT waves, and
156: \item how EIT waves can maintain coherence over distances comparable
157: to the solar diameter;
158: \item additionally, it should confirm why EIT waves sometimes
159: generate loop oscillations.
160: \end{itemize}
161:
162: \subsection{Existing Coronal Pulse Wave Models}
163: \label{sec:existing_models}
164:
165: At present, multiple published explanations exist
166: to explain EIT waves
167: \citep{Chen_etal,Warmuth,Chen_etal2005,Wang,Wu_etal,
168: Ofman_Thompson2002,Ofman2007}.
169: In each case, some of the requirements listed in \S~\ref{sec:properties}
170: are fulfilled, but no one numerical or
171: theoretical model explains all six properties.
172:
173: \citet{Chen_etal}, \citet{Warmuth},
174: and \citet{Chen_etal2005}
175: each develop models that focus on the {\it relation} between
176: the Moreton and EIT waves, leading to explanations
177: where the Moreton wave is the primary source of a
178: secondary EIT wave---a scenario which is inconsistent with the bulk of
179: ``EIT wave only'' observations. Additionally,
180: the models created by \citet{Chen_etal} and \citet{Chen_etal2005}
181: demonstrate Moreton wave propagation over large distances but EIT wave
182: propagation over much smaller distances---the opposite of what is seen
183: in observations.
184:
185: In cases where ``EIT wave only'' numerical simulations have been developed,
186: the physics driving EIT waves is derived from the original
187: \citet{Uchida1968} theory:
188: EIT waves are treated as fast-mode magnetohydrodynamic (MHD) waves.
189: Starting from this premise,
190: \citet{Wang}, \citet{Wu_etal}, \citet{Ofman_Thompson2002}, and
191: \citet{Ofman2007} have created
192: computer-generated coronal waves that imitate data very closely.
193: Unfortunately, the successful implementation of these models requires
194: unusually low quiet sun magnetic field strengths as well as a
195: high plasma-$\beta$
196: corona. Additionally, both the \citet{Wang} and the
197: \citet{Wu_etal}
198: simulations reproduce only the same oft-studied event from May 1997.
199: The wide variety of EIT wave velocities and morphologies
200: may be beyond the capability of these models; indeed, when \citet{Wang}
201: models a second event from April 1997, he reproduces the velocities of the
202: May 1997 rather than the April 1997 wave.
203:
204: The problem may lie in the treatment of EIT waves as fast-mode
205: MHD pulses. While a fast-mode MHD wave does have some of the properties
206: associated with EIT waves, many aspects of these coronal pulse waves
207: contradict predicted fast-mode behavior. Additionally,
208: \citet{Wills-DaveyPhD} and \citet{Warmuth2004II}
209: have found observational evidence that
210: these waves are highly non-linear,
211: with density perturbations of 40\% to more than
212: 100\% above the local background.
213:
214: In this paper, we discuss the discrepancies
215: between the predicted behavior of MHD waves and EUV observations,
216: and consider the ramifications of the MHD solution on
217: other aspects of coronal physics (\S~\ref{sec:plane_wave}). With these
218: discrepancies in mind, we show that aspects of
219: a single-pulse solution can account for the properties of
220: EIT waves (\S~\ref{sec:resolution}).
221:
222: \section{Inconsistencies Arising from a Fast-Mode MHD Solution}
223: \label{sec:plane_wave}
224:
225: At first glance, the choice of a fast-mode MHD solution seems the
226: most appropriate to explain EIT waves.
227: Fast- and slow-mode MHD wave mode speeds can be written as
228: \begin{equation}
229: \label{eqn:fm_sm}
230: {v_{f,s}}^2 = \frac{1}{2}[ {v_A}^2 + {c_s}^2 \pm
231: \sqrt{{v_A}^4 + {c_s}^4 - 2{c_s}^2{v_A}^2 \cos{2\theta}}]
232: \end{equation}
233: where
234: ${v_A}^2 = {B^2}/(4{\pi}{\rho})$ defines the local Alfv\'{e}n speed, and
235: ${c_s}^2 = ({\gamma}{k_B}T)/m$ the local sound speed. Note that
236: $v_f \ge v_A$ and $0 \le v_s \le c_s$, depending on $\theta$. This means
237: that any event with a speed below $v_A$ cannot be considered a fast
238: magnetosonic wave.
239:
240: To reproduce EUV observations, the chosen wave solution
241: must be a compressive MHD wave that can travel
242: ubiquitously through a magnetized plasma.
243: Pure Alfv\'en waves
244: cannot produce the necessary compression to be seen as a brightness
245: enhancement. Slow-mode magnetoacoustic waves
246: are compressive, but their propagation is limited by magnetic field
247: direction; the slow-mode velocity vanishes for propagation perpendicular to
248: field lines. Not only would this prevent the observed ubiquitous
249: propagation through the quiet corona, but
250: {\it TRACE} observations show evidence of
251: EIT waves successfully crossing coronal loop structures \citep{WD_Thompson}.
252:
253: However, fast-mode MHD waves have the double advantage of being compressive
254: and existing for all magnetic field orientations. These are the properties
255: that led \citet{Uchida1968} to use fast-mode MHD shocks for
256: his original Moreton wave solution, and motivate their continued use
257: in more recent simulations.
258: Unfortunately, a fast-mode solution presents problems when trying
259: to recreate some of the properties of EIT waves. In particular,
260: it becomes difficult to explain:
261: \begin{itemize}
262: \item observed speeds,
263: \item theoretical assumptions of a low-$\beta$ corona (due the fact that
264: many EIT waves travel slower than the local sound speed),
265: \item the variety of observed propagation speeds, and
266: \item the nature and duration of pulse coherence.
267: \end{itemize}
268:
269: \subsection{Velocity Magnitudes}
270:
271: One inconsistency between a fast-mode
272: MHD wave model and observed EIT wave behavior
273: concerns EIT wave speed magnitudes.
274: Magnetic field orientation constrains a fast-mode MHD wave to a velocity
275: range $v_A \le v_{fm} \le ({v_A}^2 + {c_s}^2)^{\frac{1}{2}}$.
276: Previous studies of EIT and Moreton waves have defined initial conditions
277: such that the resultant local Alfv\'en or fast-mode
278: speed is also the EIT wave speed as observed for a particular event
279: in the data \citep{Wang,Wu_etal};
280: therefore, since EIT waves have been observed
281: at any number of speeds \citep{Myers}, we
282: take myriad existing work into account and
283: determine as large a range of quiet sun fast-mode speeds as we can from the
284: data.
285:
286: Various studies have determined plasma conditions for the
287: base of the quiet corona. Magnetic field strengths
288: have been measured at anywhere from 2.2~G \citep{Fludra_etal,Falconer_Davila}
289: to 10~G \citep{Pauluhn_Solanki}, while multiple studies have
290: found density measurements close to $2 \times 10^8~\rm{cm}^{-3}$
291: \citep{Aschwanden_Acton,Feldman_etal,Doschek_etal}.
292:
293: Such findings lead to a wide range of possible Alfv\'{e}n speeds.
294: \citet{Gopalswamy2002} assume a magnetic field
295: strength of 2.2~G and a density of $5 \times 10^8~\rm{cm}^{-3}$,
296: resulting in $v_A = 215$~km/s and
297: $v_{fm} = 230$~km/s at the base of the quiet corona, where $v_{fm}$ is the
298: fast-mode speed perpendicular to the magnetic field. We consider
299: the \citet{Gopalswamy2002} velocities
300: a lower bound for fast-mode speeds. By taking
301: the highest measured field strength (10~G) and the
302: most predominantly measured
303: density ($2 \times 10^8~\rm{cm}^{-3}$), we find that the Alfv\'{e}n
304: speed in the quiet Sun can reasonably extend as high as 1500~km/s.
305: These values provide us with a (rather broad) range of possible
306: quiet sun Alfv\'en speeds.
307:
308: Since the minimum fast-mode speed is constrained by the Alfv\'en speed,
309: any EIT wave must travel faster than $v_A$ for a fast-mode MHD solution to
310: be valid. Until now, most studies have considered sample sets weighted
311: towards faster waves (Gopalswamy \& Kaiser(2002); Warmuth et al.(2004a);
312: Narukage et al.(2005); etc.)
313: \nocite{Gopalswamy2002,Warmuth2004I,Narukage2005}
314: because they have focused on events correlated
315: with shocked Moreton waves; in such samples
316: (with mean velocities of $\sim 200-400$~km/s) problems with the
317: fast-mode velocity are not as readily apparent.
318:
319: For comparison, Figure~\ref{fig:MT_plot} shows all the mean velocities
320: recorded by \citet{Myers}; of the 175 EIT waves
321: occurring between 25 March 1997 and 16 June 1998, 160 were observed in
322: multiple frames.
323: Only a small fraction of the \citet{Myers} EIT waves have
324: average speeds close to 300~km/s; the large majority are noticeably slower.
325: The velocities shown in Figure~\ref{fig:MT_plot} are inconsistent with
326: a minimum Alfv\'en speed of 215~km/s.
327: 101 of the 160 observed events have average speeds below
328: our minimum $v_A$. Such a large discrepancy suggests that some
329: physical assumption is incorrect.
330:
331: A lower $v_A$ would be possible if
332: we found the measured values for $\mathbf{B}$ were too high and/or
333: the measured densities too low.
334: The \citet{Lin_Kuhn} direct measurements of coronal magnetic field find
335: 4~G 75~Mm above an active region; presumably field strengths are lower
336: in the quiet corona through which the waves propagate.
337: However, since the density falls off as
338: $\sim e^{(-z / \Lambda)}$, where $\Lambda$ is the pressure scale height,
339: we actually expect the Alfv\'en speed to increase with altitude.
340: Using extrapolation methods, recent studies have also
341: found ``true'' quiet sun magnetic fields in the range of
342: 20-40~G \citep{Krivova_Solanki,Cerdena_etal}; these values
343: would increase calculated Alfv\'en speeds
344: by as much as an order of magnitude.
345: Alternatively, the problem could lie with
346: the assumption of a fast-mode solution.
347:
348: It may be possible that Figure~\ref{fig:MT_plot} actually shows a superposition
349: of two different types of wave events; there is a slight visual break at around
350: $\sim 260$~km/s, suggesting we may be observing clustering of two populations.
351: If this is the case, it is possible the higher speed events (26 of 160) may
352: be consistent with fast magnetosonic wave simulations (see, for example,
353: Wang (2000); Wu et al. (2001); Ofman \& Thompson (2002); Ofman (2007)).
354: However, the slower events would still need a separate explanation.
355: \nocite{Ofman2007}
356:
357: \subsection{Requirements of a Coronal Plasma}
358:
359: To further emphasize the potential problems with treating EIT waves as
360: fast-mode waves, we consider the requirement that $v_{fm} \ge v_A$.
361: By assuming that EIT waves (as fast-mode waves) travel at
362: $v_{fm}$, we set an upper bound
363: for $v_A$. According to Figure~\ref{fig:MT_plot}, this would
364: give us Alfv\'en speeds ranging
365: $27~{\rm km/s} \le v_A \le 438~{\rm km/s}$.
366:
367: We can determine the validity of these possible Alfv\'en speeds
368: by considering them in the context of plasma-$\beta$. $\beta$
369: describes the ratio of gas pressure to magnetic pressure, and is
370: often written $\beta = (8{\pi}p)/B^2$. This also means that
371: \begin{equation}
372: \beta \sim \frac{{c_s}^2}{{v_A}^2}.
373: \end{equation}
374: with a difference of a factor $2/{\gamma}$ (where
375: $\gamma$ is the ratio of specific heats), which is of order unity.
376: Most EIT waves are observed in the 195~\AA\ passband, which
377: is most sensitive to plasma at
378: $\sim 1.5$~MK. We take
379: the sound speed at this temperature (185~km/s) as a
380: reasonable value for $c_s$.
381: Using the average velocities shown in Figure~\ref{fig:MT_plot}
382: to define $v_A$, we find that the plasma-$\beta$
383: associated with EIT waves can extend from
384: $\beta \sim 0.20$ to as much as $\beta \sim 50$ using these values of
385: $c_s$ and $v_A$.
386:
387: It has become widely accepted that coronal morphology is magnetically
388: dominated, and is often approximated by a force-free field.
389: By definition, $\beta$ must be small in a
390: magnetically dominated plasma, and only for $\beta \ll 1$
391: is a force-free field model reasonable;
392: Some recent work has discussed the possibility of coronal
393: plasma-$\beta$s close to unity \citep{Gary,Aschwanden1999}, but
394: these measurements have typically been taken above active regions
395: and are assumed to apply to current-filled loops. In any case,
396: it is rare to find a theoretical $\beta$ much larger than unity.
397:
398: Some fully-three-dimensional Quiet Sun models
399: (such as that of \citet{Wu_etal}) have implied $\beta$ can be as high as
400: $5 \ge \beta \ge 50$ in active region latitudes, over +/-$30^\circ$
401: (see Figure 2 of
402: \citet{Wu_etal}). While it is true that we have no direct measurements for
403: values of density or magnetic field in the corona, under such conditions (and
404: over such an extended area), the morphology of the corona in EUV images
405: demonstrates that the $\beta$ parameter must be low; magnetic structures
406: dominate everywhere.
407:
408: Until observations imply that there are extensive (of order several hundred Mm)
409: areas of the corona with such high $\beta$, we will instead be swayed by
410: existing coronal limb observations.
411: If the $\beta$s found using the \citet{Myers}
412: data are valid, then large
413: portions of the quiet sun cannot be magnetically-dominated.
414: Such large possible $\beta$ values either contradict
415: the validity of the corona as
416: a low-$\beta$ plasma, or offer additional evidence that EIT waves
417: cannot be modeled using fast-mode waves.
418:
419: \subsection{Propagation Speed Differences}
420:
421: The fact that a broad range of speeds is observed at all should
422: cause us to question the validity of fast-mode waves
423: as an EIT wave solution. In a linear
424: regime, the wave speed corresponds to the reaction speed of the medium;
425: wave velocities are directly correlated to observable properties such as
426: density or magnetic field strength. Observations of EIT waves show that
427: pulses maintain coherence over global distances.
428: This lack of decoherence suggests that
429: the plasma properties of the quiet corona
430: are often uniform. If EIT waves were actually fast-mode MHD waves,
431: this underlying global sameness would constrain EIT waves to a narrow
432: range of velocities close to the expected fast-mode speed. The
433: simulations of \citet{Wang}
434: and \citet{Wu_etal}, which propagate
435: fast-mode waves through mildly-structured quiet corona, produce just
436: this type of result; \citet{Wang}
437: finds that even quiet sun changes
438: over time are not large enough to substantially affect the fast-mode speed.
439:
440: While EIT observations lack the temporal cadence to show if EIT waves
441: travel at constant speed, the range of speeds found by
442: \citet{Myers}
443: makes it difficult to justify the existence of a
444: ``preferred'' EIT wave speed. Additionally, the \citet{Myers} data
445: show strong evidence of
446: waves with different speeds traveling through the same region of quiet
447: sun in the space
448: of a few hours. In the case of one particularly productive active region,
449: seven waves were produced over a 36-hour period (from 1-May to 3-May 1998)
450: with speeds of 85 to 435~km/s, a difference of a factor
451: of 5. In each case, the wave traveled a distance of $\sim 1~R_{\odot}$
452: through the same general area of quiet Sun.
453:
454: Explaining each of these wave fronts as fast-modes would require
455: that the quiet sun fast-mode speed change globally on time-scales
456: shorter than a few hours. Since EIT waves are strongly associated with
457: CMEs, it may be that CMEs corresponding to EIT waves
458: produce large-scale topology changes which then affect the global
459: fast-mode speed. However, the lack of global changes shown by difference
460: images suggests that this is unlikely.
461:
462: \subsection{Pulse Coherence}
463: \label{sec:pulse}
464:
465: Morphologically, EIT waves appear as single-pulse fronts.
466: To date, there has only been one observation of a pulse wave
467: (in this case, a Moreton wave) that appears to include multiple fronts,
468: related to the X10 flare of 29 October 2003 \citep{Neidig2004};
469: unfortunately, no contemporaneous EUV data exist.
470:
471: Numerical simulations have shown that a fast-mode MHD solution can
472: generate a wave packet comparable to EUV observations.
473: A wave packet of fast-mode MHD waves can produce a single-pulse front;
474: however, the differing phase speeds within the packet
475: would leave it highly susceptible to
476: dispersion resulting from conditions such as density stratification
477: and magnetic field variations. If the scale of the fluctuations is
478: much smaller than the wavelength (as in the case of magnetic
479: field loops), these fluctuations will not affect the coherence of the
480: wave; however, the pulse width appears to be about a scale height
481: \citep{Wills-DaveyPhD}, allowing for significant effects due to density
482: variations. Since the dispersing medium is ubiquitous,
483: the packet would begin to break apart almost immediately, and
484: would appear as periodic ``ripples'' on either side of the main front.
485:
486: Such immediate dispersion effects appear difficult to reconcile with
487: observations of single coherent fronts propagating over global distances.
488: The lack of temporal resolution in the {\it SOHO}-EIT data may account for
489: the lack of any observed periodicity (perhaps visible as multiple fronts)
490: as the front widens and the amplitude decreases. However,
491: multiple {\it TRACE}
492: observations, despite a much higher cadence, have also failed to
493: reveal any obvious periodicity in a wave as it decays.
494:
495: Quantitative measurements of the 13 June 1998 EIT wave
496: (Figure~\ref{fig:waves}(b)) show that the
497: density enhancement cross-section maintains coherence for some
498: time (of the order of tens of minutes), and will even break apart slightly
499: and re-form in a ``pulse'' shape as it encounters different coronal structures
500: \citep{Wills-Davey2006}.
501: In this quantifiable case, the wave amplitude decreases over
502: time---in a manner consistent with radial expansion---but
503: there is no measurable increase in the pulse FWHM \citep{Wills-DaveyPhD}.
504: Additionally, no ``ripples''
505: appear around the main pulse as this occurs.
506: To the extent that the pulse was measurable
507: (before it became indistinguishable from noise),
508: the data appear to be consistent with a wave
509: propagating dispersionlessly.
510:
511: This lack of dispersion
512: also appears consistent with the wavelet analysis performed by
513: \citet{Ballai2005}.
514: Over the length of the entire 13 June 1998 data set,
515: their results show a roughly constant wavelet power spectrum band
516: ranging from 285 to 560 seconds. They interpret this as a strong signal
517: with an intensity period $\sim 400$~seconds that {\it does not degrade}.
518: While their results do not shed light on
519: the nature of the pulse-like structure
520: of the front, they do appear to confirm that the wave packet remains
521: intact throughout their measurements. Given some of the interference
522: seen in the \citet{Wills-Davey2006} cross-sections,
523: the \citet{Ballai2005} findings would suggest that
524: the pulse is unusually stable to perturbations, and does not suffer from
525: the dispersion expected for a linear fast-mode wave.
526:
527: \section{Resolving EIT Wave Inconsistencies}
528: \label{sec:resolution}
529:
530: EIT wave velocities have presented two key problems: the speeds are too
531: slow for a significant number of observations
532: to be explained using fast-mode MHD waves, and the plasma properties
533: of a largely uniform quiet corona should not lead to such a wide range
534: of constant wave speeds. However, if we instead understand EIT waves as
535: a type of coronal MHD soliton---perhaps a 2-D slow-mode soliton---the
536: velocity range becomes easier to explain.
537:
538: One key difference between plane wave and
539: soliton solutions is the velocity dependence. With a linear MHD solution,
540: wave speed is determined solely by properties of the transmission medium.
541: Soliton speed is additionally dependent on the amplitude of the
542: pulse. In the case of MHD solitons,
543: speed varies as a function of density enhancement
544: \citep{Buti,Ballai2003}.
545:
546: Consider the velocity dependence shown in
547: Figure~\ref{fig:MT_plot}. Although the ``Quality Rating'' is a
548: visually-determined observer-dependent ranking system, the data from
549: \citet{Myers} still show that well-defined
550: (more density-enhanced) waves to travel faster.
551: Speeds only approach, but do not reach,
552: the Alfv\'en speed $v_A$; \citet{Narukage2004}
553: show that large-scale pulse waves traveling at or above $v_A$
554: shock and appear as Moreton waves. The velocity-density enhancement
555: dependence also allows for events of different speeds to pass
556: through the same region of quiet sun without requiring global
557: restructuring.
558:
559: In addition to solving the velocity discrepancies, a soliton explanation
560: also provides some of the pulse stability and coherence needed to
561: explain the properties of EIT waves. Because the stability of a soliton
562: is dependent both on both nonlinearity in the pulse and dispersion in the
563: local medium, solitons are stable to small perturbations, allowing
564: them to travel through thin cross-wise loop structures
565: and over large distances of quiet sun.
566: As solitons are nondispersive, this explanation would
567: also consider the lack of dispersion observed in
568: strong events such as the 13 June 1998 event (Figure~\ref{fig:waves}(b)).
569:
570: While it is true that the slow-mode experiences greater dissipation than
571: other modes, because of their large width, EIT waves only need to remain
572: coherent over $\sim 10$ wavelengths to display typical behavior. If we
573: consider the work of \citet{Ofman1999},
574: who looked at slow MHD waves
575: in polar plumes, they found that pulses maintained coherence for a minimum of
576: three wavelengths, and showed the sort of steepening that would be
577: counteracted by nonlinearity, in the case of soliton behavior. This suggests
578: that coherence over ten wavelengths is not unreasonable.
579:
580: Of course, the dynamics observed in EIT waves
581: could not be the same as those seen by
582: \citet{Ofman1999}. Rather, since these wave fronts propagate
583: laterally through the corona and are
584: at least a scale height tall, they will rely on a different
585: steepening/dispersion mechanism to create soliton-like behavior.
586: The steepening could come from the fact that Alfv\'en speed increases with
587: altitude in the corona. The
588: dispersion mechanism could be lateral density stratification across the pulse
589: itself. The fact that the medium is itself MHD also means that the wave must
590: have a magnetic component. However, since such a large pulse must
591: remain coherent to small perturbations (such as magnetic loops), it may imply
592: that only very strong (i.e. active regions) or very defined (i.e. coronal
593: holes) magnetic structures have any noticeable effect on the wave.
594:
595: The effect of a soliton on the local medium can also account for
596: loop oscillations. As a compressive wave packet with no related rarefaction,
597: it must displace the medium
598: in the direction of propagation, where the displacment will remain
599: unless restored
600: by some other force. While this argument can account for any linear
601: compressive wave packet, it is still consistent with the effects
602: of a soliton.
603: In the case of an EIT wave, coronal material will be carried with the front.
604: \citet{WD_Thompson}
605: demonstrated this for the 13 June
606: 1998 event, as they tracked individual loops along with the wave. However,
607: since the magnetic fields of the corona are anchored in the photosphere,
608: after the wave has moved on, the individual loops will ``snap'' back.
609: \citet{Wills-DaveyPhD}
610: found that most structures behaved in an overdamped
611: manner when returning to their original positions, but some
612: loops---often aligned perpendicular to the direction of propagation---showed
613: oscillatory behavior.
614:
615: Lastly, the production of a soliton does not require the presence of a
616: shock, allowing for the existance of EIT waves in the absence of
617: Moreton waves. While this still doesn't explain the relationship
618: between Moreton and EIT waves, a soliton-like EIT wave can
619: account for the vast majority of observations.
620:
621: \section{Discussion}
622: \label{sec:discussion}
623:
624: The consistency of the properties of EIT waves has long
625: motivated solar physicists to develop a physical understanding as to
626: their nature. Developing this understanding has proved elusive
627: in previous work. Unfortunately, fast MHD compressional waves do not properly
628: describe dynamics of many EIT wave events.
629: The physical properties of EIT waves---their single-pulse, stable
630: morphology; the non-linearity of their density perturbations; the
631: lack of a single representative velocity---instead suggest that they may be
632: best explained as soliton-like phenomena.
633:
634: While most fronts travel below the expected coronal Alfv\'en speed,
635: as a general trend, larger density perturbations tend to move at
636: faster velocities.
637: There is also the observational evidence that many EIT wave
638: pulse widths are close to one to two scale heights;
639: this may be a visual effect, but is is possible that pressure and magnetic
640: forces convolve to act as a wave guide, as predicted by
641: \citet{Nye_Thomas}. It would be consistent with
642: initial findings that flux is conserved as
643: EIT waves propagate radially along the
644: solar surface rather than spherically
645: \citep{Wills-DaveyPhD}. It might also account, at least in part,
646: for the strong discrepancy between the number of
647: EIT and SXT pulse wave observations
648: \citep{Sterling_Hudson,Biesecker,Warmuth2005};
649: SXT preferentially observes hotter structures with larger scale heights,
650: and the maximum pulse height of EIT waves might be constrained by
651: smaller, cooler loops.
652:
653: The flux conservation found by \citet{Wills-DaveyPhD}
654: does demonstrate one
655: surprising, unsoliton-like behavior: after a pulse stops forming, its
656: amplitude appears to drop off as $r^{-1}$.
657: This occurs in spite of the
658: other, soliton-like properties observed in EIT waves.
659: It is likely that flux conservation is a necessary aspect of
660: radially-propagating trapped MHD solitons, and that a decrease in amplitude
661: is necessary to a conservative solution.
662:
663: We feel the solitary wave hypothesis
664: offers the most compelling explanation to date
665: for the properties of EIT waves. While the derivation of a two-dimensional
666: MHD soliton solution is perhaps beyond analytical scope, and therefore must
667: be demonstrated numerically, the properties inherent in a soliton-like
668: explanation should
669: fit the data much better than the oft-used fast-mode solutions.
670: It becomes possible to explain:
671: the lack of a ``typical'' EIT wave velocity;
672: the amplitude-velocity relationship seen by \citet{Myers};
673: and the consistent observations of single, coherent, nonlinear coronal pulses.
674: While we do not pretend to offer a comprehensive explanation on the nature of
675: EIT waves, by offering this interpretation, we hope to assist theorists and
676: modelers by providing a new direction.
677:
678: Developing a consistent theoretical understanding of EIT waves is particularly
679: important in the context of new EUV missions. {\it TRACE}
680: observations have already shown that wave parameters are easily quantitatively
681: measured with sufficient spatiotemporal resolution \citep{Wills-Davey2006}.
682: If we can correctly
683: model and reproduce EIT waves, we can use wave properties
684: extracted from observations to inverse model the plasma parameters of the
685: affected quiet corona. Missions like {\it STEREO} \citep{STEREO} and
686: {\it GOES-N} \citep{GOES-N} will
687: give us the opportunity to develop and test these observational modeling
688: tools.
689:
690: Using large-scale propagating waves for
691: {\it global coronal seismology} has been postulated since \citet{Meyer}.
692: However, its successful implementation requires a well-understood
693: theoretical model. Previous studies \citep{Meyer,Ballai2003}
694: have attempted to calculate quiet sun magnetic field strengths with
695: global coronal seismology using the assumption that Moreton waves and
696: EIT waves can be modeled as MHD fast-mode waves. \citet{Meyer} finds
697: a field strength that appears high; \citet{Ballai2003}
698: finds one that
699: appears low. An accurate wave model
700: may result in a more reasonable field strength calculation, allowing
701: EIT waves to make the transition
702: from coronal phenomena to observational tool.
703:
704: \acknowledgements
705: The authors wish to thank V. J. Pizzo for insightful comments and editting.
706: Figure~\ref{fig:waves}(a) was reproduced by permission of B. J. Thompson.
707: Figure~\ref{fig:moreton} was reproduced by permission of N. Narukage.
708: This research was funded by NASA Grant LWS 02-0000-0025.
709:
710: \clearpage
711:
712: \bibliographystyle{apj.bst}
713: \bibliography{ms.bib}
714:
715: \clearpage
716:
717: \begin{figure}
718: \plotone{f1.eps}
719: \caption{An example of a Moreton wave observed in H-$\alpha$
720: on 4 November 1997 by the Flare Monitoring
721: Telescope of Kyoto University's Hida Observatory. This event was produced
722: in conjunction with a GOES~X2.1 flare.
723: {\it This figure reproduced from \citet{Narukage2003}
724: courtesy of N. Narukage.}}
725: \label{fig:moreton}
726: \end{figure}
727:
728: \begin{figure}
729: \plotone{f2a.eps}
730: \caption{Two examples of EIT waves as observed by different EUV
731: instruments. (a) shows running difference images of
732: an EIT wave seen by {\it SOHO}-EIT on 12 May 1997,
733: and studied in detail by
734: \citet{Thompson1998}. (b) shows base difference images
735: and measured fronts from \citet{Wills-Davey2006} of an event observed
736: by {\it TRACE} on 13 June 1998.
737: (In running
738: difference images, each frame is subtracted from the one following.
739: In base difference images, all frames have a single pre-event image
740: subtracted from them.)
741: {\it Figure~\ref{fig:waves}(a) reproduced courtesy of B. J. Thompson.}}
742: \label{fig:waves}
743: \end{figure}
744:
745: \begin{figure}
746: \plotone{f3.eps}
747: \caption{Distribution of average EIT wave speeds with respect to
748: ``Quality Rating,'' a subjective measure corresponding to the observer's
749: confidence in the velocity reading, with zero being a low confidence
750: score \citep{Myers}. While ``Quality Ratings''
751: have no quantifiable validity, these data suggest a correlation between
752: density enhancement and speed. Note the substantial velocity spread, with
753: events traveling at an average speed in the range $25 < v < 438$~km/s, and
754: many traveling below the minimum calculated Alfv\'en speed of 215~km/s.}
755: \label{fig:MT_plot}
756: \end{figure}
757:
758: \end{document}
759:
760: