1: \documentclass[12pt,preprint]{aastex}

2: \slugcomment{Submitted to ApJ}

3: \shorttitle{Quantitative Measurements of CME-driven Shocks}

4: \shortauthors{Ontiveros \& Vourlidas}

5: 

6: \begin{document}

7: 

8: \title{Quantitative Measurements of CME-driven Shocks from LASCO Observations}

9: 

10: \author{Veronica Ontiveros\altaffilmark{1}}

11: \affil{Instituto de Geofisica, Universidad Nacional Autonoma de

12:   Mexico, DF, 04510, MEXICO}

13: \email{vontiver@gmu.edu}

14: \and

15: \author{Angelos Vourlidas}

16: \affil{Code 7663, Naval Research Laboratory, Washington DC 20375, USA}

17: \email{vourlidas@nrl.navy.mil}

18: 

19: \altaffiltext{1}{present address: CEOSR, George Mason University, Fairfax VA, 22030, USA}

20: 

21: \begin{abstract}

22: 

23: In this paper, we demonstrate that CME-driven shocks can be

24: detected in white light coronagraph images and in which

25: properties such as the density compression ratio and

26: shock direction can be measured.  Also, their propagation direction can be

27: deduced via simple modeling. We focused on CMEs during the ascending

28: phase of solar cycle 23 when the large-scale morphology of the

29: corona was simple. We selected events which were good candidates to

30: drive a shock due to their high speeds (V$>$1500 km s$^{-1}$). The

31: final list includes 15 CMEs. For each event, we calibrated the LASCO

32: data, constructed excess mass images and searched for indications of

33: faint and relatively sharp fronts ahead of the bright CME front. We found such

34: signatures in 86\% (13/15) of the events and measured the

35: upstream/downstream densities to estimate the shock strength. Our

36: values are in agreement with theoretical expectations and show good

37: correlations with the CME kinetic energy and momentum. Finally, we

38: used a simple forward modeling technique to estimate the 3D shape

39: and orientation of the white light shock features. We found

40: excellent agreement with the observed density profiles and the

41: locations of the CME source regions. Our results strongly suggest that the observed brightness enhancements result from density enhancements due to a bow-shock structure driven by the CME.

42: 

43: 

44: \end{abstract}

45: 

46: \keywords{Sun: activity -- Sun: corona -- Sun: coronal mass ejections

47: -- Sun: shocks}

48: 

49: \section{Introduction}

50: 

51: Coronal Mass Ejections (CMEs) are the largest transient expulsions of

52: coronal material in the heliosphere. These explosive events are

53: recorded by coronagraphs as brightness enhancements in white light

54: images because the ejected material scatters a large amount of

55: photospheric light. In image sequences, local brightness changes

56: provide most of the information on CME parameters, such as speed and

57: mass. LASCO \citep{bru95} observations have established that CME

58: speeds vary from a few hundred to more than 2500 km/s

59: \citep{yas04}. With this wide range, it is reasonable to expect that

60: CME speeds often exceed the local magnetosonic speed, and drive a

61: shock wave in the low corona \citep{hun87}.

62: 

63: There are two main observational results that provide support for the

64: existence of shocks in the low corona. Metric type-II radio bursts

65: provide indirect evidence of CME-driven shocks \citep[e.g.,][]{cli99},

66: but the scarcity of imaging radio observations precludes the reliable

67: identification of the driver. Observations of distant (from the CME)

68: streamer deflections \citep[e.g.,][]{gos74, mic84,she00} provide the

69: most reliable indication of a CME-driven wave pushing out the

70: streamers. However, there remains the question of whether this wave is

71: a shock wave, specially for the cases where the CME speed is not

72: excessively high. \cite{vou03} presented the first direct detection of

73: a CME-driven shock in white light images, combining two signatures:

74: (1) a sharp but faint brightness enhancement ahead of the CME and (2)

75: a streamer deflection well-connected to the expansion of the sharp

76: front. \cite{vou03} confirmed that the white light signature was a

77: shock wave using an MHD simulation based on the measured CME speed and

78: location.

79: 

80: Despite the large number of CME observations with LASCO, CME-driven

81: shocks in white light images remain difficult to detect. The

82: brightness enhancement due to the shock itself is faint and can be

83: easily lost in the background corona which changes from event to

84: event. Projection effects can also make it difficult to recognize and

85: separate the shock signatures from the rest of the CME because

86: deflected streamers, the shock and the CME material can all overlap

87: along a given line of sight (LOS). If the shock exists, however, it

88: will result in a density enhancement and, with proper analysis, should

89: be visible in the images.

90: 

91: We note that the earlier paper by \citet{vou03} discussed shock

92: signatures related to rather small and fast events (such as surges and

93: jets). Here we extend the detection of white light signatures to

94: standard CMEs by analyzing a set of fast CMEs during the rising phase

95: of solar cycle 23. These two papers suggest that white light shock

96: signatures must be a common feature in coronagraph images and the

97: previous scarcity of shock detections is mostly due to reduced

98: sensitivity, fields of view and temporal coverage of past

99: instruments. 

100: 

101: The data selection and methodology are described in

102: \S~\ref{selection}. The unique aspects of this work are the

103: quantitative density measurements that allow us to estimate the shock

104: strength, as presented in \S~3, and the analysis of the 3D morphology

105: and orientation of the white light shock using the Solar Corona

106: Raytrace (SCR), a software package that simulates the appearance of

107: various 3D geometries in white light coronagraph images

108: \citep{the06}. We compare the modeled images to density profiles

109: obtained from LASCO images in \S~\ref{shape} and found them in

110: excellent agreement. A summary and general discussion are presented in

111: \S~\ref{summary}.

112: 

113: \section{Event Selection and Identification of the White Light Shock \label{selection}}

114: 

115: To identify a sample of CME events with likely shock signatures, we

116: used two general criteria: (1) we searched for fast CMEs ($>$1500 km

117: s$^{-1}$) because they are more likely to drive a shock, and (2) we

118: considered only CMEs occurring at the ascending phase of the solar

119: cycle 23 (1997-1999), when the simple morphology of the background

120: white light corona offers a better chance to observe faint shock-like

121: structures with minimal confusion from overlapping structures along

122: the LOS.  Only 15 CMEs (out of about 2000) satisfied our selection

123: criteria and are shown in Table \ref{list}. The first and second

124: column are the number and the date of the event; the third column is

125: the time of first appearance in LASCO C2; the fourth, fifth and sixth

126: columns are the linear speed, angular width (AW) and the central

127: position angle (PA) respectively, as reported in the CME LASCO catalog

128: \citep{yas04}. The seventh column shows whether the CME is

129: associated with a decimetric type II radio burst \citep{gop05}.

130: 

131: To identify whether a shock signature exists in a given LASCO image,

132: we look for white light features that satisfy the following criteria:

133: (1) it must be a smooth, large scale front, (2) it must outline the

134: outermost envelope of the CME, and (3) it should be associated,

135: spatially and temporally, with streamer deflections. We choose these

136: criteria based on our expectations of how a CME-driven shock should

137: behave; namely, it should be ahead of the CME material (the driver),

138: it should expand away from the CME over large coronal volumes (but

139: avoiding coronal holes, for example), and it should affect streamers

140: when it impinges on them \citep{vou03}.

141: 

142: It turns out that such fronts exist in the majority of the images we

143: looked at but they are generally much fainter than most of the other

144: CME structures. They remain unnoticed in the running difference or

145: quick-look LASCO images normally published in the literature. These

146: fronts become visible only when the brightness scale of these images

147: is saturated to bring out the fainter structures.

148: 

149: We are able to identify and analyze these fronts because we use

150: calibrated LASCO images. The calibration brings out faint structures,

151: which may be missed in standard image processing, because it removes

152: vignetting and other instrumental effects.  We use excess

153: mass images from both LASCO C2 and C3. These are calibrated images

154: from which a pre-event has been subtracted thus removing the

155: background corona \citep{vou00}. A frame from each of our CMEs is

156: shown in figure \ref{gallery}. Most of the images have a curved line

157: to guide the reader's eye to the shock signature, while the dashed

158: line in each frame is pointing out the position angle corresponding to

159: the measurements that will be explained in section

160: \S~\ref{gammas}. Because of the large variation in the brightness of

161: the features, we have to apply different contrasts to bring out the

162: shock signature in each CME. In most images, the shock signature is

163: associated with diffuse emission on the periphery of the much brighter

164: CME material. The diffuse emission could arise from the coronal

165: material on the shock surface. For some events, like \#1, \#6, or

166: \#11, the faint emission encompasses the CME as one would expect for

167: an ideal case of a shock enveloping the driver. In other events, e.g

168: \#8, \#9 and \#12, the faint emission is only seen over a small range

169: of position angles. In all cases, the emission has a smooth front,

170: follows the general shape of the driver material (the bright CME

171: core), and is associated with a streamer bent from its pre-event

172: position. All these features are clear indications that we are dealing

173: with a wave. 

174: 

175: The reason why we are confident that what we observe is indeed a shock wave lies with the CME speed. All our events are either halo or partial halo CMEs, and their speeds are lower limits to the true speeds. Even these lower limit

176: speeds are sufficiently high to create a shock wave for typical values of plasma parameters in the low corona \citep{hun87}. Therefore, we will refer to the observed feature as shocks, from now on.

177: 

178: \section{Shock Measurements\label{res}}

179: 

180: \subsection{Shock Signature Evaluation\label{groups}}

181: 

182: We identified a shock signature in the LASCO images for 13 out of the

183: 15 selected events. For events \#5 and \#13, it was not possible to

184: find a feature that satisfied even one of our three selection criteria

185: (\S~2). We believe that the lack of a smooth front and streamer

186: deflections may be due to the presence of a previous CME. In both

187: cases, the excess mass images showed clear evidence of a disturbed

188: corona (e.g., mass depletions, streamer displacements). A shock may

189: not form if the first event has altered the background magnetic field

190: considerably. Even if it forms, as the DM Type-II emissions suggest,

191: it is unlikely to develop a smooth, large front as it propagates

192: through such a disturbed medium. Similarly for streamer deflections,

193: many of the streamers could have already been deflected by the

194: previous CME at various angles from the sky plane and any new

195: deflection may not register as a smooth front in the images. Finally,

196: the strong intensity variations left on the image by the previous CME

197: may mask any faint fronts associated with the shocks from our CMEs. It

198: seems, therefore, that a relatively unperturbed corona facilitates the

199: detection of the faint CME-driven shocks. Nevertheless, once we establish which signatures are shock-related, we expect it will become easier to analyze events in more disturbed coronal conditions.

200:  

201: For events \#4, \#10 and \#11, we found a clear shock signature in the

202: LASCO C2 images, while for the rest of the events, the clearest

203: signatures were found in the LASCO C3 images. We found at least one

204: location with a clear white light shock signature for all halo CMEs

205: (10 events). This is expected if our shock interpretation is correct

206: since halos offer the best viewing of the CME flanks due to their

207: propagation along the LOS. We also note that our interpretation

208: implies that a major part of the halo CME extent is due to the shock

209: rather than actual ejected material and analysis of CME widths need to

210: take this fact into consideration.

211: 

212: \subsection{From Mass Profiles to Density Ratios\label{dens}}

213: 

214: We can use the calibrated LASCO images not only to identify the faint

215: shock fronts but also to derive some estimates of the density profile

216: across these fronts. Each pixel in the mass LASCO images gives the

217: total mass or, equivalently, the total number of electrons along the

218: LOS. This excess electron column density (e/cm$^{2}$) can be converted

219: to electron volume density (e/cm$^{3}$), if the depth of the structure

220: along the LOS is known. This quantity is unknown and it cannot be

221: reliably estimated without some knowledge of the 3D morphology of the

222: structure. We address the 3D aspect in \S~4 but for the analysis we

223: assume a nominal depth of 1 R$_\sun$ for all the events because it is

224: a convenient scale and likely a good upper limit given the slope

225: ($\sim 0.3 R_\sun$) of the brightness profiles (e.g.,

226: figure~\ref{ev1_2}).

227: 

228: We then derive the total electron column density along the profile, by

229: integrating the density of the background equatorial corona from the

230: Saito, Poland, Munro (SPM) model \citep{sai77}. Again, the actual

231: value of the background density for each event is not known and we

232: have to resort to a density model as is often the case in analysis of

233: coronal observations. Here we assume that the same equatorial SPM

234: model applies to all events for two reasons: (1) our sample covers

235: only a small phase of the solar cycle when the average density of the

236: background corona does not vary significantly, and (2) the density

237: enhancement at the shock front must come from streamer material since

238: shocks do not propagate, nor pile up material over coronal holes.

239: 

240: \subsection{Estimation of the Shock Strength\label{gammas}}

241: 

242: For each event we chose the image with the visually clearest shock

243: signature. In that image, we obtain several profiles at different PAs

244: along the shock front. Our method averages the emission along a narrow

245: range of PAs ($\sim$5$\degr$) to improve the signal-to-noise

246: ratio. The radial extent of each profile allows us to obtain the

247: upstream and downstream brightness at different angles of the shock

248: \citep[see also][]{vou03} and convert them to densities as mentioned

249: in \S~3.2. We classify the 15 events in four groups based on the

250: appearance of the shock signatures in the images and the evidence of a

251: jump in the density profiles.  Group 1 includes those events which

252: have a clear shock signature in the image and a steep jump in the

253: brightness profiles at the location of the shock front (6 events).

254: Group 2 includes those events which show a clear shock signature in

255: the image but the density jump at the shock front is barely detectable

256: above the noise (5 events).  Group 3 includes those events which have

257: shock signatures in the image but the density profiles are too noisy

258: to identify the jump at the shock front; there were 2 events in Group

259: 3.  Finally, Group 4 are the two events (\#5 and \#13) without any

260: shock signatures in any LASCO image.

261: 

262: The shock fronts are more visible on the images rather than in the

263: density profiles because of our eyes' spatial averaging ability. We

264: believe that the density profiles can be improved by averaging over a

265: larger angular width. However, this averaging tends to smooth the

266: profiles and reduce the density jump. Until we find a better averaging

267: method we adopted the $5 \degr$ width in the current work.

268: 

269: For this reason, we concentrate on the profiles with the sharpest

270: density jump. We use the density jump as a proxy to the shock

271: strength. We define $\Gamma_{CR}$ as the compression ratio of the

272: total to background volume densities, $\Gamma_{CR} = 1

273: +\frac{\rho}{\rho_0}, $ where $\rho$ is the excess density due to the

274: shock, and $\rho_0$ the upstream density obtained from the SPM model.

275: 

276: Figure \ref{ev1_2} shows event \#1 as an example. The two parallel

277: lines mark the position angles we average over to obtain the

278: brightness profile with the sharpest density jump, and hence the

279: strongest shock signature. The jump is located at 7.9 R$_\sun$ at

280: PA$=$321$\degr$. The observed density profiles and the ratio between

281: excess and background densities are plotted on the right panels of

282: figure \ref{ev1_2}. In this case, we obtain a $\Gamma_{CR}$ of 1.6 at

283: the location of the shock. We repeat the same analysis for the

284: remaining events in our sample. We also obtain the CME mass, momentum

285: and kinetic energy from the same images following the method described

286: by \cite{vou00} and using the speed measurements from the LASCO CME

287: catalog. These measurements allow us to get global CME parameters to

288: compare with the local shock strength, which are discussed next.

289: 

290: \subsection{Statistics} \label{statistics}

291: 

292: Table \ref{results} presents our measurements for the 13

293: events. Columns 1-2 correspond to the event number (\ref{list}) and

294: its group (~\S\ref{groups}). Columns 3 to 6 are the time of LASCO

295: observation, the heliocentric distance to the shock signature, the PA

296: of the profile and the estimated density jump, $\Gamma_{CR}$, of the

297: shock for that profile. Columns 7 to 9 are the CME mass, kinetic

298: energy and momentum respectively, obtained from calibrated images. Now

299: we can assess the validity of our main assumption; namely, whether the

300: faint structures seen ahead of the main CME ejecta could indeed be the

301: white light counterpart of the CME-driven shock. If this is true, we

302: expect a correlation between the magnitude of the density jump (or

303: $\Gamma_{CR}$) and the CME dynamical parameters, such as the CME

304: kinetic energy.

305: 

306: The plots in figure \ref{trends} show the trends and correlations

307: obtained between $\Gamma_{CR}$ and some CME parameters for the best

308: events only (groups 1 and 2). We find important correlations between

309: the shock strength and the CME momentum (cc=0.80), and kinetic energy

310: (cc=0.77). Furthermore, the largest $\Gamma_{CR}$ are associated with

311: the sharpest shock signatures (group 1, see \S~\ref{groups}) and the

312: highest kinetic energies and momenta. These results suggest that our

313: $\Gamma_{CR}$ parameter is associated with the CME dynamics as

314: expected from a shock-produced density jump.

315: 

316: Perhaps surprisingly, there is no obvious correlation between the

317: strength and the linear speed of the CMEs. The quoted speeds are taken

318: from the LASCO catalog and therefore correspond to the speed at the

319: position angle of the fastest moving feature in the LASCO images. Our

320: profiles were taken at different position angles since the shock front

321: is more easily discernible at some distance away from the CME

322: front. Also, the speeds are derived from linear fits to height-time

323: measurements extending over the full range of the LASCO field of view

324: (2-30 R$_\sun$) and correspond to the average CME speed over the field

325: of view. Our density jump is derived from a single snapshot of the CME

326: at a single heliocentric distance. A plot between $\Gamma_{CR}$ and

327: the CME speed at the same PA and distance might provide a better

328: correlation. We tried to make these speed measurements. However, the

329: large CME speed and synoptic LASCO cadences did not allow us to obtain

330: a sufficient number of data points to derive reliable speed

331: estimations for any of our events.

332: 

333: Figure~\ref{polar} is related to the visibility of the shock signature

334: in the LASCO images and shows that the clearest shock signatures were

335: found above or below $15\degr$ with respect to the solar equator. This

336: holds even for the halo CMEs where there the shock is visible over

337: more position angles. Considering the phase of the solar cycle, these

338: results show that locations away from the streamers are favorable

339: angles for shock signature observations on white light images. This

340: result should be kept in mind when searching for shock signatures in

341: coronagraph images. The complexity of the background corona masks the

342: faint shock emissions during solar maximum while there are few

343: sufficiently fast events to drive a shock during solar minimum.

344: 

345: \section{Estimating the Shock Geometry with a Forward-Modeling

346:   Technique\label{shape}}

347: 

348: The results in the previous sections provide ample support for a shock

349: interpretation of the faint emission ahead of the CME front. In addition, we have devised a practical way to model the faint emission with a prescribed shape using

350: forward modeling techniques. The advantages of this approach are the

351: speed and simplicity of the software, and the resulting information on

352: the 3D morphology and direction of the shock. The disadvantage is that

353: we have no way of calculating a goodness of fit for the model other

354: than a visual judgment on whether the envelope of the model fits the

355: observed emission envelope. For the sake of brevity, we use the term

356: 'fitting' from now on to describe the 'fitting-by-eye' we actually

357: employed.

358: 

359: For simplicity, we assume that the CME-driven shock has a 3-D bow

360: shock morphology, as expected from a body moving in uniform magnetized

361: flow (e.g., Earth's magnetosphere). We first need a geometric

362: description for such a model. We found one in \citet{smith03} which is

363: used to describe the shocks around Herbig-Haro objects. It is a a

364: surface of revolution, in cylindrical coordinates and is described by

365: the form (Eq. (1) in \citet{smith03})

366: \begin{equation}

367: \frac{z}{d} = \frac{1}{s} (\frac{R}{d})^s

368: \end{equation}

369: where s controls the opening angle of the bow, d is a scale length

370: (\textsl{semilatus rectum\/}) and $R$ is the heliocentric distance of

371: the nose of the shock. The variables $d$ and $s$ control the shape of

372: the bow shock and are the most important variables for visually

373: matching the shock shape to the observations. To adopt this model for

374: coronagraphic observations, we add a narrow shell of constant density,

375: $N_{e0}$, and width, $\Delta$. In this way, we can calculate the

376: brightness of the model using the Thomson scattering equations and the

377: spacecraft geometry and analyze the model images exactly as we do the

378: observations. We justify this shell as the plasma enhancement around

379: the shock surface at a given moment. While plasma pileup at CME fronts

380: is still an open question \citep{howard05}, it is expected that the

381: shock will cause local density enhancements as it propagates through

382: the corona \cite{vou03}.

383: 

384: We use the SCR software package to create a simulated coronagraph

385: image from the model. SCR is a numerical implementation of Thomson

386: scattering that renders a total (or polarized) brightness 2D image as

387: seen by a coronagraph (e.g. LASCO C2 or C3) given a 3D density

388: structure distribution \citep{the06} and is available in

389: Solarsoft. For all events, we assumed a constant thickness of $\Delta

390: = 0.3 R_\sun$ which is comparable to the width of the brightness jump

391: in the images. Because the height of the shock varied for each event,

392: we set the density, $N_{e0}$ within the thin shell to the estimated

393: density just ahead of the shock front (\S~3.2). Once the width and

394: density of the model shock were set, we tried to match it to the LASCO

395: images by varying its orientation in space and the geometric

396: parameters of the parabola ($d$ and $s$). When we were satisfied that

397: the simulated image fit visually the observed envelope of the shock,

398: we integrated along the 3D shape, using the LASCO viewing geometry,

399: and obtained a simulated brightness image of the model shock.

400: 

401: Figure \ref{los} shows simulated white light images of our bow-shock

402: model viewed from different lines of sight: (1) Along the Sun-Earth

403: line, (2) $10\degr$ west and (3) 45$\degr$ west, 45$\degr$ south. The

404: location of the bow shock nose is at 8 $R_\sun$. The images show the

405: full model for completeness but we have restricted our integrations to

406: a volume of $30 R_\sun^3$, so the actual model is truncated and the

407: long thin extensions in Figure~\ref{los} do not appear in our images.

408: 

409: To check whether the LASCO density profiles are consistent with a

410: bow-shock geometry, we fit an SCR model to each event and obtained

411: simulated density profiles over several PAs, using the same method as

412: described in \S~\ref{dens} for the LASCO observations. This procedure

413: is currently done by hand and the LOS integration is time-consuming in

414: our hardware. We are working on improving it but so far we were able

415: to perform detailed comparisons for only three events in our

416: sample. We chose events \#1 (November 6, 1997), \#6 (June 4, 1998) and

417: \#8 (November 26, 1998), which have some of the clearest shock

418: signatures. Figure \ref{ray} shows the results. Each plot shows the

419: comparison between the SCR profiles to the LASCO ones for different

420: PAs at the shock signature.

421: 

422: Figure \ref{ray} shows that the observed density profiles are

423: consistent with a bow shock geometry for at least 30$\degr$ along the

424: shock signature. The density fits are surprisingly good given the

425: simplicity of our model. Note that we did not attempt to fit either

426: the density jump or the shock LOS extent. They were kept constant for

427: each event. This result offers a strong indication that the overall

428: envelope of enhanced emission around the CME must come from a simple

429: structure (e.g., a bow shock) probably reflecting the simplicity of

430: the minimum corona. In other words, the shock structure, and probably

431: its visibility, may depend on the overall configuration of the large

432: scale corona. It will be interesting to repeat our analysis for CMEs

433: during the solar maximum.

434: 

435: Another consistency check comes from comparing the orientation of the

436: bow shock shell in 3D space (which we get from the SCR modeling) to

437: the expected orientation of the CME. For this, we make the usual

438: assumption that the core of the CME (the driver for us) propagates

439: radially outwards from the nearest possible source region (e.g., a

440: flaring active region). The source regions for our three events are as follows:

441: \begin{itemize}

442: \item Event \#1 is associated with the flare

443: observed at 11:49 at S18W63. We took this location as the source

444: region of the CME. Since this is a front side event, it is relatively

445: easy to determine that there is no other active region with a better

446: association.

447: \item The LASCO movies suggest that event \#6 is likely associated with a

448: filament eruption on the far side of the Sun. The filament was seen

449: for several days in H$\alpha$ images as it crossed over the western

450: limb. Extrapolating from its known position on May 29th we estimate

451: that the center of the filament would be at N43W107 on the day of the

452: eruption.

453: \item Event \#8 is also a backside CME resulting in an indirect source

454: region identification. We examined the EIT and LASCO movies for a few

455: days before and during the eruption. The low corona signatures of the

456: eruption suggest that active region 8384 is the most likely source

457: and should be located around S26W134 at the time of the eruption.

458: \end{itemize}

459: 

460: We then calculated the heliographic coordinates of the nose of our

461: modeled bow shock for each of the three events. The results for the

462: three events are shown in Table~\ref{location}. Again, we did not

463: attempt to take into account the location of the source region when we

464: fit the geometric model. Only during the writing of this paper we

465: calculated the final position of the shock nose and

466: compared it with the possible source regions. We were surprised to

467: find that the direction of the modeled shock is within $30\degr$ of the

468: expected CME nose, assuming radial propagation from the source

469: region. The discrepancy could be simply due to non-radial expansion of

470: the CME or uncertainty in the source region since two of the events

471: were backside CMEs. Given these restrictions, the results in

472: Table~\ref{location} are very encouraging because they suggest that

473: our forward modeling approach can provide useful information on the 3D

474: morphology and orientation of the shock using a single viewpoint and

475: modest hardware and software resources. We plan to investigate the

476: sensitivity of the derived shock orientation to different model fits

477: and apply it to more events in a future paper.

478: 

479: \section{Summary and Discussion \label{summary}}

480: 

481: In this paper, we demonstrate that the CME-driven shock is indeed

482: visible in coronagraph images. It can be seen as the faint large scale

483: emission ahead and around the bright CME material. To establish this

484: interpretation, we started by selecting all fast CMEs ($>$1500 km s$^{-1}$)

485: observed by LASCO between 1997-1999 (15 events). We found the

486: following:

487: \begin{itemize}

488: \item Ten of our events are associated with a decametric type II radio

489:   burst, suggesting the existence of a shock wave in the outer

490:   corona. The remaining five events are backside CMEs where

491:   the detection of radio burst is not always possible and the

492:   existence of a shock cannot be ruled out. In other words, the

493:   existence of a shock at the heights of the LASCO observations (2-30

494:   Rs) is supported by other observations for all events in our sample.

495: \item 86\% of these events exhibited a relatively sharp but faint brightness

496:   enhancement ahead or at the flanks of the CME over a large area,

497:   which we interpret as the white light counterpart of the CME-driven shock.

498: \item All halo CMEs (10 events) have at least one location with such a

499:   shock signature. This is consistent with a shock draping all around

500:   the CME driver.

501: \item The clearest white light signatures were found $15^\circ$ above

502:   or below the solar equator, irrespective of heliocentric

503:   distance. It is possible that the morphology and complexity of the

504:   corona along the LOS plays a role in identifying the shock in white

505:   light images.  The two events with no white light shock signatures

506:   were also the fastest, came in the wake of a previous large-scale

507:   CME. As we discussed in \S~3.1, the disturbed background corona may

508:   be responsible for the lack of shock signatures. It is also likely

509:   that any shock signatures may have been missed because of the low

510:   observational cadence and high speed of these events.

511: \item We found only a weak dependence between the shock strength

512:   ($\Gamma_{CR}$) and the CME speed. There may be several reasons for this

513:   discrepancy: (1) the speeds are more sensitive to projection effects, (2) the strength and speeds are measured at different PAs

514:   and/or (3) the speeds correspond to the average CME speed in the LASCO

515:   field of view while the strengths are measured in a single image.

516: \item We found stronger correlations between the density jump and the

517:   CME kinetic energy (cc=0.77) and between density jump and the momentum

518:   (cc=0.80). This is a very encouraging result because it shows that

519:   our density jump is closely related to the CME dynamics and hence

520:   more likely to correspond to a true shock jump.

521: \item We are able to account for the smooth observed jumps in the

522:   brightness profiles (and the derived density profiles) as compared to the

523:   step-like jumps observed in-situ. We found that they can be

524:   reproduced by a line-of-sight integration through a thin ($\sim 0.3

525:   R_\sun$) shell of material. This material is presumably the locally

526:   enhanced corona, which has become compressed due to the passage of the shock.

527: \end{itemize}

528: 

529: The high CME speeds, the sharpness of the features and the brightness

530: jumps are all strong indicators that our interpretation of these

531: features as the white light counterpart of CME-driven shocks is

532: correct. The strong correlations of the density jump to the CME

533: kinetic energy and momentum provide additional support. Based on the

534: information presented here, it should be a

535: simple matter to identify such features in all events where a shock is

536: expected. We have found many more examples in a quick

537: survey of LASCO images throughout the mission. It is still difficult,

538: however, to extract quantitative measurements from all of these shocks

539: due to the lower signal-to-noise ratio of the individual density

540: profiles compared to the images. We are looking for ways to

541: average across the shock front without introducing unnecessary

542: smoothing to it.

543: 

544: We have examined whether the observed shock shapes are consistent with

545: expected 3D shock geometries. We used a standard bow shock geometric

546: model, adapted from astrophysical shocks, and a forward modeling

547: software package from the SECCHI solarsoft collection to test a quick

548: method of estimating the shock size and orientation for

549: coronagraphs. We found that a bow-shock geometry is indeed a good fit

550: to the observed LASCO morphology and it readily explains the observed

551: density profiles. The simulated profiles can match the observed

552: profiles over several position angles, and even at large heliocentric

553: distances. We also found that our modeled 3D shock direction is in

554: fairly good agreement with the expected direction of the CME assuming

555: radial propagation from the source region.

556: 

557: These results suggest that we can not only estimate the 3-dimensional

558: shape and direction of the CME-driven shock but we can also use the

559: model fits to separate the brightness enhancement of the shock from

560: that of the driving material and thus obtain more accurate

561: measurements of the CME and shock characteristics. One such quantity

562: is the shock kinetic energy which plays an important role in

563: understanding and modeling the production of solar energetic particles

564: from shocks. We will pursue these ideas further in a future paper.

565: 

566: \acknowledgments We thank R. A. Howard for his editorial help and

567: valuable comments, which have improved the text considerably;

568: A. F. Thernisien for his generous help with the SCR software, and the

569: anonymous referee for his/her careful reading and suggestions. This work

570: is funded by the LWS TR\&T grant NNH06AD851. The CME catalog is

571: generated and maintained by NASA and The Catholic University of

572: America in cooperation with the Naval Research Laboratory. SOHO is a

573: project of international cooperation between ESA and NASA.

574: 

575: \begin{thebibliography}{}

576: 

577: \bibitem[Brueckner et al.(1995)]{bru95} Brueckner, G.~E., Howard, R.~A., Koomen, M.~J., Korendyke, C.~M., Michels, D.~J., Moses, J.~D., Socker, D.~G., Dere, K.~P., Lamy, P.~L., Llebaria, A., Bout, M.~V., Schwenn, R., Simnett, G.~M., Bedford, D.~K. \& Eyles, C.~J. 1995,  \solphys, 162, 357

578: 

579: \bibitem[Cliver et al.(1999)]{cli99} Cliver, E.~W., Webb, D.~F. \& Howard, R.~A. 1999, \solphys, 187, 89

580: 

581: \bibitem[Gopalswamy et al.(2005)]{gop05} Gopalswamy, N., Aguilar-Rodriguez, E., Yashiro, S., Nunes, S., Kaiser, M.~L. \& Howard, R.~A. 2005, \jgr (Space Physics), 110, A9, 12

582: 

583: \bibitem[Gosling et al.(1974)]{gos74} Gosling, J.~T., Hildner, E.,

584:   MacQueen, R.~M., Munro, R.~H., Poland, A.~I. \& Ross, C.~L. 1974,

585:   \jgr, 79,4581

586: 

587: \bibitem[Howard and Vourlidas(2005)]{howard05} Howard, R. A., \&

588:   Vourlidas, A. 2005, Eos Trans. 86(18), Jt. Assem. Suppl., Abstract SH53A-05

589: 

590: \bibitem[Hundhausen(1987)]{hun87} Hundhausen, A.~J., Holzer, T.~E. \& Low, B.~C. 1987, \jgr, 92, 11173

591: 

592: \bibitem[Michels et al.(1984)]{mic84} Michels, D.~J., Sheeley, Jr., N.~R., Howard, R.~A., Koomen, M.~J. Schwenn, R. Mulhauser, K.~H. \& Rosenbauer, H. 1984, Advances in Space Research, 4, 311

593: 

594: \bibitem[Saito et al.(1977)]{sai77} Saito, K., Poland, A.~I. \& Munro, R.~H. 1977, \solphys, 55, 121

595: 

596: 

597: \bibitem[Sheeley et al.(2000)]{she00} Sheeley, N.~R., Hakala, W.~N. \& Wang, Y.-M. 2000, \jgr, 105, 5081

598: 

599: \bibitem[Smith, Khanzadyan, and Davis(2003)]{smith03} Smith, M. D., Khanzadyan, T., \& Davis, C. J. 2003, MNRAS, 339, 52

600: 

601: \bibitem[Thernisien et al.(2006)]{the06} Thernisien, A.~F.~R., Howard, R.~A. \& Vourlidas, A. 2006, \apj, 652, 763

602: 

603: \bibitem[Vourlidas et al.(2000)]{vou00} Vourlidas, A., Subramanian, P., Dere, K.~P. \& Howard, R.~A. 2000 \apj, 534, 456

604: 

605: \bibitem[Vourlidas et al.(2003)]{vou03} Vourlidas, A., Wu, S.~T., Wang, A.~H., Subramanian, P. \& Howard, R.~A. 2003, \apj, 598, 1392

606: 

607: \bibitem[Yashiro et al.(2004)]{yas04} Yashiro, S., Gopalswamy, N., Michalek, G., St.~Cyr, O.~C., Plunkett, S.~P., Rich, N.~B. \& Howard, R.~A. 2004, \jgr (Space Physics), 109, A18, 7105

608: 

609: \end{thebibliography}

610: 

611: \clearpage

612: 

613: \begin{deluxetable}{lcccccc}

614: \tabletypesize{\scriptsize}

615: 

616: \tablecaption{All high speed CMEs (V$>$1500 km s $^{-1}$) between 1997-1999\label{list}}

617: \tablewidth{0pt}

618: \tablehead{

619: \colhead{event} & \colhead{CME date} &\colhead{1st appearance} & \colhead{linear speed} & \colhead{AW} & \colhead{PA} &\colhead{type II}\\

620: & \colhead{(yymmdd)} &\colhead{(C2 UT)} & \colhead{(km s$^{-1}$)} & \colhead{(deg)} & \colhead{(deg)}&\colhead{(Dm)}

621: }

622: \startdata

623: 1 & 971106 & 12:10:00 & 1556 & 360 & 262 & yes\\

624: 2 & 980331 & 6:12:00 & 1992 & 360 & 177 & no\\

625: 3 & 980420 & 10:07:00 & 1863 & 165 & 278 & yes\\

626: 4 & 980423 & 5:27:00 & 1618 & 360 & 116 & yes\\

627: 5 & 980509 & 3:35:58 & 2331 & 178 & 262 & yes\\

628: 6 & 980604 & 2:04:00 & 1802 & 360 & 314  & no\\

629: 7 & 981124 & 2:30:00 & 1798 & 360 & 226  & no\\

630: 8 & 981126 & 6:18:05 & 1505 & 360 & 198 &no\\

631: 9 & 981218 & 18:21:00 & 1749 & 360 & 36 & yes\\

632: 10 & 990503 & 6:06:00 & 1584 & 360 & 88 &yes\\

633: 11 & 990527 & 11:06:00 & 1691 & 360 & 341 &yes\\

634: 12 & 990601 & 19:37:00 & 1772 & 360 & 359 &yes\\

635: 13 & 990604 & 7:26:54 & 2230 & 150 & 289 &yes\\

636: 14 & 990611 & 11:26:00 & 1569 & 181 & 38 & yes\\

637: 15 & 990911 & 21:54:00 & 1680 & 120 & 13 & no\\

638: \enddata

639: 

640: \end{deluxetable}

641: 

642: 

643: \clearpage

644: 

645: \begin{deluxetable}{llccccccc}

646: \tabletypesize{\scriptsize}

647: 

648: \tablecaption{Results\label{results}}

649: \tablewidth{0pt}

650: \tablehead{

651: \colhead{event} & \colhead{group} &\colhead{shock time} & \colhead{shock height} & \colhead{shock PA} & \colhead{$\Gamma_{CR}$} & \colhead{mass} & \colhead{kinetic energy} & \colhead{momentum}\\

652: & &\colhead{(UT)} & \colhead{(R$_\sun$)} & \colhead{(deg)} & &\colhead{($\times$10$^{15}$g)}& \colhead{($\times$10$^{31}$ erg)} & \colhead{($\times$10$^{24}$ dyn s)}}

653: \startdata

654: 1 & 1 & 12:41:05 & 7.9 & 321 & 1.6 & 5.48 & 6.63 & 0.85 \\

655: 2 & 1 & 7:29:37 & 19.6 & 158 & 2.4 & 15.74 & 31.23 & 3.13 \\

656: 3 & 3 & 12:42:05 & 23.7 & \nodata & \nodata & 23.52 & 40.82 & 4.38 \\

657: 4 & 1 & 5:55:22 & 4.4 & 114 & 1.2 & 5.51 & 7.21 & 0.89 \\

658: 5 & 4 & \nodata & \nodata & \nodata & \nodata & \nodata & \nodata & \nodata \\

659: 6 & 2 & 3:41:14 & 9.8 & 302 & 1.4 & 5.3 & 8.6 & 0.95\\

660: 7 & 1 & 4:42:05 & 11.7 & 201 & 2.8 & 14.62 & 22.23 & 2.55\\

661: 8 & 1 & 6:18:05 & 9 & 217 & 1.6 & 10.98 & 13.21 & 1.70\\

662: 9 & 1 & 19:41:42 & 13.6 & 73 & 1.8 & 7.43 & 11.32 & 1.30\\

663: 10 & 2 & 8:18:05 & 6 & 75 & 2.2 & 10.44 & 13.1 & 1.65\\

664: 11 & 2 & 13:38:17 & 4.4 & 298 & 1.7 & 3.7 & 5.28 & 0.63\\

665: 12 & 2 & 21:18:07 & 11.8 & 1 & 1.8 & 11.09 & 17.41 & 1.96\\

666: 13 & 4 & \nodata & \nodata & \nodata & \nodata & \nodata & \nodata & \nodata \\

667: 14 & 2 & 14:18:05 & 10.4 & 19 & 1.9 & 11.42 & 14.06 & 1.79\\

668: 15 & 3 & 23:42:05 & 17.9 & \nodata & \nodata & 2.6 & 3.67 & 0.44\\

669: \enddata

670: 

671: \end{deluxetable}

672: 

673: \clearpage

674: \begin{deluxetable}{lrr}

675: \tabletypesize{\scriptsize}

676: 

677: \tablecaption{Comparison of Modeled Shock Orientation and CME Source Regions\label{location}}

678: \tablewidth{0pt}

679: \tablehead{

680: \colhead{event} & \colhead{shock nose} &\colhead{source region}\\

681: }

682: \startdata

683: 971106 & S13W56 & S18W63 \\

684: 980604 & N47W138 & N43W107 \\

685: 981126 & S38W108 &S26W134 \\

686: \enddata

687: \end{deluxetable}

688: \clearpage

689: \begin{figure}

690: \epsscale{0.8}

691: \plotone{f1.eps}

692: \caption{The sample of the 15 fastest CMEs observed by the LASCO

693:   coronagraphs between 1997 and 1999. The image with the clearest

694:   shock signature is shown in each panel. The radial lines mark the

695:   position angles of the density profiles analyzed. The curved lines

696:   are visual guides for the location and extent of the faint shock

697:   structures. These features may be more visible in the online version

698:   of the figure. For events 5 and 13 it was not possible to determine

699:   a clear signature due to the disturbed background

700:   corona. \label{gallery}}

701: \end{figure}

702: \clearpage

703: \begin{figure}

704: \plotone{f2.eps}

705: \caption{\textit{Left panel:\/} Selected image for the November 6, 1997

706:   CME. A clear shock signature can be seen at the flanks of the

707:   CME. The parallel lines over the shock front show the profile with the strongest shock signature. \textit{Right panel:\/} The top plot shows the estimated

708:   up-stream and down-stream density profile at PA = 312 (solid) and the

709:   background corona density from the SPM model (dashed). The bottom

710:   plot shows the density ratio, $\Gamma_{CR}=1.6$, at $7.9$

711:   $R_\sun$ which we use as a proxy to the shock strength.}\label{ev1_2}

712: \end{figure}

713: \clearpage

714: \begin{figure}

715: \epsscale{0.4}

716: \plotone{f3.eps}

717: \caption{Trends and correlations between estimated shock $\Gamma_{CR}$ and

718:   select CME parameters. The top and middle panels show the correlations between

719:   the shock $\Gamma_{CR}$ and the CME kinetic energy (cc=0.77) and momentum

720:   (cc=0.80), respectively. The bottom panel shows no clear correlation

721:   between the shock $\Gamma_{CR}$ and the CME speed (group 1: stars, group

722:   2: diamonds). \label{trends}}

723: \end{figure}

724: 

725: \clearpage

726: \begin{figure}

727: \plotone{f4.eps}

728: \caption{Locations on the plane of the sky of the clearest shock

729:   signatures in our event list. The dashed lines separate favorable

730:   and non-favorable position angles for a shock observation. All

731:   events showing a clear shock signature in the image and/or in the

732:   data analysis (stars and diamonds labels), are found below or above

733:   15 degrees from the solar equator.\label{polar}}

734: \end{figure}

735: 

736: \clearpage

737: \begin{figure}

738: \plotone{f5.eps}

739: \epsscale{0.4}

740: \caption{Simulated white light images for a bow-shock model at 8

741:   R$_\sun$ observed through different lines of sight, from top to

742:   bottom:(1) Along the Sun-Earth line, (2) 10$\degr$ west and (3)

743:   45$\degr$ west, 45$\degr$ south. The hole in the center of the

744:   images shows the size of the solar disk for scale. The intensity

745:   gradient represents the white light brightness of the model as

746:   viewed by LASCO C3.\label{los}}

747: \end{figure}

748: \clearpage

749: 

750: \begin{figure}

751: \epsscale{0.7}

752: \plotone{f6.eps}

753: \caption{Comparing observed and modeled density profiles for events \#1,\#

754:   6 and \#8. From top to bottom for each column:(1) the actual LASCO

755:   image and the selected angles for obtaining the density profiles;

756:   (2) A synthetic coronagraph image of the shock model overplotted on the LASCO

757:   image; (3-5) comparison of density profiles at different PAs (solid

758:   line: LASCO density profile; dotted line: model shock density profile; dashed

759:   line: background coronal density from the SPM model).\label{ray}}

760: \end{figure}

761: 

762: 

763: \end{document}

764: 

765: