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

2: %\documentclass[12pt,preprint,xdvi]{emulateapj}

3: 

4: \usepackage{natbib}

5: \usepackage{graphicx}

6: \usepackage{times}

7: \usepackage{natbib}

8: \usepackage{graphicx}

9: %\usepackage{lineno}

10: 

11: \newcommand{\be}{\begin{equation}}

12: \newcommand{\ee}{\end{equation}}

13: \newcommand\eq{eq.}

14: \newcommand\eqs{eqs.}

15: \newcommand\fig{Fig.}

16: \newcommand\figs{Figs.}

17: \def\vb{v_{\rm b}}

18: \def\va{v_{\rm A}}

19: \def\vae{v_{_{\rm A},e}}

20: \def\ta{T^{\rm A}_e}

21: \def\bvec{{\bf B}}

22: \def\kvec{{\bf k}}

23: \def\vvec{{\bf v}}

24: \def\uvec{{\bf u}}

25: \def\vpar{v_{_{\parallel}}}

26: \def\xvec{{\bf x}}

27: \def\that{{\bf \hat{t}}}

28: \def\xhat{{\bf \hat{x}}}

29: \def\yhat{{\bf \hat{y}}}

30: \def\zhat{{\bf \hat{z}}}

31: \def\momvec{{\bf P}}

32: \def\half{\hbox{${1\over2}$}}

33: \def\quarter{\hbox{${1\over4}$}}

34: \def\disc#1{{[\![#1]\!]}}

35: \def\avg#1{{\bigl\langle#1\bigr\rangle}}

36: 

37: \bibliographystyle{/Users/danalongcope/Library/texmf/tex/misc/apj}

38: 

39: \begin{document}

40: 

41: \title{Gas-dynamic shock heating of post-flare loops due to 

42: retraction following localized, impulsive reconnection}

43: 

44: \author{D.W. Longcope,$^1$ S.E. Guidoni,$^1$ and M.G. Linton$^2$}

45: \affil{1. Department of Physics, Montana State University,

46:   Bozeman, Montana 59717\\

47: 2. Naval Research Laboratory, Washington, D.C.}

48: 

49: \keywords{MHD --- shock waves --- Sun: flares}

50: 

51: %\linenumbers

52: 

53: \begin{abstract}

54: We present a novel model in which field lines shortening after

55: localized, three-dimensional reconnection heat the

56: plasma as they compress it.  The shortening progresses

57: away from the reconnection site at the Alfv\'en speed, 

58: releasing magnetic energy and generating parallel, compressive flows.

59: These flows, which are highly supersonic when $\beta\ll1$,

60: collide in a pair of strong gas-dynamic shocks at which 

61: both the mass density and temperature are raised.

62: Reconnecting field lines initially 

63: differing by more that $100^{\circ}$ can produce

64: a concentrated knot of plasma hotter that $20$ MK at the loop's apex, 

65: consistent with observations.

66: In spite of these high temperatures, the shocks convert

67: less than $10\%$ of the liberated magnetic energy into heat --- the

68: rest remains as kinetic energy of bulk motion.  These gas-dynamic

69: shocks arise only when the reconnection is impulsive and localized in

70: all three dimensions; they are distinct from the slow

71: magnetosonic shocks of the Petschek steady-state reconnection model.

72: \end{abstract}

73: 

74: \date{Draft: \today}

75: 

76: \section{Introduction}

77: 

78: Magnetic reconnection has long been proposed as a mechanism for heating

79: coronal plasma.  In one early

80: model \citep[see \fig\ \ref{fig:geom}]{Kopp1976},

81: reconnection occurs between open field lines separated by 

82: a vertical current sheet (red line) creating 

83: new closed field lines (post-flare loops, grey).  

84: Closing these field lines stops the solar wind upflow in a

85: gas-dynamic shock (GDS) that \citet{Kopp1976} estimated would raise 

86: the temperature by $80\%$.  \cite{Cargill1982} found the direct

87: magnetic energy conversion by reconnection 

88: to be a far more effective source of heating in this same model.  It

89: could raise the temperature of post-flare loops by up to

90: a factor of three to 6 MK.  Even this higher value is, however, 

91: insufficient to explain the $15$--$20$ MK temperatures observed at

92: apices of post-flare loops in Soft X-ray and Fe {\sc xxiv} EUV

93: emission \citep{Warren1999}.

94: 

95: \begin{figure}[htp]

96: \epsscale{1.0}

97: %\epsscale{1.1}

98: \plotone{fig1.eps}

99: \caption{The classical Kopp-Pneuman model and its modifications.

100: The side view (left) shows the basic two-dimensional geometry.  Inset

101: (a) on the upper left shows how steady Petschek reconnection is configured

102: within the current sheet.  Inset (b) on the lower left shows how a

103: transient reconnection produces a finite layer of reconnected flux

104: (grey).  The front view (right) shows how the model becomes

105: three-dimensional when the reconnection is localized in third

106: dimension ($x$).  Instead of a current layer the reconnection

107: produces a $\Lambda$-shaped flux tube of closed field (grey);

108: a completely detached, V-shaped flux tube (not shown) is

109: created at the same time.}

110: 	\label{fig:geom}

111: \end{figure}

112: 

113: Recent theoretical investigations have revealed that 

114: electric fields large enough for fast magnetic reconnection

115: can be self-consistently produced by a wide range of

116: small-scale mechanisms, provided only that they are localized

117: within a segment of the 

118: current sheet \citep{Birn2001,Biskamp2001}.  All such

119: processes generate reconnection flows resembling the model of

120: \citet[see inset a]{Petschek1964}

121: with slow magnetosonic shocks outside the non-ideal region

122: (SMSs, blue lines) at which the magnetic field is deflected and

123: weakened, thereby heating the plasma.

124: When reconnection is localized in both space and time

125: the SMSs close back together across a finite layer of horizontal field, as

126: shown in inset (b) of

127: \fig\ \ref{fig:geom} \citep{Semenov1983,Biernat1987}.  This field forms

128: the top of a ``hairpin'' comprising all the

129: flux closed since the onset of reconnection.

130: 

131: As the hairpin flux sheet retracts its field lines become much

132: shorter and releases substantial magnetic energy.

133: This magnetic energy is converted almost entirely into kinetic energy

134: rather than partly into heat as in Petschek's steady state model

135: \citep{Semenov1998}.  In a strictly two-dimensional version of the 

136: model the mass of the shortened tube accumulates in the tip of the the

137: retracting hairpin.  This ``snowplowing'' artifact is absent when

138: reconnection is also localized in the third dimension ($x$) and 

139: there is a magnetic

140: field component in that direction.  With the horizontal field

141: component (sometimes called a ``guide field'') 

142: the current sheet separates field with angle $2\zeta$

143: ($\tan\zeta=B_z/B_x$, see the ``front view'' of \fig\

144: \ref{fig:geom}).

145: 

146: Flux reconnected within a patch and over a finite interval 

147: forms a $\Lambda$-shaped flux tube (grey) similar to

148: magnetospheric {\em flux transfer events} 

149: \citep{Russell1978,Lee1993,Otto1995}.

150: The tube is distinguished from the surrounding field (the flux layers)

151: by the distinct connectivity given it through reconnection.  Different

152: field line geometry, such as the bend, produce dynamics

153: in the tube entirely different from those in the surrounding flux

154: layers, with which it has little subsequent interaction.

155: 

156: Recently \citet{Linton2006} studied the relaxation of this

157: post-reconnection flux tube using three-dimensional

158: magnetohydrodynamic (MHD) simulation.  They found that the

159: perpendicular dynamics of the tube, as it moves between the flux layers,

160: is well approximated by the equations

161: for a thin magnetic flux tube \citep{Spruit1981}, although with

162: $\beta<1$.  They presented an analytic solution in which

163: the post-reconnection tube shortens at the Alfv\'en

164: speed, converting magnetic energy into kinetic energy.

165: \citet{Linton2006} did not investigate the

166: dynamics parallel to the tube or any associated thermal

167: effects of shortening.  

168: 

169: This letter demonstrates that flux tube shortening is a powerful, inevitable 

170: mechanism for heating post-flare loops, which has not been previously

171: investigated.  The magnetic forces responsible for shortening, also drive

172: compressive parallel flows at the Alfv\'en speed.  At very low

173: $\beta$, these are high-Mach number flows 

174: whose collision naturally generates very strong shocks.  

175: The shocks are distinct from the SMSs of Petschek reconnection and are driven

176: by reconnection-initiated, perpendicular dynamics. This is in contrast to

177: previous investigations of flux tube shocks wherein acoustic

178: wave-steepening or pressure differences were considered as drivers 

179: \citep{Herbold1985,Thomas1991}. 

180: 

181: \section{Post-reconnection flux tube dynamics}

182: 

183: We begin by assuming that localized, transient, fast

184: magnetic reconnection has occurred, by an unspecified 

185: physical mechanism, within an otherwise

186: static current sheet.  This will, as just discussed, leave

187: a $\Lambda$-shaped flux tube, initially at rest.  Due to its 

188: sharp bend it is out of equilibrium, and

189: magnetic forces start it 

190: sliding downward between the magnetic layers separated by the current

191: sheet.

192: 

193: The tube's retraction, unhindered by the external flux layers, 

194: can be modeled 

195: using the {\em thin flux tube} equations of \citet{Spruit1981} and

196: subsequent authors \citep[see, for example, the review

197: by][]{Fisher2000}.  While the high-$\beta$ flux tubes in those

198: previous investigations are confined by the pressure of the

199: unmagnetized convection zone, our post-reconnection tube has very low $\beta$

200: and is confined by the magnetic pressure of the

201: flux layers outside the current sheet.  

202: The tube is an isolated entity

203: distinguished from its surroundings by its connectivity

204: \citep{Linton2006,Linton2008}.

205: We assume sufficient collisionality to justify the use of MHD equations 

206: throughout.

207: 

208: The tube is assumed thin enough to be described only by

209: its axis.  Internal properties such as the

210: magnetic field strength, $B_i$, pressure, $p_i$, and mass density

211: $\rho_i$, are function only of axial position.

212: The tube is also thin enough for fast magnetosonic waves to

213: establish pressure balance across its diameter virtually

214: instantaneously.  This assumption constrains the internal properties

215: to match those outside the flux tube: 

216: $B_i^2/8\pi+p_i=B_e^2/8\pi+p_e$, assumed to be uniform and constant.

217: 

218: We will also assume the plasma $\beta$ to be always small.  The main

219: force on a section of tube is therefore the magnetic tension due to

220: curvature of the axis $B_i^2(\partial\that/\partial\ell)/4\pi$,

221: where $\ell$ is arc-length

222: and $\that=\partial\xvec/\partial\ell$ is the unit tangent vector. 

223: Since it is ultimately the Lorentz force, it is natural that this

224: force is strictly perpendicular to the axis

225: ($\that\cdot\partial\that/\partial\ell=0$).

226: 

227: The pressure gradient, $-\that\,\partial p_i/\partial\ell$, is

228: formally smaller, by a factor of $\beta$, than the magnetic tension.

229: It is, however, the only force parallel to the axis, and is

230: essential to arresting internal pile-up of mass.  We therefore retain

231: terms involving pressure to first order in

232: $\beta$.

233: 

234: Velocity evolution is governed by a momentum equation with only

235: these two forces.  Conservation properties become apparent

236: when arc-length is replaced with $\mu$, 

237: the integrated mass per unit flux:

238: $\partial\mu/\partial\ell=\rho_i/B_i$.  The value of $\mu$ never

239: changes for a given fluid element.  The resulting momentum equation

240: \be

241:   {d\vvec\over dt} ~=~

242:   {B_i\over 4\pi\rho_i}{\partial\over\partial\ell}(B_i\that)

243:   ~=~ {\partial\over\partial\mu}\left(\, {B_i\that\over4\pi } \right) ~~,

244: 	\label{eq:FTe}

245: \ee

246: includes the parallel pressure gradient after use of pressure balance,

247: $B_i=B_e + 4\pi(p_e-p_i)/B_e$, valid to first order in $\beta$.

248: The momentum per unit flux of any section of tube,

249: \be

250:   {\momvec} ~=~ \int{\rho_i\vvec\over B_i}\,d\ell ~=~ \int\vvec\,d\mu

251:   ~~,

252: 	\label{eq:mom}

253: \ee

254: changes only through forces (per unit flux)

255: from the ends of the section, $B_i\that/4\pi$,

256: directed parallel to the axis.

257: 

258: \section{Shock relations in thin flux tubes}

259: 

260: Momentum conservation leads to a set of shock relations for thin flux

261: tubes.  Consider two straight sections with uniform

262: properties (designated $1$ and $2$), 

263: separated by an abrupt change at coordinate $\mu_0$.

264: The length-scale of this change is large compared 

265: to the tube radius but otherwise small

266: enough that we hereafter call it a

267: ``discontinuity'' and ``corner''.\footnote{Analysis of the

268: discontinuity's internal structure must be done outside the thin-tube

269: approximation.  Doing so will probably reveal its length-scale to be

270: comparable to the tube's radius.

271: This is analogous to non-ideal effects resolving internal structure

272: of a hydrodynamic shock at scales comparable to the mean free path

273: \citep{Grad1951}. Following that analogy, the thin flux tube equations

274: provide external conservation laws leading to the shock relations.}

275: This feature moves through space at

276: constant velocity $\uvec$ while its Lagrangian coordinate changes at

277: constant rate $\dot{\mu}_0$. We assume that the properties of the 

278: straight sections do not change as this

279: happens.  Fluid velocities on either side of

280: the discontinuity differ from $\uvec$, but

281: components perpendicular to the tangent vector match that

282: of $\uvec$: $(\vvec-\uvec)\times\that=0$.

283: 

284: The component of relative velocity parallel to the tangent vector,

285: $\vpar=\that\cdot(\vvec-\uvec)$, represents a flow across the

286: discontinuity.  The mass flux (per magnetic flux) through the

287: discontinuity, $\dot{\mu}_0$, must equal the mass flux across points

288: on either side of it, since the corner moves

289: without changing.  For positions $\mu_1$ and $\mu_2$, separated by

290: fixed distances from $\mu_0$ this means

291: \be

292:   \dot{\mu}_2 = -\left.{\rho_i\vpar/B_i}\right\vert_2

293:   = -\left.{\rho_i\vpar/B_i}\right\vert_1 ~=~\dot{\mu}_1 

294:   ~=~\dot{\mu}_0~~.

295: 	\label{eq:mass_cons}

296: \ee

297: 

298: The flux tube between $\mu_2$ and $\mu_1$ does not change so its

299: momentum (per magnetic flux) is constant,

300: \be

301:   \dot{\momvec} = \disc{\dot{\mu}\vvec} + \disc{B_i\that/4\pi}

302:   = 0 ~~,

303: \ee

304: denoting $\disc{f} = f_2 - f_1$.  Since

305: $\vvec=\uvec+\vpar\that$ away from the discontinuity

306: momentum constancy may be written

307: \be

308:   \avg{B_i/4\pi - \rho_i\vpar^2/B_i}\disc{\that} 

309:   -\disc{p_i/B_e + \rho_i\vpar^2/B_i}\avg{\that} =0 ~~,

310: 	\label{eq:mom_cons}

311: \ee

312: where $\avg{f}=(f_2+f_1)/2$.

313: 

314: The vectors $\disc{\that}$ and $\avg{\that}$ are orthogonal

315: so \eq\ (\ref{eq:mom_cons}) represents two independent 

316: equations.  As long as the bend is not a hairpin

317: ($\avg{\that}\ne0$) the $\avg{\that}$

318: component of momentum conservation, to lowest order in $\beta$, is 

319: equivalent to gas-dynamic momentum conservation: 

320: $\disc{p_i +\rho_i\vpar^2}=0$.  To the same order in $\beta$, \eq\

321: (\ref{eq:mass_cons}) gives the gas-dynamic mass continuity equation:

322: $\disc{\rho_i\vpar}=0$.  Provided there is no heat flow from the

323: external field or across $\mu_1$ or $\mu_2$, then the sum of kinetic

324: and thermal energy within the tube section will also be conserved.

325: This provides one additional, independent constraint on

326: $\rho_i$, $p_i$ and $\vpar$ across the discontinuity.

327: (Since $B_i$ can be related directly to $p_i$ it can be

328: eliminated from energy conservation.)

329: 

330: The foregoing describes three relations between six

331: quantities which do not include the magnetic

332: field's strength or direction.  The

333: relations are the traditional gas-dynamic Rankine-Hugoniot

334: conditions \citep{Courant1948}, but for flows inside a flux tube 

335: \citep{Ferriz-Mas1987}.  The other component of \eq\ (\ref{eq:mom_cons}), to

336: lowest order in $\beta$,

337: \be

338:   \avg{1 - (4\pi\rho_i/B_e^2)\vpar^2}\,

339:   \bigl\vert\disc{\that}\bigr\vert^2 = 0 ~~.

340: 	\label{eq:bend_eq}

341: \ee

342: constitutes one more relation, which does involve the field direction.

343: 

344: From values on one side the Rankine-Hugoniot relations may be

345: satisfied in two different ways by values on the other \citep[see for

346: example][]{Courant1948}.  They may be satisfied non-trivially by a unique

347: set of {\em different} values, or they may be satisfied trivially by

348: the {\em same} set of values

349: (i.e.\ $\disc{\rho_i}=\disc{\vpar}=\disc{p_i}=0$).

350: In the non-trivial case, the three different

351: quantities satisfy three independent constraints and cannot be forced,

352: in general, to satisfy a fourth.  The leading factor of \eq\

353: (\ref{eq:bend_eq}), however, constitutes a fourth independent

354: constraint so it will not in general vanish; it is therefore

355: necessary that $\disc{\that}=0$ (there is no bend).

356: This means that a thin, low-$\beta$ flux

357: tube can support discontinuities in internal quantities

358: only at a GDS within a straight section of tube.

359: 

360: If, on the other hand, there

361: {\em is} a bend in the flux tube ($\disc{\that}\ne0$), the only

362: way to satisfy \eq\ (\ref{eq:bend_eq}) is for the

363: pre-factor to vanish.  Since the non-trivial solution of all three 

364: Rankine-Hugoniot conditions would over-determine the system, they must

365: be satisfied trivially, without discontinuity.  In other words

366: $\rho_i$, and $\vpar$ are continuous at $\mu_0$ and satisfy the

367: relation $\vpar=B_e/\sqrt{4\pi\rho_i}=\va$ the Alfv\'en speed.  This

368: is similar to an intermediate shock \citep{Priest2000}, but includes the

369: influence of fast magnetosonic waves assumed to be maintaining pressure

370: balance across the tube.

371: 

372: \section{Shocks in the retracting flux tube}

373: 

374: The $\Lambda$-shaped bend in the post-reconnection flux tube shown in

375: \fig\ \ref{fig:geom} will immediately decompose into four different

376: shocks of the kinds described above.  Two intermediate shocks (bends,

377: B) propagate along the field lines, forming a straight horizontal

378: section between them (see \fig\ \ref{fig:bend}), as previously found

379: by \citet{Linton2006}.

380: Two GDSs propagate away

381: from the center, at $\pm v_s$, along the horizontal section.

382: This symmetric arrangement divides the

383: tube into sections labeled, $3$, $2$ and $1$, outward from the center.

384: 

385: \begin{figure}[htp]

386: \epsscale{0.95}

387: %\epsscale{0.8}

388: \plotone{fig2.eps}

389: \caption{The flux tube retracting from the $\Lambda$-shaped initial

390: condition shown in \fig\ \ref{fig:geom}.  This initial configuration is

391: shown by thick dashed lines, while its current state is shown in

392: grey.  The past trajectory of the fluid, shown by solid lines, is

393: initially focused inward.  The gas-dynamic shocks (GDS) 

394: redirect the flow vertically; their past positions are shown by a thin

395: dashed line.  Circled numbers designate the

396: states, $1$, $2$ and $3$, separated by these shocks.}

397: 	\label{fig:bend}

398: \end{figure}

399: 

400: Section $1$ consists of the flux tube at rest and in its initial state:

401: $\vvec_1=0$, $\rho_1=\rho_e$ and $p_1=p_e$.  The initial field lines

402: are inclined at angle $\zeta$, so 

403: $\that_1=\xhat\cos\zeta-\zhat\sin\zeta$, while

404: $\that_2=\xhat$ in the horizontal section.  The bend

405: propagates along the initial field at the Alfv\'en speed:

406: $\uvec=\vae\that_1$.  The relative parallel flow is continuous

407: across the bend

408: $v_{\parallel,1}=v_{\parallel,2}=-\vae$.  These facts can be combined

409: into the post-bend fluid velocity

410: \be

411:   \vvec_2 = \uvec + \vpar\that_2 = 

412:   -2\vae\sin^2(\zeta/2)\xhat-\vae\sin\zeta\zhat ~~,

413: \ee

414: directed along the bisector of the bend 

415: (the mean direction of the curvature force).

416: 

417: Internal quantities are continuous across the bends, so

418: $\rho_2=\rho_1=\rho_e$ and $p_2=p_1=p_e$.

419: From a reference frame moving downward at $-\zhat\vae\sin\zeta$, sections

420: $2$ and $3$ appear to form a classic shock tube, with inflow at

421: \be

422:   M_2 = {|v_{x,2}|\over c_{s,e}} = 

423:   \sqrt{8\over\gamma\beta_e}\sin^2(\zeta/2) ~~,

424: \ee

425: where $\gamma=5/3$ is the ratio of specific heats.

426: Since $\beta_e\ll1$, this Alfv\'enic inflow can have extremely high

427: Mach number when field lines of significantly different 

428: orientation reconnect.

429: 

430: This supersonic inflow is brought to rest, $v_{x,3}=0$, by a GDS

431: moving outward at $v_s$.  This stopping shock, equivalent to a piston

432: moving into stationary fluid at speed $|v_{x,2}|$, is a classic

433: problem \citep[see \S 69 from][for example]{Courant1948}, for which the

434: solution is

435: \be

436:   {v_s/|v_{x,2}|} = \sqrt{M_2^{-2}+(\gamma+1)^2/16}

437:   - \quarter(3-\gamma) ~~.

438: \ee

439: The density ratio, $\rho_3/\rho_2=1+|v_{x,2}|/v_s$, following from mass

440: conservation, approaches the well-known limit

441: $\rho_3/\rho_e=(\gamma+1)/(\gamma-1)=4$, at large Mach numbers.

442: 

443: The post-shock pressure,

444: $\beta_3 = \beta_e[1+\gamma M_2^2(1+v_s/|v_{x,2}|)]$,

445: also follows

446: from the shock relations, and

447: even in the limit of vanishing pre-shock pressure ($\beta_e\ll1$), 

448: it can be significant:

449: $\beta_3\simeq4(\gamma+1)\sin^4(\zeta/2)$.

450: The low-$\beta$ assumption thereby imposes a limit on the reconnection angle; 

451: only for $\zeta<36^{\circ}$ is $\beta_3<0.1$ and 

452: $\beta_3>1$ when $\zeta>67^{\circ}$.

453: 

454: The post-shock temperature is more conveniently 

455: expressed with respect to 

456: $\ta=(m_p/k_B)\vae^2/2$, than to

457: the initial temperature.

458: For example, $15$ G field immersed in 

459: $n_e=3\times10^8\,{\rm cm}^{-3}$ plasma has a characteristic

460: temperature $\ta=2\times10^8$ K.

461: The plasma beta is $\beta_e = T_e/\ta$, so a

462: 2 MK coronal plasma will have $\beta_e=10^{-2}$.

463: Figure

464: \ref{fig:temp}a--b shows the post-shock temperature,

465: $T_3=\ta\beta_3\rho_e/\rho_3$, over a range of

466: $\beta_e$--$\zeta$ parameter space.  For $\beta_e=0.01$ and

467: $\beta_3=1/3$ (the edge of the light grey area, 

468: $\zeta\simeq50^{\circ}$) the 

469: post-shock plasma will be $T_3=0.1\ta=20$ MK.

470: 

471: \begin{figure}[htb]

472: \epsscale{0.75}

473: %\epsscale{0.85}

474: \plotone{fig3.ps}

475: \caption{The post-shock temperature and thermal energy across the

476: $\zeta$--$\beta_e$ parameter space.  The top panel (a) shows contours

477: of $T_3$ whose levels can be determined from intersection with the 

478: left axis ($\zeta=0$) since  $T_3=T_e$ there.

479: Solid contours show  $T_3=0.01,\, 0.1$ and 1 (all in units of $\ta$).

480: Light and dark grey regions show $0.333<\beta_3<1$ and $\beta_3>1$

481: respectively.  The middle panel (b) shows $T_3$ versus $\zeta$ 

482: for the values $T_e=\beta_e=0.01$ (squares) and $0.1$ (triangles).

483: The bottom dashed curve is limit for the case $\beta_e=0$.

484: The bottom panel (c) shows, for the same values of $T_e=\beta_e$, 

485: the fraction of released energy thermalized

486: by the GDS.  Symbols show the same point in each panel.}

487: 	\label{fig:temp}

488: \end{figure}

489: 

490: The total energy release initiated by reconnection,

491: $\Delta W=2\vae^2L\rho_e\sin^2(\zeta/2)$, 

492: is the initial kinetic energy of all 

493: mass within the $L=2\vae t$ of tube affected by retraction.  This 

494: is greater than the energy decrease resulting from 

495: shortening the tube by $\Delta L=2L\sin^2(\zeta/2)$, due to additional

496: work done by the background magnetic field expanding into the vacated

497: volume.  In our idealized model this work exactly doubles the energy,

498: but in more realistic scenarios the factor may be somewhat

499: different.

500: 

501: The moving mass is deflected downward by the

502: GDS, converting a portion of its kinetic

503: energy into thermal energy.  The kinetic energy (per area) 

504: converted to thermal energy, $\rho_3v_{x,2}^2v_st$, constitutes a

505: fraction

506: \be

507:   {\Delta E_t\over\Delta W}=

508:   {\rho_3v_{x,2}^2v_s\over\rho_e|\vvec_2|^2\vae} =

509:   2\left(1+{v_s\over|v_{x,2}|}\right)\sin^4(\zeta/2) ~~,

510: 	\label{eq:ET}

511: \ee

512: of the total (see \fig\ \ref{fig:temp}c).

513: Cases of vanishing initial pressure ($\beta_e\ll1$) will have a

514: fraction $(\gamma+1)\sin^4(\zeta/2)$ of the released energy

515: thermalized.  Compression work done on larger initial pressure will

516: raise this fraction slightly, but all cases with $\beta_3<1/3$ 

517: thermalize less than $10\%$ of the released energy.

518: 

519: \section{Summary}

520: 

521: The foregoing illustrates, through a simplified analytic model, 

522: a process

523: we believe must be common in the flaring corona.  Localized reconnection

524: does not directly dissipate magnetic energy, but rather initiates its

525: release through subsequent shortening of field lines.  This shortening

526: propagates from the reconnection site at the Alfv\'en speed,

527: converting energy from magnetic into kinetic form.  The shortening

528: flux tubes compress plasma within them, raising its temperature as

529: they do so.  Due to the supersonic (Alfv\'enic) flows this compression

530: occurs at strong shocks, which raise the temperature far beyond that from

531: adiabatic compression.

532: 

533: Both the GDS in our model and the SMS in Petshcek's model

534: result, ultimately, from shortening or weakening of magnetic field

535: lines.  The two-dimensional Petschek model permits shortening only

536: perpendicular to the symmetry direction, accompanied 

537: by weakening.  The shock is therefore a SMS whose normal is mostly

538: perpendicular to the sheet.  After three-dimensional patchy

539: reconnection, on the other hand, field lines also shorten in the

540: erstwhile symmetry direction which is the orientation of the shock

541: normal ($\pm\xhat$).

542: The GDSs are disconnected from the diffusion

543: region, and are instead features of the ideal relaxation {\em following}

544: reconnection.  

545: This new scenario requires two-step energy conversion, from magnetic

546: to kinetic to thermal energy, in contrast to the SMSs which

547: thermalize magnetic energy directly where they weaken the field.  

548: % As a consequence, there is no

549: % weakening (indeed no change) in the magnetic field strength at the

550: % GDS, as there must be at the SMS.

551: 

552: \medskip

553: 

554: This work was supported by NASA and the NSF.

555: 

556: %\bibliography{/Users/danalongcope/stuff/short_abbrevs.bib,/Users/danalongcope/stuff/full_lib.bib}

557: 

558: \begin{thebibliography}{}

559: 

560: \bibitem[\protect\citeauthoryear{Biernat, Heyn, \& Semenov}{Biernat

561:   et~al.}{1987}]{Biernat1987}

562: Biernat, H.~K., Heyn, M.~F.,  \& Semenov, V.~S. 1987, JGR, 92, 3392

563: 

564: \bibitem[\protect\citeauthoryear{{Birn} et~al.}{{Birn} et~al.}{2001}]{Birn2001}

565: {Birn}, J., et~al. 2001, JGR, 106, 3715

566: 

567: \bibitem[\protect\citeauthoryear{Biskamp \& Schwarz}{Biskamp \&

568:   Schwarz}{2001}]{Biskamp2001}

569: Biskamp, D.,  \& Schwarz, E. 2001, Phys.~Plasmas, 8, 4729

570: 

571: \bibitem[\protect\citeauthoryear{Cargill \& Priest}{Cargill \&

572:   Priest}{1982}]{Cargill1982}

573: Cargill, P.~J.,  \& Priest, E.~R. 1982, Solar~Phys., 76, 357

574: 

575: \bibitem[\protect\citeauthoryear{Courant \& Friedrichs}{Courant \&

576:   Friedrichs}{1948}]{Courant1948}

577: Courant, R.,  \& Friedrichs, K.~O. 1948, Supersonic Flows and Shock Waves (New

578:   York: Interscience Publishers)

579: 

580: \bibitem[\protect\citeauthoryear{Ferriz-Mas \& Moreno-Insertis}{Ferriz-Mas \&

581:   Moreno-Insertis}{1987}]{Ferriz-Mas1987}

582: Ferriz-Mas, A.,  \& Moreno-Insertis, F. 1987, A\&A, 179, 268

583: 

584: \bibitem[\protect\citeauthoryear{Fisher et~al.}{Fisher

585:   et~al.}{2000}]{Fisher2000}

586: Fisher, G.~H., Fan, Y., Longcope, D.~W., Linton, M.~G.,  \& Abbett, W.~P. 2000,

587:   Phys.~Plasmas, 7, 2173

588: 

589: \bibitem[\protect\citeauthoryear{Grad}{Grad}{1951}]{Grad1951}

590: Grad, H. 1951, Comm. in Pure and App. Math., 5, 257

591: 

592: \bibitem[\protect\citeauthoryear{Herbold et~al.}{Herbold

593:   et~al.}{1985}]{Herbold1985}

594: Herbold, G., Ulmschneider, P., Spruit, H.~C.,  \& Rosnr, R. 1985, A\&A, 145,

595:   157

596: 

597: \bibitem[\protect\citeauthoryear{{Kopp} \& {Pneuman}}{{Kopp} \&

598:   {Pneuman}}{1976}]{Kopp1976}

599: {Kopp}, R.~A.,  \& {Pneuman}, G.~W. 1976, Solar~Phys., 50, 85

600: 

601: \bibitem[\protect\citeauthoryear{Lee et~al.}{Lee et~al.}{1993}]{Lee1993}

602: Lee, L.~C., Ma, Z.~W., Fu, Z.~F.,  \& Otto, A. 1993, JGR, 98, 3943

603: 

604: \bibitem[\protect\citeauthoryear{Linton, De{V}ore, \& Longcope}{Linton

605:   et~al.}{2008}]{Linton2008}

606: Linton, M.~G., De{V}ore, C.~R.,  \& Longcope, D.~W. 2008, Earth, Planets and

607:   Space (in press)

608: 

609: \bibitem[\protect\citeauthoryear{Linton \& Longcope}{Linton \&

610:   Longcope}{2006}]{Linton2006}

611: Linton, M.~G.,  \& Longcope, D.~W. 2006, ApJ, 642, 1177

612: 

613: \bibitem[\protect\citeauthoryear{Otto}{Otto}{1995}]{Otto1995}

614: Otto, A. 1995, JGR, 100, 11863

615: 

616: \bibitem[\protect\citeauthoryear{Petschek}{Petschek}{1964}]{Petschek1964}

617: Petschek, H.~E. 1964, in AAS-NASA Symposium on the Physics of Solar Flares, ed.

618:   W.~N. Hess (Washington, DC: NASA), 425

619: 

620: \bibitem[\protect\citeauthoryear{Priest \& Forbes}{Priest \&

621:   Forbes}{2000}]{Priest2000}

622: Priest, E.~R.,  \& Forbes, T.~G. 2000, Magnetic Reconnection. {MHD} theory and

623:   applications (Cambridge University Press)

624: 

625: \bibitem[\protect\citeauthoryear{Russell \& Elphic}{Russell \&

626:   Elphic}{1978}]{Russell1978}

627: Russell, C.~T.,  \& Elphic, R.~C. 1978, Space Sci. Rev., 22, 681

628: 

629: \bibitem[\protect\citeauthoryear{Semenov, Heyn, \& Kubyshkin}{Semenov

630:   et~al.}{1983}]{Semenov1983}

631: Semenov, V.~S., Heyn, M.~F.,  \& Kubyshkin, I.~V. 1983, Sov.\ Astron., 27, 600

632: 

633: \bibitem[\protect\citeauthoryear{Semenov, Volkonskaya, \& Biernat}{Semenov

634:   et~al.}{1998}]{Semenov1998}

635: Semenov, V.~S., Volkonskaya, N.~N.,  \& Biernat, H.~K. 1998, Phys.~Plasmas, 5,

636:   3242

637: 

638: \bibitem[\protect\citeauthoryear{Spruit}{Spruit}{1981}]{Spruit1981}

639: Spruit, H.~C. 1981, A\&A, 98, 155

640: 

641: \bibitem[\protect\citeauthoryear{Thomas \& Montesinos}{Thomas \&

642:   Montesinos}{1991}]{Thomas1991}

643: Thomas, J.~H.,  \& Montesinos, B. 1991, ApJ, 375, 404

644: 

645: \bibitem[\protect\citeauthoryear{{Warren} et~al.}{{Warren}

646:   et~al.}{1999}]{Warren1999}

647: {Warren}, H.~P., {Bookbinder}, J.~A., {Forbes}, T.~G., {Golub}, L., {Hudson},

648:   H.~S., {Reeves}, K.,  \& {Warshall}, A. 1999, ApJ, 527, L121

649: 

650: \end{thebibliography}

651: 

652: 

653: \end{document}

654: