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

2: \usepackage[dvips]{color}

3: 

4: \renewcommand{\d}{\mbox{d}}

5: \newcommand{\dpp}[2]{\frac{\partial #1}{\partial #2}}

6: \newcommand{\ddt}[1]{\dpp{#1}{t}}

7: \newcommand{\ddtt}[1]{\frac{\partial^2 #1}{\partial t^2}}

8: \newcommand{\ddx}[1]{\dpp{#1}{x}}

9: \newcommand{\ddxx}[1]{\frac{\partial^2 #1}{\partial x^2}}

10: \newcommand{\ddy}[1]{\dpp{#1}{y}}

11: \newcommand{\ddz}[1]{\dpp{#1}{z}}

12: \newcommand{\DDt}[1]{\frac{\d #1}{\d t}}

13: \newcommand{\mi}[1]{\mbox{\boldmath$#1$}}

14: \newcommand{\Nbv}{\mathcal{N}^{\,2}}

15: \newcommand{\Gam}{\Gamma_{\!1}}

16: 

17: \shorttitle{MHD waves in inclined magnetic field}

18: \shortauthors{Parchevsky \& Kosovichev}

19: 

20: \begin{document}

21: \title{Numerical simulation of excitation and propagation of helioseismic

22: MHD waves: Effects of inclined magnetic field}

23: 

24: \author{Parchevsky, K.V., Kosovichev, A.G.}

25: \affil{Stanford University, HEPL, Stanford CA 94305, USA}

26: \email{kparchevsky@solar.stanford.edu}

27: 

28: \begin{abstract}

29: Investigation of propagation, conversion, and scattering of MHD

30: waves in the Sun is very important for understanding the mechanisms

31: of observed oscillations and waves in sunspots and active regions.

32: We have developed 3D linear MHD numerical model to investigate

33: influence of the magnetic field on excitation and properties of the

34: MHD waves. The results show that the magnetic field can

35: substantially change the properties of the surface gravity waves

36: ($f$-mode), but their influence on the acoustic-type waves

37: ($p$-modes) is rather moderate. Comparison our simulations with the

38: time-distance helioseismology results from SOHO/MDI shows that the

39: travel time variations caused by the inclined magnetic field do not

40: exceed 25\% of the observed amplitude even for strong fields of

41: 1400--1900 G. This can be an indication that other effects (e.g.

42: background flows and non-uniform distribution of magnetic field) can

43: contribute to the observed travel time variations. The travel time

44: variations caused by the wave interaction with magnetic field are in

45: phase with the observations for strong fields of 1400--1900 G if

46: Doppler velocities are taken at the height of 300 km above the

47: photosphere where plasma parameter $\beta\ll1$. The simulations show

48: that the travel times only weakly depend on the height of velocity

49: observation. For the photospheric level the travel times are

50: systematically smaller on approximately 0.12 min then for the hight

51: of 300 km above the photosphere for all studied ranges of the

52: magnetic field strength and inclination angles. The numerical MHD

53: wave modeling and new data from the HMI instrument of the Solar

54: Dynamics Observatory will substantially advance our knowledge of the

55: wave interaction with strong magnetic fields on the Sun and improve

56: the local helioseismology diagnostics.

57: 

58: 

59: \end{abstract}

60: 

61: \keywords{Sun: oscillations---sunspots }

62: 

63: \section{Introduction}

64: Local helioseismology has provided important results about the

65: structures and dynamics of the solar plasma below the visible

66: surface of the Sun, associated with sunspots and active regions

67: \citep[e.g.][]{Duvall1996,Kosovichev1996,Kosovichev2000,Zhao2001,Haber2000,Komm2008}.

68: The helioseismic inferences help us to understand the complicated

69: processes of the origin solar magnetic structures, formation and

70: evolution of sunspots and active regions. These studies are based on

71: measurements and inversions of variations of acoustic travel times

72: and oscillation frequencies in the areas occupied by magnetic field

73: and around them.

74: 

75: There are several factors that may cause the observed variations of

76: the oscillation properties, and it is very important to

77: painstakingly investigate their effects for improving the

78: reliability of the helioseismic inference \citep{Bogdan2000}. Such

79: studies are carried out both observationally by doing various

80: experiments with the data analysis procedure, e.g. by masking the

81: regions of strong field, doing "double-skip" experiments etc.

82: \citep[e.g.][]{Zhao2006}, and theoretically by simulating wave

83: propagation in various conditions of the solar convection zone and

84: calculating how these conditions affect the helioseismic

85: observables, such as the oscillation power spectrum and acoustic

86: travel times \citep{Georgobiani2007,Zhao2007}. We emphasize that for

87: correct interpretation of helioseismic results the theoretical

88: modeling must include calculations of the actual observables, taking

89: into account all important aspects of the data measurement

90: procedure, such as data filtering and averaging, and geometrical

91: factors \citep{Nigam2007}. Unfortunately, in many theoretical

92: studies the actual helioseismic measurement procedure is not

93: modeled, and this may lead to incorrect conclusions about the role

94: of various factors in the helioseismic results.

95: 

96: In general, the main factors causing variations in helioseismic

97: travel times in solar magnetic regions, can be divided in two types:

98: direct and indirect. The direct effects are due to the additional

99: magnetic restoring force, which changes the wave speed and may

100: transform acoustic waves into different types of MHD waves. The

101: indirect effects are due to the changes in the convective and

102: thermodynamic properties in magnetic regions. These include

103: depth-dependent variations of temperature and density, large-scale

104: flows and changes in wave source distribution and strength. Both

105: direct and indirect effects may be present in the observed

106: travel-time and frequency variations and cannot be easily

107: disentangled by data analyses causing confusions and

108: misinterpretations. Thus, it is very important to investigate the

109: various aspects by numerically modeling the individual factors

110: separately.

111: 

112: In particular, we have investigated the effects of the suppressed

113: excitation of acoustic waves in sunspot regions, where strong

114: magnetic field inhibits convective motions, which are the primary

115: source of solar waves \citep{Parchevsky2007a}. The results showed

116: that the suppression of acoustic sources may explain most of the

117: observed deficit of acoustic power in sunspot regions

118: \citep{Parchevsky2007b}, and also cause systematic shifts in the

119: travel-time measurements. However, these shifts are significantly

120: smaller than the observed variations of the travel times and and

121: also have the opposite sign \citep{Parchevsky2008}, and, thus,

122: cannot affect the basic conclusions on the sunspot sound-speed

123: structure, contrary to previous suggestions

124: \citep[e.g.][]{Rajaguru2006}.

125: 

126: The goal of this paper is to model the excitation and propagation of

127: helioseismic waves (both f- and p-modes) in the presence of inclined

128: magnetic field and investigate the importance of the inclined field

129: in the time-distance helioseismology measurements by carefully

130: modeling the measurement procedure and comparing with the

131: observational results, obtained by \citet{Zhao2006}. The issue of

132: the influence of the inclined magnetic field was raised by

133: \citet{Schunker2005}, who found that the phase shift of the signal

134: in the penumbra of a sunspot, measured by the acoustic holography

135: technique varies with the sunspot position on the disk. They

136: attributed this to the variations of the angle between the inclined

137: magnetic field of the penumbra and the line-of-sight. They suggested

138: that the variations of the phase shift may affect the inferences of

139: the sound-speed distribution below sunspots, inferred by

140: time-distance helioseismology. However, \citet{Zhao2006} repeated

141: the analysis of the same sunspot by the time-distance technique, and

142: found substantially smaller variations with the position on the

143: disk, and no significant effect on the wave-speed profile. They also

144: found that the variations due to the inclination angle exist only

145: for the wave measurements using the Doppler-shift signal, and that

146: the variations are absent when the travel times are measured from

147: the simultaneous intensity observations from SOHO/MDI. This result

148: indicates that the observed variations with the inclination angle of

149: the Doppler-shift measurements are likely to be related to changes

150: of the ratio between the vertical and horizontal components of the

151: displacement vector of the solar oscillations in the penumbra, and

152: not the wave transformation or other effects, which could affect the

153: modal structure of the oscillations. The solar oscillation theory

154: predicts that the ratio between the vertical and horizontal

155: components mostly depends on the surface boundary conditions

156: \citep[e.g.][]{Unno1989}. In the sunspot umbra, the boundary

157: conditions may change due to the inclined magnetic field or/and near

158: surface flows, the Evershed effect, which is observed directly in

159: the Doppler-shift data and shows a significant center-to-limb

160: variation.

161: 

162: In this paper, we present the results of numerical modeling of the

163: inclined magnetic field on the time-distance helioseismology

164: measurements by isolating this affect in a simple magnetic

165: configuration, and show that only 25\% of travel time variations

166: measured by \citet{Zhao2006} can be explained by a direct influence

167: of the inclined magnetic field on acoustic waves. In Sec. 2, we

168: present the governing equation and describe the numerical method of

169: 3D modeling of helioseismic MHD waves. In Sec.~3, we present the

170: code verification results comparing the numerical results with

171: analytical solutions for simple cases. In Sec.~4, we present the

172: simulation results of the wave propagation in regions with inclined

173: magnetic field, calculation of the center-to-limb variations of the

174: time-distance helioseismology measurements and comparison with the

175: observational results.

176: 

177: \section{Numerical Model}

178: 

179: \subsection{Governing equations and numerical scheme}

180: 

181: Propagation of MHD waves inside the Sun in the presence of magnetic

182: field is described by the following system of linearized MHD

183: equations:

184: \begin{equation}\label{Eq:MHD_3D}

185: \begin{array}{l}

186: \displaystyle \ddt{\rho'} + \nabla\cdot \mi{m}'=0,\\[12pt]

187: \displaystyle \ddt{\mi{m}'} + \nabla p' - \frac{1}{4\pi\rho_0}

188: [(\nabla\times \mi{B}_0)\times\mi{B}' + (\nabla\times

189: \mi{B}')\times\mi{B}_0] = \mi{g}_0\rho' + \mi{S},\\[12pt]

190: \displaystyle \ddt{p'} + c_{s0}^2\nabla\cdot \mi{m}' +

191: c_{s0}^2\frac{\Nbv_0}{g_0}m_z = 0,

192: \end{array}

193: \end{equation}

194: where $\mi{m}'=\rho_0\mi{v}'$ is the momentum perturbation,

195: $\mi{v}'$, $\rho'$, $p'$, and $\mi{B}'$ are the velocity, density,

196: pressure, and magnetic field perturbations respectively, $\mi{S}$ is

197: the wave source function. The quantities with subscript 0, such as

198: gravity $\mi{g}_0$, sound speed $c_{s0}$, and Br\"unt-V\"ais\"al\"a

199: frequency $\mathcal{N}_0$ correspond to the background model. The

200: spatial and temporal behavior of the wave source is modeled by

201: function $f(x,y,z,t)$:

202: \begin{equation}

203: f(x,y,z,t)=\left\{

204: \begin{array}{ll}

205: \displaystyle

206: A\left[1-\frac{r^2}{R_{src}^2}\right]^2\left(1-2\tau^2\right)e^{-\tau^2} & \mbox{if } r\leq R_{src}\\

207: \displaystyle 0 & \mbox{if } r>R_{src},

208: \end{array}

209: \right.

210: \end{equation}

211: where $R_{src}$ is the source radius,

212: $r=\sqrt{(x-x_{src})^2+(y-y_{src})^2+(z-z_{src})^2}$ is the distance

213: from the source center, $\tau$ is given by equation

214: \begin{equation}\label{Eq:MHD_3Dsource}

215: \tau=\frac{\omega (t-t_0)}{2} - \pi, \qquad t_0\leq t\leq

216: t_0+\frac{4\pi}{\omega},

217: \end{equation}

218: where $\omega$ is the central source frequency, $t_0$ is the moment

219: of the source initiation. In our simulations we used sources of two

220: types: source of vertical force $\mi{S}=(0,0,f)^T$ and pressure

221: source $\mi{S}=\nabla f$. Superposition of such randomly distributed

222: sources describes very well the observed solar oscillation spectrum

223: \citep{Parchevsky2007a}.

224: 

225: For numerical solution of equations (\ref{Eq:MHD_3D}) a

226: semi-discrete finite difference scheme of high order was used. At

227: the top and bottom boundaries non-reflective boundary conditions

228: based on the perfectly matched layer (PML) technique were set. We

229: used the standard solar model S \citep{Christensen-Dalsgaard1996}

230: with a smoothly joined model of the chromosphere of

231: \citet{Vernazza1976}, as the background model. The background model

232: was modified near the photosphere to make it convectively stable.

233: Details of numerical realization of the code and the background

234: model can be found in \citet{Parchevsky2007a}.

235: 

236: 

237: \subsection{Code verification for different types of MHD waves}

238: To verify the code and estimate the total error of the method we

239: compare numerical results with simple analytical solutions. We

240: assume that all quantities depend on time $t$ and one spatial

241: coordinate $x$ only, and consider the following linearized adiabatic

242: 1D system of the MHD equations in Cartesian coordinates for a

243: uniform background model

244: \begin{equation}\label{Eq:MHD_1DxEq}

245: \begin{array}{l}

246: \displaystyle \ddt{\rho}+\bar{\rho}\ddx{u}=0,\\[12pt]

247: \displaystyle \ddt{u} + \frac{1}{\bar\rho}\ddx{p} +

248: \frac{\bar{B}_z}{4\pi\bar{\rho}}\ddz{B_z}=0,\\[12pt]

249: \displaystyle \ddt{v} - \frac{\bar{B}_x}{4\pi\bar{\rho}}\ddx{B_y}=0,\\[12pt]

250: \displaystyle \ddt{w} - \frac{\bar{B}_x}{4\pi\bar{\rho}}\ddx{B_z}= 0,\\[12pt]

251: \displaystyle \ddt{B_y}-\bar{B}_x\ddx{v} =0,\\[12pt]

252: \displaystyle \ddt{B_z} + \bar{B}_z\ddx{u} - \bar{B}_x\ddx{w}

253: =0,\\[12pt]

254: \displaystyle \ddt{p}+\bar c_s^2\bar\rho\ddx{u} = 0,

255: \end{array}

256: \end{equation}

257: where $\bar c_s$ is the sound speed in the background model, $p$,

258: $\rho$ are the pressure and density, $u$, $v$, and $w$ are the

259: velocity components, $B_x$, $B_y$, $B_z$ are the components of the

260: magnetic field respectively. The quantities related to the

261: background model are marked here by overbar. By rotating of the

262: coordinate frame around OZ axis we set $\bar{B}_y=0$. Equations

263: (\ref{Eq:MHD_1DxEq}) can be rewritten in matrix notations

264: \begin{equation}\label{Eq:MHD_1DxMat}

265: \begin{array}{l}

266: \displaystyle \ddt{\mi{U}}+\mi{A}\ddx{\mi{U}}=0,

267: \end{array}

268: \end{equation}

269: where $\mi{U}=(\rho,u,v,w,B_y,B_z,p)^T$. We seek a solution of

270: equations (\ref{Eq:MHD_1DxMat}) in infinite interval $-\infty < x <

271: \infty$ in the form of plane waves

272: \begin{equation}\label{Eq:MHD_1DxPlnWaveSolution}

273: \mi{U}=\hat{\mi{U}}\exp\left[i(kx-\omega t)\right]

274: \end{equation}

275: Substituting equation (\ref{Eq:MHD_1DxPlnWaveSolution}) into system

276: (\ref{Eq:MHD_1DxMat}) we obtain an eigenvalue problem for amplitude

277: $\hat{\mi{U}}$

278: \begin{equation}

279: \mi{A}\hat{\mi{U}} = \frac{\omega}{k}\hat{\mi{U}},

280: \end{equation}

281: where $V=\omega/k$ is the phase speed of the wave. Thus, the

282: amplitudes of various MHD quantities are the components of

283: eigenvectors of matrix $\mi{A}$, and eigenvalues of this matrix

284: represent phase velocities of the corresponded waves. From this we

285: calculate the amplitudes of the entropy, Alfven, slow, and fast MHD

286: waves:

287: \begin{equation}\label{Eq:MHD_1DxAmplitudes}

288: \begin{array}{ccccccc}

289: \hat{\mi{U}}_s=(\bar\rho,&0,&0,&0,&0,&0,&0)^T\\[5pt]

290: \hat{\mi{U}}_{\mp A}=(0,&0,&\pm V_A,&0,&\bar{B}_x,&0,&0)^T\\[5pt]

291: \hat{\mi{U}}_{\mp S}=(\bar\rho,&\mp V_S,&0,

292: &\displaystyle\mp\frac{c_A^2V_S\sin\theta\cos\theta}{V_A^2-V_S^2},&0,

293: &\displaystyle-\frac{B_0V_S^2\sin\theta}{V_A^2-V_S^2},&c_s^2\bar\rho)^T\\[12pt]

294: \hat{\mi{U}}_{\mp F}=(\bar\rho,&\mp V_F,&0,

295: &\displaystyle\mp\frac{c_A^2V_F\sin\theta\cos\theta}{V_A^2-V_F^2},&0,

296: &\displaystyle-\frac{B_0V_F^2\sin\theta}{V_A^2-V_F^2},&c_s^2\bar\rho)^T.

297: \end{array}

298: \end{equation}

299: The phase velocities of these waves are $0$, $\mp V_A$, $\mp V_S$,

300: and $\mp V_F$ respectively, where

301: \begin{equation}\label{Eq:MHD_1DxEigs}

302: \begin{array}{l}

303: V_A=c_A\cos\theta,\\

304: \displaystyle

305: V_S=\frac{1}{\sqrt{2}} \sqrt{c_s^2+c_A^2-\sqrt{c_s^4+c_A^4-2c_s^2c_A^2\cos2\theta}},\\[12pt]

306: \displaystyle V_F=\frac{1}{\sqrt{2}}

307: \sqrt{c_s^2+c_A^2+\sqrt{c_s^4+c_A^4-2c_s^2c_A^2\cos2\theta}}.

308: \end{array}

309: \end{equation}

310: Here $c_A=\bar B/\sqrt{4\pi\bar\rho}$ is the Alfven speed, $\theta$

311: is the angle between the wave vector and the background magnetic

312: field ($\bar B_x=\bar B\cos\theta,\; \bar B_z=\bar B\sin\theta$).

313: Equations (\ref{Eq:MHD_1DxPlnWaveSolution}),

314: (\ref{Eq:MHD_1DxAmplitudes}), and (\ref{Eq:MHD_1DxEigs}) give us an

315: analytical solution to 1D system of MHD equations

316: (\ref{Eq:MHD_1DxEq}), which includes all types of MHD waves. Initial

317: conditions for these waves are obtained by setting $t=0$ in equation

318: (\ref{Eq:MHD_1DxPlnWaveSolution}).

319: 

320: For testing the numerical simulations we use our 3D code with

321: initial conditions depending only on $x$ variable. The calculations

322: are carried out in the Cartesian geometry in the domain of

323: 15.46~Mm$\times$15.46~Mm$\times$3.05~Mm with the numbers of grid

324: points: $N_x=N_y=104,\; N_z=71$. All boundary conditions are chosen

325: periodic simulating an infinite spatial domain. Wave vector

326: $k=10\pi/(N_x+1)\Delta x$ is chosen in a way to match the periodic

327: boundary conditions. In dimensionless variables

328: $\tilde\rho=\rho/\bar\rho$, $\tilde p=p/\bar\rho \bar c_s^2$,

329: $\tilde{\mi{v}}=\mi{v}/\bar c_s$, and $\tilde B^2=B^2/4\pi\bar\rho

330: \bar c_s^2$ parameters of the background model are

331: \begin{equation}

332: \bar\rho=1,\quad \bar c_s=1,\quad \bar B = 0.5,\quad c_A =0.5,\quad

333: \theta=\pi/4.

334: \end{equation}

335: The amplitude of the Alfven wave (in chosen dimensionless variables)

336: traveling in the positive direction of the $x$-axis is

337: $\hat{\mi{U}}_{+A}=(0,\:0,\:-1,\:0,\:1,\:0,\:0)^T$. For the slow and

338: fast MHD waves traveling in the same direction these amplitudes are

339: $\hat{\mi{U}}_{+S}=(1,\:0.33108,\:0,\:2.6894,\:0,\:-2.5184,\:1)^T$

340: and

341: $\hat{\mi{U}}_{+F}=(1,\:1.0679,\:0,\:-0.13146,\:0,\:0.39708,\:1)^T$

342: respectively. The cyclic frequencies of these waves are

343: $\omega_A=kV_A$, $\omega_S=kV_S$, and $\omega_F=kV_F$ respectively.

344: The results of our numerical and analytical solutions for the moment

345: of time $t$ = 20 min. for all three waves are shown in Figure

346: \ref{Fig:1Dtest_PlnWaves}. Panels a, b, and c represent the results

347: for the Alfven, slow, and fast MHD waves respectively. Only

348: variables $v$ and $B_y$ are nonzero in the Alfven wave. They are

349: shown by solid and dashed curves in panel a) respectively. The exact

350: solution for $v$ is shown by circles. In the slow and fast MHD waves

351: variables $\rho$, $u$, $w$, $B_z$, and $p$ are all nonzero. The

352: dimensionless pressure coincides in amplitude and phase with the

353: density and is not shown in Figure \ref{Fig:1Dtest_PlnWaves}.

354: Variables $\rho$, $u$, $w$, and $B_z$ are shown in panels b) and c)

355: by solid, dash-dotted, dashed, and dotted curves respectively. The

356: exact analytical solution for $w$ is shown by circles. The

357: analytical solutions for other variables coincide with the numerical

358: curves and thus are not shown. We see that our numerical solutions

359: reproduce the amplitudes, phases, and velocities of the Alfven,

360: slow, and fast MHD waves very well.

361: 

362: 

363: \section{Results and discussion}

364: \subsection{MHD waves generated by a single source in uniform and non-uniform

365: background magnetic field} In this section we present our results of

366: numerical simulation of excitation and propagation of MHD waves

367: generated by a single source of vertical force with central

368: frequency $\nu$ = 3.5 mHz placed at depth $h_{src}=100$ km in a

369: rectangular region of size 15.5$\times$15.5$\times$12.5 Mm$^3$

370: (104$\times$104$\times$70 nodes). The horizontal grid is uniform

371: with $\Delta x=\Delta y = 150$ km. The vertical grid is non-uniform.

372: The grid step $\Delta z$ varies from 50 km near the photosphere to

373: 600 km near the bottom of the computational domain. Time step

374: $\Delta t=0.5$ s was chosen to satisfy the Courant stability

375: condition. Vector $\mi{B}_0=(B_0 \sin\gamma,0,B_0\cos\gamma)^T$ of

376: the uniform inclined background magnetic field lies in XZ plane and

377: has inclination angle of $\gamma$ = 45$^\circ$ with respect to the

378: top boundary normal. The magnetic field strength, $B_0$, varied from

379: 0 to 2500 G in our simulations. The ratio of the gas pressure to the

380: magnetic pressure, plasma parameter $\beta$ for typical sunspot

381: penumbra values is shown in Fig.~\ref{Fig:beta}.

382: 

383: We describe here two examples of wave propagation in the uniform

384: background magnetic field with a wave source located inside the

385: magnetic region, and of a non-uniform magnetic field with the

386: source located outside the magnetic regions. Since the convective

387: motions on the Sun are inhibited in strong magnetic field regions,

388: the acoustic sources there are suppressed. Thus, the second case

389: better describes the realistic situation on the Sun.

390: 

391: The simulation results for the uniform inclined magnetic field of

392: $B_0 = 625$ G are shown in Fig.~\ref{Fig:SingleSrc_UniB}. The

393: vertical map of $B_z'$ (panel c, right) reveals a strong Alfven

394: wave, which propagates along the background inclined magnetic field

395: lines. As expected, the Alfven wave is not presented in the map of

396: density perturbations. The Alfven wave is generated due to the

397: interaction between the wave source and the magnetic field at the

398: source location. The concentric waves in the left panels represent a

399: mixture of the fast MHD wave (analogous to p-modes in absence of the

400: magnetic field) and the surface magnetic-gravity wave (analogous to

401: f-mode). Since the wave speed depends on the angle between the

402: vectors of magnetic field vector and wavenumber, the wave fronts are

403: anisotropic. The separation of these two types of waves will be

404: discussed in Sec.~3.2 together with analysis of phase and group

405: travel times.

406: 

407: The simulation results for the second case, when the source is

408: placed outside the region containing the background magnetic field

409: are shown in Figure \ref{Fig:SingleSrc_NonUnB}. The domain size and

410: grid parameters are the same, as in Figure \ref{Fig:SingleSrc_UniB}.

411: The background magnetic field $\mi{B}_0=(0,B_0

412: \sin\gamma,B_0\cos\gamma)^T$ lies in YZ plane, has inclination angle

413: $\gamma$ = 45$^\circ$ with the top boundary normal and is parallel

414: to the planes showed in left columns of figure  by dashed lines. The

415: magnetic field $B_0$ has constant values of 0 and 2500 G in the

416: regions on the left and right of the dashed lines respectively.

417: Between the dashed lines the strength of the background magnetic

418: field is linearly decreases from maximum value to zero. Such

419: configuration of the background field satisfies condition

420: $\mbox{div} \mi{B}_0=0$. The magnetic field strength in this example

421: is chosen higher than the typical penumbra value for a better

422: demonstration of the magnetic effects.

423:  The wave source is placed in the region

424: free from the background magnetic field. The waves generated by such

425: a source are pure acoustic and surface gravity waves. When they

426: enter the region occupied by the inclined magnetic field they are

427: transformed into the fast MHD and slow magneto-gravity waves

428: respectively. The Alfven wave does not appear in these simulations.

429: Evidently, the fast MHD wave travels faster than the original

430: acoustic wave, and its amplitude is reduced. In these simulations we

431: do not notice significant wave transformation effects in the near

432: surface reflection layers where the plasma parameter, $\beta$, equal

433: to 1. The helioseismic waves are trapped below the surface and

434: according to our simulations are not affected by transformation into

435: other types (slow MHD and Alfven modes). We will discuss in more

436: detail the role of the transformation for waves of different

437: frequencies in a future publication. The main purpose of this paper

438: is to discuss the effects of the inclined magnetic field on the

439: observed travel time variations in the sunspot penumbra.

440: 

441: 

442: \subsection{Phase and group travel time variations along the wavefront}

443: To study the travel time variations we performed simulations in a

444: rectangular box of size 48$\times$48$\times$12.5 Mm$^3$

445: (320$\times$320$\times$70 nodes) for different values $B_0$ and

446: inclination angles $\gamma$ of the uniform background magnetic

447: field: (625 G, 70$^\circ$), (1400 G, 45$^\circ$), and (1900 G,

448: 30$^\circ$). The grid step size, source type and depth were chosen

449: the same as in previous Section 3.1. Total simulation time equals 6

450: hours of solar time. The simulation results for $B_0$=625 G and

451: $\gamma$=70$^\circ$ are shown in Figure \ref{Fig:kw_diagr}. The top

452: row represents $k$--$\nu$ diagrams, the bottom one shows

453: corresponded horizontal snapshots of the z-component of velocity at

454: the level of the photosphere for the moment of time $t=30$ min. The

455: usual technique of fitting of the cross-covariance function by

456: Gabor's wavelet was used to calculate the travel times. This

457: technique was developed for $p$-modes \citep{Kosovichev1997}. The

458: source of the vertical (z-component) of force generates a strong

459: gravity wave (the lowest ridge of $k$--$\nu$ diagram on panel a

460: corresponds to the $f$-mode) which has to be filtered out. Results

461: of separation of $p$- and $f$-modes are shown in panels (b) and (c)

462: respectively. The maps of $V_z$ for $p$- and $f$-modes (bottom row)

463: were obtained by taking the inverse Fourier transform of the

464: corresponding 3D spectra. It is clear, even without applying an

465: $f$-mode filter, that starting form the distance of about 15 Mm from

466: the source the $p$- and $f$-modes are spatially separated due to the

467: different velocities. Due to the different dispersion relations the

468: magnetic-gravity and fast MHD waves easily separated on

469: time-distance diagram (see Figure \ref{Fig:rho_TD}). The solid black

470: curve represents a theoretical time-distance curve for $p$-modes and

471: standard solar model in absence of the magnetic field. We see, that

472: the fast MHD waves are not significantly affected by the magnetic

473: field and follow the theoretical $p$-mode curve, calculated for the

474: quiet Sun in the ray approximation \citep{Kosovichev1997}. The

475: magnetic-gravity waves have characteristic "zebra" structure due to

476: the difference between phase and group velocities. Their speed is

477: less than the speed of the fast MHD wave.

478: 

479: The magnetic field results in anisotropy of the wave properties.

480: Therefore, the wave travel times measured from the line-of-sight

481: component of the displacement velocity depend on the direction of

482: the wave propagation and also on the viewing angle. We have

483: investigate this effect for the case of the uniform inclined

484: magnetic field. The choice of the coordinate system and geometry are

485: shown in Figure \ref{Fig:LoS_geometry}. Horizontal XY-plane

486: coincides with the photosphere. The origin of the Cartesian

487: coordinate system is placed at the point O above the wave source

488: (the source itself is at the depth of 100 km below the photosphere).

489: The uniform inclined background magnetic filed lies in the XZ-plane

490: ($B_y=0$) and has angle $\gamma=45^\circ$ with the normal to the

491: photosphere. The location of point of observations

492: $P_{\mathrm{obs}}$ is defined by the distance $\Delta$ from the wave

493: source and the azimuthal angle $\alpha$. The line-of-sight (LoS)

494: direction is defined by two angles: angle $\theta$ between LoS

495: direction and local normal, and azimuthal angle $\psi$ between OX

496: axis and projection of the LoS on the local horizontal plane. First,

497: we build 3D $k$--$\nu$ diagrams for each velocity component $u$,

498: $v$, and $w$, filter out the $f$-modes as shown in Figure

499: \ref{Fig:kw_diagr} and calculate the Cartesian components of

500: velocity perturbation $u_p$, $v_p$, and $w_p$ for $p$-modes by

501: taking the inverse Fourier transform of corresponding spectra. Then,

502: we calculate cross covariance

503: \begin{equation}

504: C(\mi{r}_1,\mi{r}_2,\tau)=\frac{1}{T}\int_0^T

505: v_{LoS}^{(p)}(\mi{r}_1,t) v_{LoS}^{(p)}(\mi{r}_2,t+\tau) dt

506: \end{equation}

507: of LoS velocities

508: \begin{equation}

509: v_{LoS}^{(p)}=u_p\cos\psi\sin\theta + v_p\sin\psi\sin\theta +

510: w_p\cos\theta

511: \end{equation}

512: in the origin of the system of coordinates and in the observational

513: point $P_{\mathrm{obs}}$ ($\Delta=7.9$ Mm). Then we fit the cross

514: covariance function with Gabor's wavelet

515: \begin{equation}

516: G(\Delta,\tau)=A\cos[\omega_0(\tau-\tau_p)]\exp\left[

517: -\frac{\delta\omega^2}{4}(\tau-\tau_g)^2\right],

518: \end{equation}

519: where $\Delta=|\mi{r}_1 - \mi{r}_2|$ is the distance between points

520: where LoS velocities $v_{LoS}^{(p)}(\mi{r}_1,t)$ and

521: $v_{LoS}^{(p)}(\mi{r}_2,t)$ are measured, $A$ is the amplitude,

522: $\omega_0$ is the central frequency, $\tau_p$ and $\tau_g$ are the

523: phase and group travel times respectively, and $\delta\omega$ is the

524: bandwidth. Parameters $A$, $\omega_0$, $\tau_p$, $\tau_g$, and

525: $\delta\omega$ are free and have to be determined from the fitting

526: procedure. Repeating this procedure for different observational

527: points with the same $\Delta$ but different azimuthal angle we

528: obtain $\tau_p$ and $\tau_g$ as functions of $\alpha$.

529: 

530: Travel time variations along the wave front for different $B_0$ and

531: different LoS angles are shown in Figure \ref{Fig:TravelTimes_alph}.

532: Panels a, b, and c correspond to $B_0$ of 625 G, 1400 G, and 1900 G

533: respectively. The solid, dashed, and dash-dotted curves correspond

534: to $\psi$ = \{0$^\circ$, 90$^\circ$, 180$^\circ$\} respectively. The

535: angle between the LoS and the local normal is $\theta=20.5^{\circ}$.

536: The mean phase travel times show variations of about 0.5 min along

537: the wave front, the variation amplitude is a little smaller for

538: strong (1400--1900 G) magnetic fields than for the weak 625 G field.

539: For the weak magnetic field changes of $\psi$ change the shape of

540: the curve, but not its average value. The strong magnetic fields

541: cause anisotropy, and the mean value of the travel times changes

542: with angle $\psi$. Thus, averaging along the wave front for the

543: strong magnetic fields gives variations of the observed mean travel

544: times of about 0.1--0.3 min with angle $\psi$ (see detailed

545: discussion in the next section).

546: 

547: 

548: \subsection{Azimuthal dependence of phase travel times in sunspots:

549: comparison with observations} \citet{Zhao2006} observed different

550: behavior of azimuthal dependence of phase travel times obtained from

551: the SOHO/MDI Doppler shift data \citep{Scherrer1995} in sunspots

552: depending on their position on the disk. As far as we want to

553: reproduce similar conditions (and geometry) in our simulations we

554: give a detailed description of their algorithm of the phase travel

555: time calculation. We will apply the same technique to our simulated

556: data.

557: 

558: The sunspot is located at latitude $\phi$ and longitude $\lambda$ on

559: the eastern part of the disk as shown in Figure

560: \ref{Fig:SunSpot_geometry}. The sunspot meridian plane is shown by

561: gray color. Orth $\mi{i}$ of the global system of coordinate with

562: the origin in the center of the Sun is aimed at the center of the

563: visible solar disk, orth $\mi{k}$ points to the north, and orth

564: $\mi{j}$ points to the west. The local system of coordinates with

565: the origin in the center of the sunspot is chosen in such a way that

566: $\mi{e}_z$ coincides with the local normal to the surface,

567: $\mi{e}_x$ is directed to the west, and $\mi{e}_y$ is directed along

568: the meridian and forms a right vector triplet with $\mi{e}_z$ and

569: $\mi{e}_y$.

570: 

571: Observational point $P_{\mathrm{obs}}$ is chosen inside the sunspot

572: penumbra. Azimuthal angle $A$ is counted counterclockwise from local

573: west direction $\mi{e}_x$. Two signals are calculated: the LoS

574: velocity at $P_{\mathrm{obs}}$ and the LoS velocity averaged along

575: the annulus (between inner and outer radii of a ring with average

576: radius $\Delta$ = 8 Mm with the origin in the observational point).

577: Cross covariance of these two signals is fit with the Gabor’s

578: wavelet. Fitting procedure gives us mean phase and group travel

579: times. We assume, that the annulus is small enough that sunspot

580: magnetic field $\mi{B}_0$ inside the annulus can be considered as

581: uniform. Hence the problem is reduced to simulations with the

582: uniform inclined magnetic field shown in Figure \ref{Fig:kw_diagr}.

583: To compare results of the simulations with the observations we have

584: to find from what direction we have to look at our simulation domain

585: to match the observations. In other words, we have to find a

586: relation between azimuthal angle $A$ of the sunspot center and

587: $\psi$. Angle $\psi$ here has the same meaning as in Figure

588: \ref{Fig:LoS_geometry}. This is the angle between the projection of

589: the background magnetic field on the local horizon (XY-plane) and

590: projection of the LoS on the same plane. The unit vector in LoS

591: direction in the sunspot local system of coordinates is given by

592: equation

593: \begin{equation}

594: \mi{i}=\sin\lambda\;\mi{e}_x -\sin\phi\cos\lambda\; \mi{e}_y +

595: \cos\phi\cos\lambda\; \mi{e}_z.

596: \end{equation}

597: The projection of $\mi{i}$ on local horizontal plane (defined by

598: vectors $\mi{e}_x$ and $\mi{e}_y$) has coordinates

599: $\mi{i}_p=(-\sin\lambda,-\sin\phi\cos\lambda,0)^T$ (non-unit

600: vector). Angle $\psi$ is given by equation

601: \begin{equation}

602: \psi=-A-\angle\mi{i}_p\mi{e}_x=-A-\arccos\left(

603: \frac{-\sin\lambda}{\sqrt{\sin^2\lambda+\sin^2\phi\cos^2\lambda}}\right).

604: \end{equation}

605: 

606: We performed numerical simulations of propagation of MHD waves in

607: presence of the uniform inclined magnetic field for different

608: inclination angles and strengths of the magnetic field. The goal of

609: these simulations was to calculate contribution of the inclined

610: magnetic field effect to variations of the mean travel times with

611: the azimuthal angle. This is why we used horizontally uniform

612: standard solar model as a background model. The same model was used

613: to obtain Figures \ref{Fig:SingleSrc_UniB},

614: \ref{Fig:SingleSrc_NonUnB}. The mean travel times obtained for the

615: photospheric level (panel a) and the level of 300 km above the

616: photosphere (panel b) are shown in Figure \ref{Fig:TravTimeAvr_A}.

617: The solid curve corresponds to the magnetic field $B_0=625$ G with

618: inclination angle $\gamma=70^\circ$. The dashed and dash-dotted

619: curves are corresponded to cases $B_0=1400$ G, $\gamma=45^\circ$ and

620: $B_0=1900$ G, $\gamma=30^\circ$ respectively. The systematic shift

621: of the calculated mean travel times with respect to the observations

622: is caused by temperature effects (the sound speed is smaller inside

623: sunspots than in the quiet Sun), which are not included in these

624: simulations. The travel times systematically decrease when the

625: strength of the magnetic field increases because the speed of the

626: fast MHD wave increases with the magnetic field. For studying the

627: inclined field effect we are mostly interested in relative

628: variations of the mean travel times, provide the absolute values to

629: give a general impression about the magnitude of contributions to

630: the mean travel times from the temperature effect and variations of

631: the fast MHD wave speed with the magnetic field.  A model of a

632: sunspot with both temperature and magnetic field effect will be

633: presented in a future publication.

634: 

635: The amplitude of the mean travel time variations for weak (625 G)

636: magnetic field is 10 times smaller than the observed quantity for

637: the height of 300 km and even smaller for the photospheric level.

638: For stronger magnetic fields (1400--1900 G) the amplitude of travel

639: time variations is about 25\% of the observed amplitude. Behavior of

640: the travel times depends on the magnetic field strength and the

641: level of observation of Doppler velocities. For the weak magnetic

642: field (625 G) the phase of the travel time variations is opposite to

643: the observations for both levels. For the height of 300 km above the

644: photosphere where plasma parameter $\beta\ll1$ simulated travel

645: times show the same phase as in observations for both magnetic field

646: strengths (1400 G and 1900 G). For the photospheric level and field

647: strength of 1900 G the phase of travel time variations coincides

648: with the observations, while the travel time variations for

649: $B_0=1400$ G are in antiphase with the observations. Comparison of

650: the curves at different heights for the same inclination angles and

651: magnetic field strengths shows that for the photospheric layer the

652: mean travel times are systematically smaller by only about 0.12 min

653: than for the layer of 300 km above the photosphere. Thus, the

654: variations in the height of Doppler shift measurement do not have a

655: significant effect.

656: 

657: The curves in Figure \ref{Fig:TravTimeAvr_A} were obtained for the

658: annulus radius of 8 Mm and the sunspot located at heliospheric

659: latitude $\phi=19.5^\circ$ and longitude $\lambda=-6.5^\circ$. The

660: angle between LoS and local normal to the photosphere at the center

661: of the sunspot ($\theta$ angle) is $20.5^\circ$. The parameters are

662: chosen to match corresponded angles for Figure 1b of

663: \citet{Zhao2006}.

664: 

665: 

666: \section{Conclusion}

667: The numerical 3D simulations of excitation and propagation of

668: magneto-acoustic-gravity waves in the solar interior and atmosphere

669: show that the presence of the inclined background magnetic field

670: significantly alters properties of $f$-modes but has little effect

671: on $p$-modes. The interaction of the wave source with the background

672: magnetic field generates a strong Alfven wave. However, the Alfven

673: wave does not appear when the wave source is located outside the

674: magnetic region, and the acoustic-gravity waves propagate from the

675: region without magnetic field into the region with magnetic field.

676: 

677: The helioseismic travel times, obtained from cross covariance of the

678: $p$-mode LoS velocities at the observation point and the source

679: point, show variations of about 1 min along the wave front (the

680: amplitude depends on inclination $\theta$ of the LoS). Due to the

681: anisotropy, the travel time averaged along the wave front (like in

682: the observational procedure) is not zero. Comparison of the

683: variations of the mean travel times vs. the azimuthal angle of the

684: observing point shows that the simulation results are in phase with

685: the observations when Doppler velocities are taken at the level of

686: 300 km above the photosphere (at the same height as the observed

687: velocities are obtained). The travel time weakly depends on the

688: height of observations. The amplitude of variations of the travel

689: times obtained form simulations is about 25\% of the observed

690: amplitude even for strong fields of 1400--1990 G. It can be an

691: indication that other effects (for example background flows or

692: non-uniform distribution of magnetic field) can contribute to the

693: observed travel time variations. The developed 3D MHD wave

694: propagation code provides an important tool for further

695: investigations of local helioseismology in regions with the strong

696: magnetic field.

697: 

698: \section{Acknowledgements}

699: We thank to Junwei Zhao for providing us data of variation of the

700: mean travel times obtained from observations of sunspots.

701: Calculations were carried out on Columbia supercomputer of NASA Ames

702: Research Center.

703: 

704: \begin{thebibliography}{99}

705: 

706: \bibitem[Bogdan(2000)]{Bogdan2000} Bogdan T.J. \ 2000, \solphys, 129,

707: 373

708: 

709: \bibitem[Christensen-Dalsgaard et al.(1996)]{Christensen-Dalsgaard1996}

710: Christensen-Dalsgaard, J., et al.\ 1996, Science, 272, 1286

711: 

712: \bibitem[Duvall et al.(1996)]{Duvall1996} Duvall, T.~L.~J.,

713: D'Silva, S., Jefferies, S.~M., Harvey, J.~W., \& Schou, J.\ 1996,

714: \nat, 379, 235

715: 

716: \bibitem[Georgobiani et al.(2007)]{Georgobiani2007} Georgobiani, D.,

717: Zhao, J., Kosovichev, A.~G., Benson, D., Stein, R.~F., \& Nordlund,

718: {\AA}.\ 2007, \apj, 657, 1157

719: 

720: \bibitem[Haber et al.(2000)]{Haber2000} Haber, D.~A., Hindman,

721: B.~W., Toomre, J., Bogart, R.~S., Thompson, M.~J., \& Hill, F.\

722: 2000, \solphys, 192, 335

723: 

724: \bibitem[Komm et al.(2008)]{Komm2008} Komm, R., Morita, S.,

725: Howe, R., \& Hill, F.\ 2008, \apj, 672, 1254

726: 

727: \bibitem[Kosovichev(1996)]{Kosovichev1996} Kosovichev, A.~G.\ 1996,

728: \apjl, 461, L55

729: 

730: \bibitem[Kosovichev \& Duvall(1997)]{Kosovichev1997} Kosovichev, A.~G., \& Duvall,

731: T.~L., Jr.\ 1997, SCORe'96 : Solar Convection and Oscillations and

732: their Relationship, 225, 241

733: 

734: \bibitem[Kosovichev et al.(2000)]{Kosovichev2000} Kosovichev, A.~G.,

735: Duvall, T.~L.~\_.~J., \& Scherrer, P.~H.\ 2000, \solphys, 192, 159

736: 

737: \bibitem[Nigam et al.(2007)]{Nigam2007} Nigam, R., Kosovichev,

738: A.~G., \& Scherrer, P.~H.\ 2007, \apj, 659, 1736

739: 

740: \bibitem[Parchevsky \& Kosovichev(2007a)]{Parchevsky2007a}

741: Parchevsky, K.~V. and Kosovichev, A.~G.\ 2007a, \apj, 666, 547

742: 

743: \bibitem[Parchevsky \& Kosovichev(2007b)]{Parchevsky2007b} Parchevsky, K.~V., \&

744: Kosovichev, A.~G.\ 2007b, \apjl, 666, L53

745: 

746: \bibitem[Parchevsky et al.(2008)]{Parchevsky2008} Parchevsky, K.~V., Zhao, J., \&

747: Kosovichev, A.~G.\ 2008, ArXiv e-prints, 802, arXiv:0802.3866

748: 

749: \bibitem[Rajaguru et al.(2006)]{Rajaguru2006} Rajaguru, S.~P.,

750: Birch, A.~C., Duvall, T.~L., Jr., Thompson, M.~J., \& Zhao, J.\

751: 2006, \apj, 646, 543

752: 

753: \bibitem[Scherrer et al.(1995)]{Scherrer1995} Scherrer, P.~H., et

754: al.  \ 1995, \solphys, 162, 129

755: 

756: \bibitem[Schunker et al.(2005)]{Schunker2005} Schunker, H., Braun,

757: D.~C., Cally, P.~S., \& Lindsey, C.\ 2005, \apjl, 621, L149

758: 

759: \bibitem[Vernazza et al.(1976)]{Vernazza1976} Vernazza, J.~E.,

760: Avrett, E.~H., and Loeser, R.\ 1976, \apjs, 30, 1.

761: 

762: \bibitem[Unno et al.(1989)]{Unno1989} Unno, W., Osaki, Y., Ando,

763: H., Saio, H., \& Shibahashi, H.\ 1989, Nonradial oscillations of

764: stars, Tokyo: University of Tokyo Press, 1989, 2nd ed.,

765: 

766: \bibitem[Zhao et al.(2001)]{Zhao2001} Zhao, J., Kosovichev,

767: A.~G., \& Duvall, T.~L., Jr.\ 2001, \apj, 557, 384

768: 

769: \bibitem[Zhao \& Kosovichev(2006)]{Zhao2006} Zhao, J. and

770: Kosovichev, A.,G.\ 2006, \apj, 643, 1317.

771: 

772: \bibitem[Zhao et al.(2007)]{Zhao2007} Zhao, J., Georgobiani, D.,

773: Kosovichev, A.~G., Benson, D., Stein, R.~F., \& Nordlund, {\AA}.\

774: 2007, \apj, 659, 848

775: 

776: \end{thebibliography}

777: 

778: \begin{figure}

779: \epsscale{0.8}\plotone{f1a.eps}

780: \epsscale{0.8}\plotone{f1b.eps}

781: \epsscale{0.8}\plotone{f1c.eps}

782: \caption{\label{Fig:1Dtest_PlnWaves}

783: Comparison of analytic (cirves) and numerical (markers) solutions

784: for plain Alfven (a), slow MHD (b), and fast MHD (c) waves

785: respectively. For Alfven wave only $v$ and $B_y$ (marked by solid

786: and dashed curves respectively) are non-zero. Circles mark an

787: analytical solution for $v$. Solid, dash-dotted, dashed, and dotted

788: curves in panels (b) and (c) correspond to $\rho$, $u$, $w$, and

789: $B_z$ respectively. Analytical solution for $w$ is shown by

790: circles.}

791: \end{figure}

792: 

793: \begin{figure}

794: \epsscale{1.0}\plotone{f2.eps} \caption{\label{Fig:beta} The ratio

795: of the gas pressure to the magnetic pressure (plasma parameter

796: $\beta$) as a function of depth for the background model for three

797: values of the magnetic field strength.}

798: \end{figure}

799: 

800: 

801: \begin{figure}

802: \epsscale{0.8}\plotone{f3a.eps}

803: \epsscale{0.8}\plotone{f3b.eps}

804: \epsscale{0.8}\plotone{f3c.eps}

805: \caption{\label{Fig:SingleSrc_UniB}

806: Snapshots of horizontal (left column) and vertical (right column)

807: slices of the 3D domain for perturbations of density $\rho$ (a),

808: z-momentum $\rho_0 w$ (b), and z-component of the magnetic field

809: $B_z$ (c). Strong Alfven wave is generated due to the interaction of

810: the wave source and the background magnetic field at the source

811: location. In this example, $B_0=625$ G, and inclined by $45^\circ$

812: in the XY-plane.}

813: \end{figure}

814: 

815: \begin{figure}

816: \epsscale{0.8}\plotone{f4a.eps}

817: \epsscale{0.8}\plotone{f4b.eps}

818: \epsscale{0.8}\plotone{f4c.eps}

819: \caption{\label{Fig:SingleSrc_NonUnB} Snapshots of horizontal (left

820: column) and vertical (right column) slices of the 3D domain for

821: perturbations of density $\rho$ (a), z-momentum $\rho_0 w$ (b), and

822: z-component of the magnetic field $B_z$ (c). The strength of the

823: background magnetic field (inclined by $45^\circ$ in the YZ-plane)

824: is 2500 G in the region to the right from vertical dashed lines in

825: left column and 0 in the left region. The wave source is located in

826: the region free of magnetic field. Propagating into the region with

827: the background magnetic field acoustic and surface gravity waves are

828: transformed into the fast MHD wave and magneto-gravity wave. The

829: Alfven wave does not appear in this case.}

830: \end{figure}

831: 

832: \begin{figure}

833: \epsscale{1.0}\plotone{f5a.eps}

834: \epsscale{1.0}\plotone{f5b.eps}

835: \caption{\label{Fig:kw_diagr} Spectra ($k$--$\nu$ diagrams) and

836: corresponding maps of perturbation of the z-component of velocity

837: are shown on top and bottom rows respectively. Results for the

838: original MHD wave field and wave fields after filtering out $f$- and

839: $p$-modes are shown on panels a, b, and c respectively.}

840: \end{figure}

841: 

842: \begin{figure}

843: \epsscale{1.0}\plotone{f6.eps}

844: \caption{\label{Fig:rho_TD}

845: Time-distance diagram for density perturbations. Solid curve

846: represents a theoretical time-distance curve for $p$-modes and the

847: standard solar model in absence of the magnetic field. Fast MHD wave

848: and magnetic-gravity waves are separated due to the different

849: dispersion relations.}

850: \end{figure}

851: 

852: 

853: \begin{figure}

854: \epsscale{1.0}\plotone{f7.eps}

855: \caption{\label{Fig:LoS_geometry}

856: Choice of the coordinate system. The line-of-sight direction is

857: defined by angles $\psi$ and $\theta$. Position of observational

858: point $P_{\mathrm{obs}}$ on the photosphere is fixed by azimuthal

859: angle $\alpha$ and distance $\Delta$ from the projection of wave

860: source O.}

861: \end{figure}

862: 

863: \begin{figure}

864: \epsscale{0.8}\plotone{f8a.eps}

865: \epsscale{0.8}\plotone{f8b.eps}

866: \epsscale{0.8}\plotone{f8c.eps}

867: \caption{\label{Fig:TravelTimes_alph} Mean phase travel times vs.

868: azimuthal angle $\alpha$ calculated from Doppler LoS velocities at

869: the hight of 300 km above the photosphere for different strengths of

870: the uniform inclined background magnetic field (625 G for panel a,

871: 1400 G for panel b, and 1900 G for panel c), and different LoS

872: directions. Solid, dashed, and dash-dotted curves represent travel

873: time variations for $\psi$ = $0^{\circ}$, $90^{\circ}$, and

874: $180^{\circ}$ respectively.}

875: \end{figure}

876: 

877: \begin{figure}

878: \epsscale{1.0}\plotone{f9.eps}

879: \caption{\label{Fig:SunSpot_geometry}

880: To derivation of the relation between $\psi$ and $A$. Angle $A$ is

881: the azimuthal angle of the observation point in the local coordinate

882: system associated with the sunspot. Angle $\psi$ is one of two

883: angles determining the LoS direction.}

884: \end{figure}

885: 

886: \begin{figure}

887: \epsscale{0.8}\plotone{f10a.eps}

888: \epsscale{0.8}\plotone{f10b.eps}

889: \caption{\label{Fig:TravTimeAvr_A} Phase travel time obtained from

890: simulations averaged along with annulus with radius 8 Mm as a

891: function of azimuthal angle $A$.  The solid, dashed, and dot-dashed

892: curves corresponds to the following combinations of the magnetic

893: field strength and inclination angles: (i) $B_0=625$ G,

894: $\gamma=70^\circ$, (ii) $B_0=1400$ G, $\gamma=45^\circ$, (iii)

895: $B_0=1900$ G, $\gamma=30^\circ$. Data points with errorbars

896: represent observations.}

897: \end{figure}

898: 

899: \end{document}

900: