1: \documentclass[preprint2]{aastex}
2: \received{2003 June 27}
3: \begin{document}
4:
5: \title{The Rotation Of The Deep Solar Layers}
6:
7: \author{S. Couvidat$^{1,2}$, R. A. Garc\'\i a$^2$, S. Turck-Chi\`eze$^2$, T.
8: Corbard$^3$, C. J. Henney$^3$, \&
9: S. Jim\'enez-Reyes$^4$}
10:
11: \email{couvidat@stanford.edu, rgarcia@cea.fr, cturck@cea.fr}
12:
13: \affil{$^1$ W.W. Hansen Experimental Physics Laboratory, Stanford University,
14: Stanford, CA 94305, USA\\
15: $^2$ CEA/DSM/DAPNIA/SAp, CE Saclay, 91191 Gif sur Yvette, France\\
16: $^3$ National Solar Observatory, 950 North Cherry Avenue, Tucson, AZ
17: 85726, USA\\
18: $^4$ Themis, Instituto de Astrof\'\i sica de Canarias, c/V\'\i a L\'actea s/n,
19: La Laguna, Tenerife, Spain}
20: % \received{updated 06/26/2003}
21:
22: \shorttitle{Solar Rotation with the GOLF+MDI data}
23: \shortauthors{Couvidat, Garci\'\i a, \& Turck-Chi\`eze}
24:
25: \begin{abstract}
26: From the analysis of low-order GOLF+MDI sectoral modes ($\ell \le 3, 6 \le n \le 15, |m|=\ell $) and LOWL data ($\ell > 3$), we derive the radial rotation profile assuming no latitudinal dependance in the solar core.
27: These low-order
28: acoustic modes contain the most statistically significant information about rotation of the deepest solar layers and should be least influenced by internal variability associated with the solar dynamo. After correction of the sectoral splittings for their
29: contamination by the rotation of the higher latitudes,
30: we obtain a flat rotation profile down to 0.2 R$_{\odot}$.
31: \end{abstract}
32:
33: \keywords{Sun: helioseismology, rotation --- Instruments: GOLF, LOWL, MDI}
34:
35: \section{Introduction}
36:
37: Helioseismologists use the oscillations of acoustic waves that propagate inside the Sun to infer its
38: rotation profile.
39: Due to the solar rotation (and magnetic fields), the frequencies of two modes of
40: the same degree $\ell$ and radial
41: order $n$, but with different azimuthal orders $m$, are separated by a small
42: amount referred to as splitting
43: ($\Delta \omega_{\ell n m}$).
44:
45: The understanding of the angular momentum redistribution in the deep interior
46: requires a precise derivation
47: of the rotation profile below the convection zone, but this profile is still a
48: matter of debate.
49: In the past an increase of the rotation rate near the core has
50: long been favored (\mbox{e.g.} Lazrek et \mbox{al.} 1996), but recent results by
51: Chaplin et \mbox{al.} (1999) and
52: Eff-Darwich, Korzennik, \& Jim\'enez-Reyes (2002) favor a slight decrease.
53: Corbard et \mbox{al.} (1997), using LOWL data, and Ehgamberdiev et al. (2001),
54: using IRIS data,
55: also derived such a decrease.
56:
57: Here we focus on the rotation in the solar core derived from the GOLF
58: (Global Oscillations at Low
59: Frequency) and MDI (Michelson Doppler Imager) instruments, supplemented for modes
60: with $\ell >3$ by LOWL data.
61: These data sets and the interest of the low-order modes are presented in sections 2 and 3 respectively. In section 4, the method utilized to extract the rotational splittings is discussed. In section 5 we discuss the rotation profile. We conclude in section 6.
62:
63: \section{The Data}
64:
65: We use data from GOLF/SoHO and MDI/SoHO for the modes $\ell \le 3$.
66: GOLF detects the global solar oscillations by the Doppler shift they produce at
67: the surface on the sodium
68: lines. It is specifically designed to be more sensitive to the low-degree modes,
69: and only detects those with $\ell+m$ equal to an even number.
70: We utilize the 2034-day long GOLF series (starting in April 1996) for the GOLF-alone
71: splittings.
72: MDI detects solar surface velocity from a nickel spectral line and resolves the
73: solar surface. To produce
74: integrated-disk temporal series sensitive to the low-$\ell$ modes, we apply two masks
75: (gaussian, and
76: gaussian zero mean) to the 2243-day long MDI series (Henney et
77: \mbox{al.} 1999). These two
78: series have been properly combined with GOLF 2243-day long series to increase
79: the signal-to-noise ratio of the
80: low-$\ell$ modes, and to reduce the leakage of higher-$\ell$ modes present on
81: the MDI spectrum (Garc\'\i a et
82: \mbox{al.} 2003). The use of the MDI+GOLF time series allows for the precise measurement of lower-$n$
83: modes. This point of interest is discussed further
84: in section 3.
85: We supplemented these GOLF-alone and MDI+GOLF data with LOWL rotational splittings
86: for $\ell > 3$. We used six
87: LOWL series of 1 year, from February 1994 to February 2000. The splittings have
88: been
89: extracted for each year but the final result uses the 6 years statistics
90: (Jim\'enez-Reyes 2001).
91:
92: \section{Information Value of Low-Order Modes}
93:
94: The derivation of the rotation profile in the radiative zone requires
95: high-precision splittings. For
96: instance, only $\simeq 3.5 \%$ of the splitting ($406$ nHz) of the mode $\ell=1$
97: $n=9$ $m=1$ (at $1472.85 \,
98: \mu$Hz) is due to the rotation below $0.2 \, R_{\odot}$ (Couvidat 2002). Thus,
99: we need to derive $\Delta
100: \omega_{1,9,1}$ with a precision better than $14.2$ nHz to obtain information
101: on the core.
102: The error associated with this splitting is $\simeq \pm 5.3$ nHz (see Table
103: \ref{tab1}). It contains more information
104: than the mode $\ell=1$ $n=20$ (at $2963.43 \, \mu$Hz), for which $6.5 \%$ of its
105: splitting ($401$ nHz) is due to
106: the core, \mbox{i.e.} $26$ nHz, but the uncertainty is $34.6$ nHz. This point is
107: illustrated by the left panel of Fig \ref{fig1}.
108: Even though the inner turning point is closer to the core as $n$ increases, the
109: outer turning point
110: is also closer to the surface, where stochastic excitation and solar cycle effects
111: become more important.
112: Therefore, the most favorable trade-off between sensitivity to the core rotation and
113: uncertainty in the rotational splitting
114: determination occurs for low-$\ell$ low-$n$ p modes. Consequently we use
115: only these modes in the present
116: analysis.
117:
118: \begin{figure}[ht]
119: \plotone{f1.eps}
120: \epsscale{1.4}
121: \caption{\label{fig1} Left panel: Ratio of the splitting due to the layers below
122: $0.2 \, R_{\odot}$, to the uncertainty on the total splitting for $\ell=1$ (solid line), $\ell=2$ (dashed line), and $\ell=3$ (dash-dotted line).
123: Right panel: synodic splittings for $n \le 15$ used for the inversion of the solar rotation,
124: obtained with GOLF and MDI, as a function of the inner turning point for $\ell=1$ (crosses), $\ell=2$ (stars), and $\ell=3$ (diamonds).}
125: \end{figure}
126:
127: \section{The Rotational Splittings}
128:
129: The splittings are obtained by fitting the components of each mode
130: to asymmetric Lorentzian profiles (Nigam \& Kosovichev 1998) using
131: a maximum likelihood code based on Appourchaux, Gizon, \& Rabello-Soares (1998).
132: The modes $\ell=0$ \& $2$ are fitted together, likewise the modes $\ell=1$ \& $3$.
133: For the GOLF-alone series, we use the periodogram of the data as the power spectrum, except for the modes $\ell=1$ $n=7$ \& $8$, and
134: $\ell=2$ $n=8$ \& $9$,
135: for which we use a multitaper spectral estimator (\mbox{e.g.} Percival \& Walden 1993).
136: For the combined MDI+GOLF series
137: we again use the multitaper.
138: We restricted the search of the splittings to modes with a frequency less than
139: $\simeq 3000 \, \mu$Hz, and use for the inversion only the modes with $n \le 15$. This
140: way we limit the impact of the solar cycle and surface effects. This approach is consistent with the results of
141: Chaplin et \mbox{al.} (2001) who demonstrated
142: that the splittings are strongly biased at higher frequencies, and Chaplin et \mbox{al.} (2003), who showed that the solar cycle does not affect the splittings of the low-$n$ $\ell=2$ modes.
143:
144: The free parameters of the fitted profiles for each mode are: the full width at
145: half maximum (FWHM, the same
146: for all the components of a mode), the amplitude (the relative amplitudes of the
147: different components are fixed
148: empirically), the asymmetry (the same for each component), the central
149: frequency, the noise level (assumed to
150: be constant in the fit window of 40 $\mu$Hz), and the splittings. The latter parameters are sectoral values ($|m|=\ell$).
151: We restrict the fits to frequencies $\le 2000-2200 \, \mu$Hz on the MDI+GOLF
152: data, because of the growing influence of
153: higher-$\ell$ modes in the fit window (for instance, $\ell=5$ modes perturb the
154: $\ell=2$ profiles).
155:
156: In Table \ref{tab1}, we list the synodic sectoral splittings and their
157: uncertainties. The rotational splittings
158: common to the MDI+GOLF and GOLF-alone datasets generally agree within the
159: $1\sigma$ uncertainty.
160: Right panel of Fig \ref{fig1} presents these splittings as a function of the
161: inner turning point. This plot emphasizes the
162: high quality of the SoHO data after 5 years of observation in comparison with
163: previous data sets.
164:
165: \begin{deluxetable}{llcclllccl}
166: \tabletypesize{\scriptsize}
167: \tablewidth{0pt}
168: \tablecaption{\label{tab1} Synodic sectoral splittings ($\Delta \omega_{nlm}$)
169: from GOLF-alone (left column)
170: and MDI+GOLF (right column).
171: The errors (err.) on the splittings are the $1\sigma$ uncertainties provided by
172: the Hessian matrix of the fit.
173: $r_{t}$ is the inner turning point of the mode.}
174: \tablehead{\colhead{$\ell$} & \colhead{$n$} & \colhead{$\Delta \omega_{nlm}$
175: (nHz)} & \colhead{err. (nHz)} &
176: \colhead{$r_{t}$ ($R_{\odot}$)}& \colhead{$\ell$} & \colhead{$n$} &
177: \colhead{$\Delta \omega_{nlm}$ (nHz)} &
178: \colhead{err.(nHz)} & \colhead{$r_{t}$ ($R_{\odot}$)}}
179: \startdata
180: 1 & 6 & ------/398.49 & ----/1.53 & 0.102 & 2 & 11& 401.07/401.33 & 8.20/6.45 & 0.132 \\
181: 1 & 7 & 393.39/401.05 & 6.46/3.57 & 0.096 & 2 & 12& 379.60/380.34 & 10.72/4.42& 0.124 \\
182: 1 & 8 & 400.67/398.70 & 6.06/1.49 & 0.090 & 2 & 13& 402.37/396.45 & 10.64/11.75& 0.118\\
183: 1 & 9 & 406.40/403.96 & 5.28/3.43 & 0.085 & 2 & 14& 395.17/------ & 15.27/-----& 0.112\\
184: 1 & 10& 414.63/412.37 & 9.50/4.22 & 0.080 & 2 & 15& 402.65/------ & 18.94/-----& 0.107\\
185: 1 & 11& 400.16/401.96 & 10.47/4.60& 0.076 & 3 & 6 & ------/403.69 & -----/3.64 & 0.252\\
186: 1 & 12& 406.56/397.39 & 14.19/6.22& 0.073 & 3 & 7 & ------/403.84 & -----/4.36 & 0.232\\
187: 1 & 13& 436.49/426.98 & 15.63/15.29&0.069 & 3 & 8 & ------/405.23 & -----/6.39 & 0.215\\
188: 1 & 14& 411.15/417.97 & 23.50/23.23&0.066 & 3 & 9 & ------/403.92 & -----/9.23 & 0.201\\
189: 1 & 15& 408.08/395.40 & 28.94/27.53&0.063 & 3 &10 & ------/404.79 & -----/14.28& 0.190\\
190: 2 & 6 & ------/399.84 & -----/2.23 & 0.184& 3 &11 & ------/400.58 & -----/15.30& 0.179\\
191: 2 & 7 & ------/401.37 & -----/4.07 & 0.170& 3 &12 & 401.52/402.58 & 10.10/18.34& 0.172\\
192: 2 & 8 & 397.00/399.10 & 5.70/2.83 & 0.158 & 3 &13 & 388.53/416.07 & 11.47/9.09 & 0.162\\
193: 2 & 9 & 412.01/406.42 & 6.49/2.43 & 0.147 & 3 &14 & 413.62/384.35 & 11.65/10.90& 0.155\\
194: 2 & 10& 406.55/400.97 & 6.95/5.06 & 0.139 & 3 &15 & 409.72/410.05 & 16.37/20.18& 0.148\\
195:
196:
197: \enddata
198: \end{deluxetable}
199:
200: \section{The Solar Rotation Profile}
201:
202: The 1D inversion of sectoral splittings provides information about the rotation mainly
203: along the solar equator, but with contamination by the higher latitudes. We applied a 1D MOLA inversion
204: technique (Multiplicative Optimally
205: Localized Averages, see \mbox{e.g.} Corbard et \mbox{al.} 1998).
206: The MDI+GOLF and GOLF-alone splittings (when the former are unavailable) were
207: completed for $\ell > 3$ with splittings from LOWL.
208: The direct inversion of the data leads to a rigid rotation from the base of the
209: convective zone at $0.713 \,
210: R_{\odot}$ down to $\simeq 0.35 \, R_{\odot}$ (see upper panel of Fig
211: \ref{fig2}). Below there is a decrease
212: in the rotation rate of the solar equator, as mentioned previously by several authors.
213: Depending on the regularization
214: parameter, this decrease is more or less pronounced (here we show the profile with the most pronounced decrease).
215: However, the 1D inversion with the sectoral splittings relies on the assumption
216: that the rotation rate at any latitude is equal to the rate at the equator.
217:
218: As the differential rotation of the convective zone implies the higher latitudes
219: rotate more slowly than the equator, we
220: need to correct the splittings for this contribution.
221: Therefore, we add $12$ nHz to the $\ell=1$ splittings, $8$ nHz to the $\ell=2$, and
222: $6$ nHz to the
223: $\ell=3$, following Corbard et \mbox{al.} (1998). The values of these corrections were derived from a 2D Regularized Least-Squares inversion of MDI data in the convective zone, by comparing the splittings computed with the exact linear relation (that involves the 2D rotational kernels and rotation profile) with the splittings computed with the 1D sectoral approximation.
224: We present on the lower panel of Fig \ref{fig2} the profile derived with the
225: corrected splittings. We obtain a flat
226: profile in the radiative interior. The dip around $0.3 \, R_{\odot}$ might be
227: due to the absence of correction
228: of the LOWL splittings but we cannot rule out a physical phenomenon at the limit of the nuclear core.
229:
230: This careful inversion of our most precise seismic data therefore favors a flat rotation
231: curve at $430$ nHz in the solar
232: radiative interior.
233: This lends support for a magnetic field strong enough to suppress any differential rotation
234: that might arise from an angular momentum redistribution through the gravity waves (Talon, Kumar, \& Zahn 2002).
235: We think that the absence of a decrease or increase in the rotation rate, compared with previous works, is mainly
236: due to the quality of the present data and the limitation in the degradation of
237: these data by the external layers.
238: We note that an increase of the rotation rate seems now very unlikely down to
239: $0.2 \, R_{\odot}$,
240: but cannot be ruled out below $0.2$ by the present gravity-mode analysis (Turck-Chi\`eze et
241: \mbox{al.} 2002).
242:
243: We have also reduced the MDI+GOLF splittings to $a_1$ coefficients (Schou
244: et \mbox{al.} 1994)
245: by removing $a_3$ (for $\ell=2,3$) and $a_5$ (for $\ell=3$) estimated from our knowledge
246: of the rotation in the convection zone.
247: This allows us to combine them directly with LOWL $a_1$ coefficients and
248: carry a 1D inversion
249: for a latitudinally averaged rotation rate. This gives results nearly identical to the one shown on the bottom panel of Fig \ref{fig2}.
250: Finally, we also suppressed the solar cycle effects for intermediate- and high-$\ell$ by using only the first year of LOWL data. Of course this increases the vertical error bars but does not change the rotation profile.
251:
252: \begin{figure}[!ht]
253: \centering
254: \epsscale{0.95}
255: \plotone{f2.eps}
256: \caption{\label{fig2} Upper panel: rotation profile directly derived with GOLF \&
257: MDI, combined with LOWL
258: splittings for $\ell > 3$.
259: Lower panel: rotation profile obtained after correction of the splittings to
260: account for the differential
261: rotation in the convective zone.
262: The vertical error bars result from propagating splitting measurement errors
263: through the inversion process, while
264: the horizontal error bars give the FWHMs of the corresponding averaging kernels,
265: and is an estimate of the
266: resolution achieved at each depth.}
267: \end{figure}
268:
269:
270: \section{Summary}
271:
272: By using the high-quality seismic data from the spatial instruments GOLF \&
273: MDI, and by using only the low-$n$ modes,
274: we limit the effects of the variable magnetic field that takes place in the
275: outer layers and obtain a very coherent dataset for rotation below the solar convection zone. While a proper treatment
276: of the latitudinal dependence of the rotation speed is still needed to extract the
277: rotation in the core from 1D inversions, the method we apply here removes the effect of latitudinal variation of the rotation in the convection zone, and we obtain a flat rotation profile down to 0.2
278: $R_{\odot}$. This puts a strong
279: constraint on the redistribution of the angular momentum. The
280: uncertainties in the rotation rate are still quite large
281: below $0.3 \, R_{\odot}$: progress on this point can be achieved by the
282: detection of mixed low-$\ell$ pressure modes, and gravity modes.
283:
284:
285: \section*{Acknowledgments}
286: GOLF \& MDI instruments are cooperative efforts (of French and Spanish
287: institutes for GOLF) to whom we are deeply
288: indebted. SoHO is a project of international cooperation between ESA and NASA.
289: S.C. acknowledges the support of
290: a CEA/Saclay doctoral Grant and the NASA Grant NAG5-10483. We wish to thank ECHO
291: team for providing us the LOWL
292: mode parameters.
293:
294: % \footnotesize
295:
296: \begin{thebibliography}{}
297:
298: \bibitem[]{311} Appourchaux, T., Gizon, L., \& Rabello-Soares, M.-C. 1998, A\&AS, 132,
299: 107
300: \bibitem[]{313} Chaplin, W. J., et al. 1999, MNRAS, 308, 405
301: \bibitem[]{314} Chaplin, W. J., et al. 2001, MNRAS, 327, 4, 1127
302: \bibitem[]{315} Chaplin, W. J., et al. 2003, MNRAS, in press
303: \bibitem[]{316} Corbard, T., et al. 1997, \aap, 324, 298
304: \bibitem[]{317} Corbard, T., et al. 1998, in SoHO 6/GONG98 workshop proceedings, ESA
305: SP-418, 741
306: \bibitem[]{319} Couvidat, S. 2002, PhD dissertation, Univ. Paris VII, Paris
307: \bibitem[]{320} Eff-Darwich, A., Korzennik, S. G., \& Jim\'enez-Reyes, S. J. 2002,
308: \apj, 573, 857
309: \bibitem[]{322} Ehgamberdiev, Sh. A., et al. 2001, in Proceedings of the IAU symposium, 203,
310: 97
311: \bibitem[]{324} Garc\'\i a, R. A., et al. 2003, in SoHO 12/ GONG+ 2002 meeting
312: proceedings, ESA SP-517
313: \bibitem[]{326} Henney, C. J., et al. 1999, A\&A, 348, 627
314: \bibitem[]{327} Jim\'enez-Reyes, S. J. 2001, PhD dissertation, Univ. La Laguna,
315: Tenerife
316: \bibitem[]{329} Lazrek, M., et al., 1996, Sol. Phys., 166, 1
317: \bibitem[]{330} Nigam, R., \& Kosovichev, A. G. 1998, ApJ, 505, L51
318: \bibitem[]{331} Percival, D. B., \& Walden, A. T. 1993, Spectral Analysis for
319: Physical Applications. Multitaper and
320: Conventional Univariate Techniques, Cambridge University Press
321: \bibitem[]{334} Schou, J., Christensen-Dalsgaard, J., \& Thompson, M. J. 1994, ApJ, 433, 389
322: \bibitem[]{335} Talon, S., Kumar, P., \& Zahn, J.-P. 2002, \apj , 574, L175
323: \bibitem[]{336} Turck-Chi\`eze, S., et al. 2002, ApJ, submitted
324: \end{thebibliography}
325:
326: \end{document}