1: \documentclass[12pt,preprint]{aastex}
2: %\usepackage{epsf}
3:
4: %======================================================================
5: % author's own macros
6: %======================================================================
7:
8: \newcommand{\sect}[1]{\S\,\ref{#1}}
9: \newcommand{\mem}[1]{\ensuremath{\mathrm{ #1}}}
10: \newcommand{\msun}{\ensuremath{\, {M}_\odot}}
11: \newcommand{\be}{\begin{displaymath}}
12: \newcommand{\ee}{\end{displaymath}}
13: \newcommand{\bea}{\begin{eqnarray}}
14: \newcommand{\eea}{\end{eqnarray}}
15: \newcommand\msol{{M_{\odot}}}
16: \newcommand\mstar{{M}}
17: \newcommand\mr{{M}_r}
18: \newcommand\teff{{T_{\rm eff}}}
19: \newcommand\amlt{{\alpha_{\rm MLT}}}
20: \newcommand\logl{\log\,L/L_\odot}
21: \newcommand\cs{\,cm$^2$\,s$^{-1}$}
22: \newcommand\kms{\,km\,s$^{-1}$}
23: %======================================================================
24:
25: %\received{}
26: %\accepted{}
27: %\journalid{}{}
28: %\articleid{}{}
29:
30: %\slugcomment{Astrophysical Journal, submitted}
31:
32: \shortauthors{Denissenkov, Pinsonneault, and MacGregor}
33: \shorttitle{Gravity Waves and Solar Uniform Rotation}
34:
35: \begin{document}
36:
37: \title{WHAT PREVENTS INTERNAL GRAVITY WAVES FROM DISTURBING THE SOLAR UNIFORM ROTATION?}
38:
39: \author{Pavel A. Denissenkov\altaffilmark{1,2}, Marc Pinsonneault\altaffilmark{1}, and Keith B. MacGregor\altaffilmark{3}}
40: \altaffiltext{1}{Department of Astronomy, The Ohio State University, 4055 McPherson Laboratory,
41: 140 West 18th Avenue, Columbus, OH 43210; dpa@astronomy.ohio-state.edu, pinsono@astronomy.ohio-state.edu.}
42: \altaffiltext{2}{On leave from Sobolev Astronomical Institute of St. Petersburg State University,
43: Universitetsky Pr. 28, Petrodvorets, 198504 St. Petersburg, Russia.}
44: \altaffiltext{3}{High Altitude Observatory, National Center for Atmospheric Research, P.O. Box 3000,
45: Boulder, CO 80307-3000; kmac@hao.ucar.edu.}
46:
47: \begin{abstract}
48: Internal gravity waves (IGWs) are naturally produced by convection in stellar envelopes, and they
49: could be an important mechanism for transporting angular momentum in the radiative interiors of stars.
50: Prior work has established that they could operate over a short enough time scale to explain the internal
51: solar rotation as a function of depth. We demonstrate that the natural action of IGWs is to produce large scale
52: oscillations in the solar rotation as a function of depth, which is in marked contrast to the nearly uniform rotation
53: in the outer radiative envelope of the Sun. An additional angular momentum transport mechanism is therefore
54: required, and neither molecular nor shear-induced turbulent viscosity is sufficient to smooth out the profile.
55: Magnetic processes, such as the Tayler-Spruit dynamo, could flatten the rotation profile. We therefore conclude that
56: IGWs must operate in conjunction with magnetic angular momentum transport processes if they operate at all.
57: Furthermore, both classes of mechanisms must be inhibited to some degree by mean molecular weight gradients in
58: order to explain the recent evidence for a rapidly rotating embedded core in the Sun.
59: \end{abstract}
60:
61: \keywords{stars: interiors --- Sun: rotation --- waves}
62:
63: \section{Introduction}
64: \label{sec:intro}
65:
66: During their pre-main sequence contraction, young solar-type stars are spun up
67: to rotational velocities of the order of 100\,\kms. However, during their subsequent
68: main sequence (MS) evolution the surface rotation slows down as a result of angular momentum loss
69: through magnetized stellar winds (\citealt{k88,mp08}). If a convective envelope and
70: a radiative core of a solar-type MS star rotated independently of one another then the surface spindown
71: would lead to a strong differential rotation beneath the envelope. In contradiction with
72: this, helioseismic data (e.g., \citealt{cea03}) reveal that the solar radiative core
73: rotates as a solid body and almost synchronously with the convective envelope
74: at least down to the radius $r\approx 0.2\,R_\odot$.
75: Besides, the spindown of young cluster stars (\citealt{sh87,kea95,bea97,kea97,b03})
76: requires that the internal differential rotation only persists for
77: timescales of the order of 20 Myr to 100 Myr. These data indicate that, in radiative interiors of solar-type MS stars,
78: there is an efficient mechanism of angular momentum redistribution that couples their core and
79: envelope rotation. Unfortunately, its physical nature remains elusive in spite of many
80: attempts to understand it made in the last years. Matters are further complicated by the requirement that
81: an appropriate model of angular momentum transport in the solar-type MS stars should also reproduce
82: the intricate variations (depletion) of the surface Li abundance as functions of age and effective temperature
83: observed in the same stars (\citealt{sr05}).
84:
85: A breakthrough in solving this complex problem has recently been announced by \cite{tch05} and
86: \cite{cht05}. They have shown how internal gravity waves (hereafter, IGWs) generated by
87: the envelope convection can extract angular momentum from rapidly rotating radiative cores of
88: the solar-type MS stars on short enough timescales to explain both the spindown of young cluster stars
89: and the quasi-solid-body rotation of the Sun. Furthermore, they have found that the quick
90: flattening of the internal rotation profile
91: by IGWs reduces element diffusion coefficients associated with hydrodynamic instabilities induced by
92: differential rotation to values consistent with those constrained by the Li data.
93:
94: Pursuing the goal of uncovering intrinsic causes of canonical extra mixing in low-mass red giant branch stars
95: (\citealt{dvb03}) and trying to understand the origin of fast rotation of red horizontal branch stars (\citealt{sp00}),
96: we have been undertaking a critical review of different transport mechanisms in
97: stellar radiative zones. When applying our test model of angular momentum redistribution by IGWs to a model of the present-day Sun,
98: we have found that IGWs strongly disturb the solar internal rotation making it disagree with the helioseismic data.
99: The disturbance can be eliminated only if there is another transport mechanism competing with IGWs. Its efficiency should
100: exceed that of the rotational shear mixing used by \cite{tch05} by more than three orders of magnitude. Alternatively,
101: a magnetic transport mechanism could be a competitor for IGWs. But in that case the mechanism preventing
102: IGWs from disturbing the solid-body rotation of the Sun could itself be responsible for both shaping the
103: internal rotation of the solar-type MS stars and assisting in the depletion of their surface Li abundance. This paper presents
104: a discussion of computational results that support our conclusions.
105:
106: \section{Angular Momentum Transport by IGWs}
107: \label{sec:transport}
108:
109: In a gravitational field, any perturbation exerted on a fluid excites in it both acoustic (p-modes) and
110: internal gravity waves (g-modes). The driving force for the latter is the buoyancy, as opposed to the pressure for the p-modes.
111: A general discussion of their combined linear theory can be found in the book by \cite{l78}.
112: Our application of the IGW theory first addresses the following three questions: How do IGWs propagate through radiative layers?
113: How do they redistribute angular momentum in a rotating star? How are IGWs generated? Answering the last question necessarily includes
114: a discussion of the IGW spectrum. We start general comments on IGWs, then attempt to examine the underlying issues.
115:
116: The physics of angular momentum transport and chemical mixing by IGWs in stellar radiative zones
117: has been comprehensively described by \cite{p81}, \cite{gls91}, \cite{zea97}, \cite{r98}, \cite{kea99}, \cite{ms00}, \cite{kmg01},
118: \cite{tea02}, and \cite{tch05}. In all of these papers, IGWs are considered to be generated by large-scale
119: turbulent fluid motions either in the convective envelope or at the interface between
120: the radiative core and convective envelope. The net energy flux of IGWs at the core/envelope interface $r_{\rm c}$ is usually estimated
121: as $F_E(r_{\rm c})\approx {\cal M}_{\rm t}F_{\rm c}$, where $F_{\rm c}=\rho_{\rm c}v_{\rm c}^3\approx 0.1\,(L/4\pi r_{\rm c}^2)$ is
122: the convective flux, $L$ is the star's luminosity, and ${\cal M}_{\rm t}=\omega_{\rm c}/N_{\rm c}$ is the turbulent Mach number.
123: In the last ratio, $N_{\rm c}$ is the buoyancy frequency immediately beneath
124: the interface, while $\omega_{\rm c}=2\pi v_{\rm c}/\lambda_{\rm c}$ is the turnover frequency
125: of the largest convective eddy approaching the interface with the velocity $v_{\rm c}$.
126: Everywhere in this paper $\omega$ denotes the circular frequency. Although it should be measured in rad\,s$^{-1}$, we will
127: always express its values in units of $\mu$Hz, assuming that $1\,\mu$Hz\,$\equiv 10^{-6}$\,rad\,s$^{-1}$.
128: In the mixing-length theory (MLT) of convection, that we apply,
129: the diameter of the largest turbulent eddy, which also measures its mean free path $\lambda_{\rm c}$, is
130: an $\alpha_{\rm MLT}$ fraction of the local pressure scale height $H_P$.
131: We employ the stellar evolution code described by \cite{dea06}. Gravitational settling is not included because it does not affect
132: the IGW propagation. Indeed, IGWs can easily penetrate inner radiative cores of low-mass MS stars (\citealt{tch05}) where radial variations of the mean molecular weight $\mu$
133: are much stronger than those incurred from the operation of gravitational settling. We use the \cite{gn93} mixture of
134: heavy elements. Our calibrated solar model reproduces the solar luminosity
135: ($L_\odot = 3.85\times 10^{33}$\,erg\,s$^{-1}$) and radius ($R_\odot = 6.96\times 10^{10}$\,cm) at
136: the solar age of 4.57 Gyr; this procedure yields the helium and heavy-element mass fractions
137: $Y = 0.273$, $Z = 0.018$, and the mixing length $\alpha_{\rm MLT}$ of 1.75.
138: Our solar model has $r_{\rm c}\approx 0.713\,R_\odot$ and ${\cal M}_{\rm t,\odot}\approx 8.3\times 10^{-4}$, so that the energy luminosity of
139: IGWs at the bottom of its convective envelope $L_E(r_{\rm c})=4\pi r_{\rm c}^2F_E(r_{\rm c})$ comprises only
140: $\sim$\,0.0083\,\% of $L_\odot$.
141:
142: On their way from the bottom of the convective envelope toward the center of a solar-type MS star,
143: IGWs experience radiative damping. This can be taken into account by applying a wave attenuation factor
144: $\exp(-\tau)$ to $L_E(r_{\rm c})$. To calculate the effective optical depth, we use the relation
145: \bea
146: \tau = [l(l+1)]^{3/2}\int_r^{r_{\rm c}}K\frac{NN_T^2}{\sigma^4\sqrt{1-(\sigma/N)^2}}\frac{dr'}{r'^3}
147: \label{eq:tau}
148: \eea
149: derived by \cite{zea97}. Here,
150: $N^2 = N_T^2 + N_\mu^2$,
151: where
152: \be
153: N_T^2 = \frac{g\delta}{H_P}(\nabla_{{\rm ad}}-\nabla_{{\rm rad}}),
154: \ \ \mbox{and}\ \
155: N_\mu^2 = g\varphi\,\left|\frac{\partial\ln\mu}{\partial r}\right|
156: \ee
157: are the $T$- and $\mu$-component of the square of the
158: buoyancy frequency. In the last two expressions,
159: $\nabla_{\rm rad}$ and $\nabla_{\rm ad}$ are the radiative
160: and adiabatic temperature gradients (logarithmic and with respect to
161: pressure), and $g$ is the local gravity.
162: The quantities
163: $\delta =
164: -\left(\partial\ln\rho/\partial\ln T\right)_{P,\,\mu}$ and
165: $\varphi = \left(\partial\ln\rho/\partial\ln\mu\right)_{P,T}$
166: are determined by the equation of state. In our IGW computations,
167: the equation of state for the ideal gas is used. In this particular case, $\delta = \varphi = 1$.
168: Also in equation (\ref{eq:tau}), $K = 4acT^3/3\kappa\rho^2C_P$
169: is the radiative diffusivity with $\kappa$ and $C_P$ representing the Rosseland mean opacity and
170: the specific heat at constant pressure, respectively. The quantities $l$ and $\sigma$ are introduced below.
171:
172: Turbulent eddies with different length and overturn time scales present in the convective envelope
173: can generate a whole spectrum $S_E(r_{\rm c},l,m,\omega)$ of IGWs
174: with different spherical degrees $l\geq 1$, azimuthal numbers $m$ ($|m|\leq l$), and frequencies $\omega$.
175: If the star rotates
176: and its angular velocity $\Omega$
177: varies with $r$ then the optical depth (\ref{eq:tau}) depends on all three of the wave's
178: spectral characteristics,
179: the latter two entering it through the doppler-shifted frequency
180: $\sigma = \omega - m[\Omega(r)-\Omega(r_{\rm c})]$
181: (we use a cartesian coordinate system in which
182: the z-axis is colinear with the vector $\mathbf{\Omega}$).
183: Values of this frequency should
184: be watched to remain between 0 and $N$. When $\sigma\rightarrow 0$, the optical depth $\tau$
185: approaches the infinity, and the wave is completely absorbed by the surrounding medium.
186: On the other hand, when $\sigma\rightarrow N$, the wave is totally reflected back (\citealt{r98}).
187: If only the IGWs with $1\leq l\leq l_{\rm max}$ and $\omega_{\rm min}\leq\omega\leq\omega_{\rm max}$
188: are excited and propagate into the radiative core then their net energy flux at the core/envelope interface is
189: \bea
190: F_E(r_{\rm c}) = \sum_{l=1}^{l_{\rm max}}\sum_{m=-l}^{l}\int_{\omega_{\rm min}}^{\omega_{\rm max}}S_E(r_{\rm c},l,m,\omega)d\omega = {\cal M}_{\rm t}F_{\rm c}.
191: \label{eq:neteflux}
192: \eea
193:
194: Besides energy, IGWs can also carry angular momentum the flux of which is
195: \bea
196: F_J = \frac{m}{\sigma}F_E
197: \label{eq:fjfe}
198: \eea
199: (\citealt{r98,kea99}).
200: Since both $F_E$ and $\sigma$ are positive, this relation means that
201: prograde waves (those with $m > 0$) carry positive angular momentum while
202: retrograde waves ($m < 0$) transport negative momentum.
203: Note that before Ringot's elucidating paper the improper (negative) sign was used in equation (\ref{eq:fjfe})
204: by many researchers creating some confusion in the field.
205:
206: Let us choose a frame of reference co-rotating with
207: the convective envelope and denote $\Delta\Omega(r)\equiv\Omega(r)-\Omega(r_{\rm c})$.
208: We will assume that, in spite of the action of the
209: Coriolis force in this noninertial frame of reference, the spectrum of IGWs generated by
210: the envelope convection is still axisymmetric, i.e. $S_E(r_{\rm c},l,m,\omega) = S_E(r_{\rm c},l,-m,\omega)$;
211: if necessary, corrections due to the Coriolis force can be taken into account later on (e.g., \citealt{tch03}).
212: If $\Delta\Omega(r)=0$, i.e. if the whole star rotates uniformly, then $\sigma = \omega$, and $\tau$ does not depend on $m$.
213: In this case, the total angular momentum luminosity associated with IGWs
214: \bea
215: L_J(r) = 4\pi r_{\rm c}^2\,\sum_{l=1}^{l_{\rm max}}\sum_{m=-l}^{l}m\int_{\omega_{\rm min}}^{\omega_{\rm max}}
216: S_E(r_{\rm c},l,m,\omega)\exp\{-\tau(r,l,m,\omega)\}\,\frac{d\omega}{\omega}
217: \label{eq:netjflux}
218: \eea
219: is equal to zero at any radius $r$ below the convective envelope.
220:
221: On the other hand, if $\Delta\Omega(r)\not= 0$, i.e. if the star rotates differentially,
222: then IGWs will experience selective damping in the radiative core. Indeed, let us assume
223: for example that $\Delta\Omega(r)$ increases with depth (when $r$ decreases).
224: In this case, the frequency $\sigma$ of a prograde wave with spectral characteristics $l$, $m$, and $\omega$ will be doppler shifted to values
225: smaller than the intrinsic frequency $\omega$, while that of a retrograde wave with the characteristics $l$, $-m$ and $\omega$ will
226: become larger than $\omega$ as the waves propagate inward. Hence, the difference
227: between their attenuation factors $\exp\{-\tau(r,l,m,\omega)\} - \exp\{-\tau(r,l,-m,\omega)\}$
228: will diminish with the depth. This happens because the prograde waves carrying the positive angular momentum
229: experience stronger damping (if $\Delta\Omega(r) > 0$) on their way into the radiative core
230: than the retrograde waves carrying the negative angular momentum.
231: The positive angular momentum absorbed by the surroundings at the beginning of the waves' path will spin up
232: rotation locally compared to $\Omega(r_{\rm c})+\Delta\Omega$ while the negative momentum accumulating in the waves
233: as they advance inward will be deposited deeper where the retrograde waves get eventually absorbed.
234: This briefly sketched physics of angular momentum redistribution by IGWs is incorporated into
235: the following PDE:
236: \bea
237: \rho r^2\,\frac{\partial\,\Delta\Omega}{\partial t} =
238: \frac{1}{r^2}\frac{\partial}{\partial r}\left(\rho r^4\nu\,\frac{\partial\,\Delta\Omega}{\partial r}\right) +
239: \frac{3}{8\pi}\,\frac{1}{r^2}\frac{\partial L_J}{\partial r},
240: \label{eq:ampdf}
241: \eea
242: where $\nu$ is a viscosity.
243: Supplemented with eqs. (\ref{eq:tau}), (\ref{eq:netjflux}), an expression for
244: the IGW spectrum at $r=r_{\rm c}$, and appropriate initial and boundary conditions,
245: this equation describes how the star's internal rotation profile evolves
246: in the presence of IGWs and a viscous force.
247:
248: \section{Spectra of IGWs}
249: \label{sec:spectrum}
250:
251: Given that the optical depth $\tau$, that determines the efficiency of damping of an IGW, strongly depends
252: on the wave's spectral characteristics $l$, $m$, and $\omega$, knowing
253: the spectral energy distribution for IGWs at the core/envelope interface
254: is as important as estimating their net energy. In this paper, we will use two analytical prescriptions
255: (eqs. \ref{eq:zea97} and \ref{eq:tea02} below) for
256: the spectra of IGWs. They correspond to two different physical mechanisms of the IGW excitation by turbulence
257: in the convective envelope. The formal lower and upper limits for the intrinsic frequency of these IGWs
258: are $\omega_{\rm min} = \omega_{\rm c}$ and $\omega_{\rm max} = N_{\rm c}$.
259:
260: Unfortunately, multidimensional hydrodynamic simulations of turbulent convection in
261: the solar envelope and its penetration into the radiative core give contradictory results
262: on the spectrum of IGWs generated in these numerical experiments.
263: For example, in their 2D simulations \cite{rg05} have found an IGW energy flux evenly distributed in
264: frequency at least for $l\leq 20$ with a peak energy that is three orders of magnitude smaller than
265: that predicted by equation (\ref{eq:tea02}). On the other hand, \cite{kea05} claim that
266: the broad frequency IGW spectrum reported in their earlier
267: publication was an artefact of the 2D approximation. They emphasize that the IGW flux obtained in
268: their new 3D simulations is of the same order as the one calculated from the simple parametric model of
269: \cite{gls91} based on the MLT. This finding encourages us to use spectrum
270: (\ref{eq:zea97}) as our primary IGW model.
271:
272: \subsection{The Garc\'{\i}a L\'{o}pez \& Spruit Spectrum}
273:
274: This prescription approximates the spectrum of IGWs generated at the core/envelope interface
275: as a result of dynamic hitting at the radiative side of the interface by breaking convective eddies.
276: This approximation was proposed and developed by \cite{p81}, \cite{gls91}, and \cite{zea97}. We use an expression derived in the last work
277: \bea
278: S_E(r_{\rm c},l,m,\omega) = \frac{1}{2}\,\rho_{\rm c}v_{\rm c}^3\,\frac{\omega_{\rm c}}{N_{\rm c}}\,\frac{1}{l_{\rm c}\,\omega_{\rm c}}
279: \left(\frac{\omega_{\rm c}}{\omega}\right)^3\frac{1}{2l}.
280: \label{eq:zea97}
281: \eea
282: Equation (\ref{eq:zea97}) is obtained under the following assumptions: {\it (i)} the dynamic pressure in the waves of a frequency $\omega$
283: matches that produced by the convective eddies with the overturn time $\sim$\,$\omega^{-1}$;
284: {\it (ii)} the kinetic energy spectrum of the convective motions in the envelope is represented
285: by the Kolmogorov law; {\it (iii)} besides waves at their own length scale $\lambda$, the convective eddies
286: also excite IGWs with horizontal wave lengths $\lambda_{\rm h} > \lambda$ by the superposition of their incoherent
287: action on the interface; these longer waves have a velocity amplitude reduced by the factor $\lambda/\lambda_{\rm h}$ (\citealt{gls91}).
288: In this prescription, $l_{\rm max} = l_{\rm c}(\omega/\omega_{\rm c})^{3/2}$, where $l_{\rm c} = 2\pi r_{\rm c}/\lambda_{\rm c}$.
289: Applying the integration and summation (\ref{eq:neteflux}) to the spectrum (\ref{eq:zea97}) and noticing
290: that $\omega_{\rm c}\ll N_{\rm c}$, we find that $F_{\rm E}(r_{\rm c}) = {\cal M}_{\rm t}F_{\rm c}$ as expected.
291:
292: \subsection{The Goldreich et al. Spectrum}
293:
294: The second prescription originated from the investigation of the stochastic excitation of p-modes
295: by turbulent convection in the Sun carried out by \cite{gea94}. In this model, IGWs are
296: generated by the fluctuating Reynolds stresses produced by turbulent fluid motions in the convective envelope.
297: Following \cite{tea02}, the emerging IGW spectrum at the core/envelope interface can be estimated as
298: \bea
299: S_E(r_{\rm c},l,m,\omega) & = & \frac{1}{2l}\,\frac{\omega^2}{4\pi}\int_{r_{\rm c}}^{R_{\rm c}}dr\,\frac{\rho^2}{r^2}
300: \left[\left(\frac{\partial\xi_r}{\partial r}\right)^2 + l(l+1)\left(\frac{\partial\xi_{\rm h}}{\partial r}\right)^2\right] \nonumber \\
301: & & \times\exp\left[-l(l+1)\frac{h_\omega^2}{2r^2}\right]
302: \frac{v^3\lambda^4}{1+(\omega\tau_\lambda)^{15/2}},
303: \label{eq:tea02}
304: \eea
305: where $R_{\rm c}$ is the outer radius of convective envelope, $\tau_\lambda = \lambda/v$ is the overturn time of
306: convective elements of the size $\lambda = \alpha_{\rm MLT}H_P$ moving with the velocity $v$ (at $r=r_{\rm c}$,
307: $\lambda$ and $v$ coincide with the parameters $\lambda_{\rm c}$ and $v_{\rm c}$ defined in \sect{sec:transport}),
308: and $h_\omega = \lambda\min\{1,\,(2\omega\tau_\lambda)^{-3/2}\}$. Like equation (\ref{eq:zea97}),
309: the latter equation has been derived under the assumption that the turbulent motions in the convective envelope
310: obey the Kolmogorov law.
311:
312: The radial displacement wave function $\xi_r$ and the horizontal one $l(l+1)\xi_{\rm h}$,
313: the latter being related to the former by the continuity equation (\citealt{zea97}), are normalized
314: to the unit IGW energy flux just below the convection zone.
315: For the radial function, we use the WKB solution (\citealt{p81,zea97,kea99})
316: \bea
317: \xi_r(r) = C[l(l+1)]^{1/4}\frac{\exp\left[-\int_{r_{\rm c}}^r\frac{|N|}{\omega}
318: \frac{\sqrt{l(l+1)}}{r'}dr'\right]}{\omega(\rho\, r\, |N|\,)^{1/2}}.
319: \label{eq:ksir}
320: \eea
321: This equation shows that in the convective envelope, where $N^2<0$, IGWs are evanescent. Therefore only those of them
322: that are excited close to the core/envelope interface will effectively contribute to the IGW flux at $r=r_{\rm c}$.
323: The constant $C$ is adjusted in our computations
324: for the spectrum (\ref{eq:tea02}) to yield $F_E(r_{\rm c}) = {\cal M}_{\rm t}F_{\rm c}$ with $F_{\rm c} = 0.1\,(L/4\pi r_{\rm c}^2)$.
325: The factor $(2l)^{-1}$ in eqs. (\ref{eq:zea97}) and (\ref{eq:tea02})
326: takes into account the assumed energy equipartition between the wave counterparts with opposite signs of the azimuthal number ($-l\leq m\leq l$).
327:
328: In Fig.~\ref{fig:f1}, we have plotted the logarithms of the IGW energy luminosity summed
329: over all available values of $m$ at the core/envelope interface
330: in our solar model. Those were calculated using the spectra (\ref{eq:zea97}) (solid curve) and (\ref{eq:tea02})
331: (dashed curves) with the same value of ${\cal M}_{\rm t}=8.3\times 10^{-4}$ that gives $L_E(r_{\rm c})\approx 3.2\times 10^{29}$\,erg\,s$^{-1}$.
332: Apparently, they have quite different
333: dependences on the spherical degree and frequency. Whereas the first spectrum
334: does not depend on $l$ at all\footnote{The factor $(2l)^{-1}$ disappears after
335: the summation over all $m=-l,\ldots,l$.}
336: (eq. \ref{eq:zea97}), the second is estimated to be proportional to $l^p$, where
337: $p\approx 1.6$ for the first 4 degrees.
338: Besides, the first spectrum declines with increasing $\omega$ much slower (with a power $-3$) than the second
339: spectrum (with a power $\sim$\,$-4.5$).
340: As a result, in the second case the dipole wave ($l=1$), that experiences the least damping (eq. \ref{eq:tau}),
341: carries much less energy toward the radiative core than it does in the first case,
342: especially at higher frequencies (compare the lower dashed and solid curve).
343:
344: \subsection{Uncertainties in the MLT}
345:
346: Fig.~\ref{fig:f2} shows how the convective and buoyancy frequency, $\omega_{\rm c}$ and $N$, vary
347: with the radius on the opposite sides of the core/envelope interface in our solar model.
348: From this figure, it is evident that neither the minimum frequency of IGWs $\omega_{\rm min}=\omega_{\rm c}$ nor
349: the ratio ${\cal M}_{\rm t}=\omega_{\rm c}/N_{\rm c}$ estimating the net energy flux
350: $F_E(r_{\rm c})\approx{\cal M}_{\rm t}F_\odot$
351: (dotted curve in the figure shows that the latter approximation becomes valid at $r\ga r_{\rm c}+0.5H_P$),
352: can be predicted with a confidence by the MLT of convection employed by us.
353: The uncertainties are caused by the rapid growth of $\omega_{\rm c}$
354: and $N$ with an increasing distance from the interface. Strictly speaking, both $\omega_{\rm c}\propto v_{\rm c}$
355: and $N\propto |\nabla_{\rm rad} -\nabla_{\rm ad}|^{1/2}$ should vanish at $r=r_{\rm c}$.
356: However, as soon as we step aside from the interface, both of them jump up to finite values.
357: In stellar model computations, their first nonzero values depend on spatial resolution:
358: the higher the resolution is, the smaller these values are. For example,
359: in our computations of the present-day Sun's model, the results of which are plotted in Fig.~\ref{fig:f2},
360: we find $\omega_{\rm c}\approx 0.30\,\mu$Hz and $N_{\rm c}\approx 360\,\mu$Hz, hence ${\cal M}_{\rm t}\approx 8.3\times 10^{-4}$.
361: If we had taken into account convective overshooting beyond the formal lower boundary of convective envelope
362: located at the radius $r=r_{\rm c}$, where $\nabla_{\rm rad}=\nabla_{\rm ad}$ (the Schwartzschild criterion),
363: then we would have obtained a larger value of $N_{\rm c}$. In a value of ${\cal M}_{\rm t}=\omega_{\rm c}/N_{\rm c}$,
364: the increase of $N_{\rm c}$ due to the overshooting can partly be compensated
365: by choosing a larger representative value of $\omega_{\rm c}$ that should account of its rapid growth with radius
366: on a length scale much less than $H_P$ immediately above the core/envelope interface (dashed curve).
367:
368: Besides the aforementioned uncertainties in the choice of representative values for $\omega_{\rm c}$ and $N_{\rm c}$,
369: the MLT does not account for the fact, established by laboratory experiments, observations in
370: the Earth's atmosphere and numerical simulations (e.g., see \citealt{sn89,cea91,rz95,ms00,rg06}), that
371: downward flows in a strongly stratified convection zone are much more energetic and confined in their
372: horizontal extent than upward flows. This may result in an underestimate of the IGW net energy flux, especially
373: in stars with deep convective envelope, like the Sun (\citealt{kea05}).
374:
375: Given the uncertainties in the MLT and the fact that we want to investigate the
376: stability of the solar uniform rotation against perturbations by the IGWs that were supposed to be powerful enough
377: to produce that uniform rotation in the past, we tentatively choose $\omega_{\rm min}=\omega_{\rm c}=0.30\,\mu$Hz
378: while considering the turbulent Mach number ${\cal M}_{\rm t}$
379: as a free parameter with its minimum value equal to our estimated ratio
380: $\omega_{\rm c}/N_{\rm c}\approx 8.3\times 10^{-4}$. This may actually put a conservative lower limit on the energetics of such IGWs
381: because other authors, including \cite{cht05} who demonstrated how the IGWs could shape the Sun's uniform rotation,
382: employed larger values of ${\cal M}_{\rm t}$ and sometimes also a higher $\omega_{\rm min}$.
383: Therefore, their low-degree high-frequency waves, that can penetrate deep into the radiative core,
384: carried more kinetic energy than they do in our basic case,
385: assuming, of course, that we use the same IGW spectrum. For the convective parameter
386: $l_{\rm c}$, we use an MLT value $l_{\rm c}=30$ from our calibrated solar model.
387:
388: \section{Qualitative Description of Expected Solutions}
389:
390: After the proper (positive instead of negative) sign in relation (\ref{eq:fjfe}) had been defined
391: by \cite{r98} it became clear that the addition of IGWs to other processes
392: responsible for the redistribution of angular momentum in stellar radiative zones would not be a trivial problem.
393: Indeed, instead of resulting in an exponential decay of perturbations of the internal rotation profile
394: toward a solid body rotation (e.g., \citealt{kq97}), the damping of IGWs leads to progressively growing deviations of
395: local rotation from its initial state, even if this state is close to the solid body rotation.
396: This is caused by the fact that in a region rotating faster
397: than the convective envelope the prograde waves experience stronger damping than the retrograde waves,
398: which results in a deposit of positive angular momentum there and hence in a further spin-up of this region.
399: The opposite is true for a region rotating slower than the convective envelope. Of course, the perturbations of
400: the rotation profile cannot grow infinitely large. First of all, an excess of (positive or negative) angular momentum in them
401: can slowly be dissipated through the molecular viscosity. Second and more important, when the rotational shear produced by the
402: perturbations becomes strong enough to trigger a shear instability
403: the turbulent viscous friction associated with the shear-induced turbulence will start to
404: contribute to smoothing the perturbations out.
405:
406: Apparently, the outcome of the competition between the disturbing action of IGWs and the smoothing effect of viscous friction
407: depends on their relative strength. Let us take a constant viscosity $\nu =\nu_0$. It is obvious that,
408: as long as $\nu_0$ is kept extremely large, the rotation profile will remain stationary because
409: any perturbation of $\Omega$ by IGWs will be quickly neutralized by the viscous dissipation.
410: Of course, this does not preclude gradual temporal changes of the rotation profile as a whole
411: due to the redistribution of angular momentum by IGWs as described by eq. (\ref{eq:ampdf}).
412:
413: If we begin to decrease $\nu_0$ then at some critical value of it, which is proportional to
414: the local IGW energy flux, regular oscillations of $\Omega$ will set in.
415: In the Earth's atmosphere, this bifurcation of a stationary solution toward an oscillatory solution
416: is believed to be observed experimentally as the quasi-biennial oscillations (QBO) of the mean zonal
417: wind in the equatorial stratosphere disturbed by IGWs coming down from the troposphere
418: (e.g., \citealt{lh68,p77,yh88}). In models of the solar-type MS stars, a behavior of the $\Omega$-profile resembling that of the QBO
419: has been found as a solution of equation (\ref{eq:ampdf}) near the top of radiative core by
420: \cite{kea99,kmg01,tea02,tch03}, and \cite{cht05}. \cite{tch05} have
421: called it the shear layer oscillations (SLOs). They have proposed that the SLOs
422: work as a filter for IGWs on their way toward the radiative core. If $\Omega$ increases with the depth in the core, as is expected
423: in a solar-type MS star losing its angular momentum from the surface through a magnetized stellar wind,
424: then the SLOs predominantly filter out the retrograde waves. These are absorbed closer to the center.
425: Possessing the minimum angular momentum, it is the very central part of the core that slows down the first.
426: Following it, increasingly more and more distant from the center layers get successively decelerated (\citealt{tch05}).
427: Note, however, that this theoretical prediction has recently been challenged by results of
428: a preliminary analysis of the GOLF data on solar g-mode oscillations reported by \cite{gea07}.
429: They suggest that the solar inner core at $r\la 0.15\,R_\odot$ may rotate three to five times as fast as
430: the rest of the radiative zone. If these results are confirmed by a further analysis they will likely rule out
431: the angular momentum redistribution in the Sun by IGWs.
432:
433: If we continue to reduce $\nu_0$ further on then the oscillations will grow up in
434: amplitude until finally they will turn into chaotic variations of $\Omega$ (e.g., \citealt{kmg01}).
435:
436: \section{Results: Two Spectra, Five Viscosities}
437: \label{sec:results}
438:
439: In this section, we present and discuss results of our numerical solutions of
440: the PDE (\ref{eq:ampdf}) that have been obtained with the IGW spectra
441: (\ref{eq:zea97}) and (\ref{eq:tea02}) using the viscosity prescriptions
442: summarized in Appendix~B. Besides restricting the azimuthal number
443: $m$ by the values of $\pm l$, we have also considered a limited set of
444: the spherical degree. In most of our computations, we have only used a set of numbers
445: $l=1,\,2,\,3$, and 4. We do not think that adding higher degrees
446: would qualitatively change our results and conclusions.
447: Indeed, the optical depth in the wave attenuation factor $\exp(-\tau)$
448: increases as $\tau\propto l^3$ for $l > 1$, therefore at a same radius in the Sun's radiative core
449: a contribution to the local energy flux of a wave with a higher $l$ is made from
450: a part of the spectrum at a higher frequency $\omega\sim l^{3/4}$ (solid curves in Fig.~\ref{fig:f3}).
451: As both of the IGW spectra quickly decline with a growth of $\omega$, our neglect of
452: waves with the higher spherical degrees is unlikely to lead to a serious mistake.
453: The same argument justifies our choice of the limited frequency interval
454: $0.3\,\mu\mbox{Hz}\leq\omega\leq 3\,\mu\mbox{Hz}$ because waves with higher frequencies
455: transport negligible amounts of kinetic energy and angular momentum.
456:
457: Unlike the other papers cited in the preceding section, in this work we have not tried to solve the full problem of
458: shaping the uniform rotation of the solar radiative core by IGWs. A tentative solution of it has been provided
459: by \cite{cht05}. We look at this problem from another perspective. Let us assume that the redistribution of
460: angular momentum in the Sun's core has already established its solid body outer envelope rotation as revealed by the helioseismic
461: data (e.g., \citealt{cea03}). Despite this, IGWs still continue
462: to be generated by the envelope convection. We cannot expect {\it a priori} that the stability of
463: the $\Omega$-profile against perturbations by IGWs strongly depends on its shape. Therefore, we consider it worth investigating
464: whether these perturbations are sufficiently weak in the present-day Sun for them not to disturb noticeably the Sun's uniform rotation,
465: provided that the IGWs producing these perturbations had enough energy in the past to couple the Sun's core and envelope rotation.
466: Given our reformulation of the problem, we have chosen the following initial and boundary conditions for eq. (\ref{eq:ampdf}):
467: \bea
468: \label{eq:init}
469: \Delta\Omega\,(0,r) = \frac{r_{\rm c}-r}{r_{\rm c}-r_{\rm b}}\Delta\Omega\,(0,r_{\rm b})\ \
470: \mbox{for}\ \ r_{\rm b}\leq r\leq r_{\rm c},\\
471: \mbox{and}\ \ \Delta\Omega\,(t,r_{\rm c}) = 0,\ \ \Delta\Omega\,(t,r_{\rm b})=\Delta\Omega\,(0,r_{\rm b}),
472: \label{eq:bound}
473: \eea
474: where $r_{\rm b}=r_{\rm c}-0.6\,R_\odot$. Thus, we assume that $\Delta\Omega$ initially increases with the depth linearly
475: up to its maximum value $\Delta\Omega\,(0,r_{\rm b})$. Taking into account that for the present-day Sun $\Omega\,(r_{\rm c})\approx 2.9\,\mu$Hz,
476: the initial rotational shear in our computations decreases with the depth $z=(r_{\rm c}-r)/R_\odot$ as $q\equiv (\partial\ln\Omega/\partial\ln r)\approx 3.1(x_{\rm c}-z)u_{\rm max}$, where
477: $x_{\rm c} = r_{\rm c}/R_\odot$, and $u_{\rm max}=\Delta\Omega\,(0,r_{\rm b})/10^{-6}$. In our basic parameter set, we will
478: use a value of $u_{\rm max} = 10^{-4}\,\mu$Hz\,$\ll\Omega(r_{\rm c})$.
479:
480: To solve eq. (\ref{eq:ampdf}) one needs to know stellar structure parameters,
481: such as $\rho$, $N$, $K$, and others, as functions of radius and time. However, because the internal structure of
482: the Sun has not changed appreciably in the last billion years, we will use
483: our model of the present-day Sun as a background for all of our IGW computations
484: and we will watch that the total integration time in each of them does not exceed $\sim$\,1\,Gyr.
485: We have solved eq. (\ref{eq:ampdf}) using an original method described in Appendix~A.
486:
487: \subsection{Constant Viscosity}
488:
489: Fig.~\ref{fig:f4} shows our results obtained for different values of the constant viscosity $\nu_{0,n}\equiv \nu/10^n$
490: using the spectra (\ref{eq:zea97}) (panel a) and (\ref{eq:tea02}) (panel b).
491: This figure illustrates the aforementioned bifurcation
492: from stationary to oscillatory solutions that occurs at $\nu_{0,8}\approx 5\times 10^{-4}$ in panel a and
493: at $\nu_{0,8}\approx 8\times 10^{-4}$ in panel b, a sequence of oscillatory solutions (these are analogs of the SLOs
494: discussed by \citealt{tch05}) for $5\times 10^{-5} \leq \nu_{0,8} \leq 5\times 10^{-4}$ in panel a and
495: for $7\times 10^{-5} \leq \nu_{0,8} \leq 8\times 10^{-4}$ in panel b, and a transition to
496: the chaotic behavior of $\Delta\Omega$ at $\nu_{0,8} = 3.8\times 10^{-5}$ in panel a and
497: at $\nu_{0,8} = 4.9\times 10^{-5}$ in panel b.
498: Although in these computations we have only taken into account IGWs with frequencies from
499: the narrower interval $0.3\,\mu\mbox{Hz}\leq\omega\leq 0.6\,\mu\mbox{Hz}$, this truncation does not
500: depreciate our results because
501: waves with these low frequencies carry 75\% and 91\% of the total energy for the spectral distributions
502: (\ref{eq:zea97}) and (\ref{eq:tea02}), respectively. A comparison of panels a and b shows
503: that the use of either of the two IGW spectra leads to similar qualitative results.
504:
505: Since the low-frequency waves get absorbed
506: very close to the core/envelope interface, we have taken the zooming parameter $k=2.6$ (see Appendix~A) in order to resolve
507: the short lengthscale variations of the rotation profile produced by them. For this $k$ and for the used value of $n=8$,
508: we get a time scaling factor of $9.7\,$yr for our dimensionless PDE (\ref{eq:ampdf2}).
509: The real timescales of the oscillations of $\Delta\Omega$ are somewhat longer than this because, additionally,
510: they are inversely proportional to the amplitude of the second term on the right-hand side of the PDE
511: which is of order $10^{-2}$\,--\,$10^{-3}$. Therefore, the minimum time intervals between consecutive
512: curves in Fig.~\ref{fig:f4} are $10^3$ yr (panel a) and $10^2$ yr (panel b).
513: A decrease of $\nu_0$ leads to both longer timescales and larger amplitudes of the oscillations
514: because, in order to compete with the disturbing action of IGWs,
515: a smaller viscosity needs a stronger shear to be built up, which means longer viscous dissipation times.
516:
517: So far, we have used our basic parameter set that includes a limited number of $l=1,\,2,\,3$, and 4, and uses
518: $u_{\rm max} = 10^{-4}\,\mu$Hz in the initial and boundary conditions (\ref{eq:init}\,--\,\ref{eq:bound}).
519: Panel a in Fig.~\ref{fig:f5} demonstrates that neither the addition of 4 extra $l$ values nor the increase of
520: $u_{\rm max}$ by the factor $10^3$, which results in the initial shear $q\approx 0.22$ near
521: the interface, change much the period and amplitude of the oscillations of $\Delta\Omega$.
522:
523: Although the changes caused by the increase of $\Delta\Omega(0,r_{\rm b})$ ($u_{\rm max}$)
524: turn out to be unimportant for our investigation of the ability of IGWs to disturb
525: the internal uniform rotation of the present-day Sun, they were shown by \cite{tch05} to be a matter
526: of great importance in the problem of angular momentum extraction by IGWs from the radiative core of a young
527: solar-type MS star. Indeed, a steep initial $\Delta\Omega$-profile results in the SLOs that are asymmetric with respect
528: to the line $\Delta\Omega=0$ (red curves in Fig.~\ref{fig:f5}a). Such SLOs may work as a filter that predominantly
529: absorbs prograde waves. This means that among low-degree high-frequency waves that arrive at a rapidly rotating central part of the star
530: retrograde waves transporting negative angular momentum will be over-represented. Hence, when
531: being damped in the core they will spin it down.
532:
533: \subsection{Shear-Induced Viscosity}
534:
535: The constant viscosity has been adjusted by hand to get one of the three possible outcomes of the competition
536: between the disturbing action of IGWs and the smoothing effect of the viscous force on the rotation profile.
537: It turns out that the quasi-periodic oscillations of $\Delta\Omega$ are settled only when $\nu_0$ takes on a value
538: from a rather narrow interval. It is unlikely that this accidentally happens in real stars. For the IGW filter
539: composed of quasi-regular temporal and radial variations of $\Delta\Omega$ near the top of radiative core
540: to work as proposed by \cite{tch05} some self-regulating mechanism for adjusting the proper viscosity values should
541: apparently be operating there. One such mechanism could be a shear-induced viscous friction.
542: In this case, a shear is readily built up as a result of selective damping of
543: IGWs. Absorbed waves deposit their angular momentum locally, thus pushing rotation away from its stationary state.
544: The viscosity coefficient (\ref{eq:nuv}) used by us is appropriate for describing mixing due to the rotation-induced secular shear
545: instability (\citealt{mm96}). It develops more easily than the dynamical shear instability but it acts
546: on a smaller length scale $l_{\rm t}$, such that a turbulent eddy of size $\sim$\,$l_{\rm t}$
547: can effectively exchange heat with its surroundings while it travels a mean free path of order $\sim$\,$l_{\rm t}$.
548:
549: \cite{tch05} have implemented this mechanism as follows.
550: They averaged the turbulent diffusion coefficient (\ref{eq:nuvtz}) over a complete oscillation cycle as well as
551: over a radial extent of the SLOs using a Gaussian of width $0.2H_P$. Thus obtained stationary viscosity profile
552: was then used in their IGW computations. As is said in Appendix~B, we actually use the same viscosity (\ref{eq:nuv})
553: but we allow it to vary with time and we do not average it over radius.
554: For the same choice of IGW parameters as in the preceding section but
555: for the shear-induced viscosity (\ref{eq:nuv}) calculated with $f_{\rm v}=1$,
556: results of our solution of the PDE (\ref{eq:ampdf2}) with the spectrum (\ref{eq:zea97})
557: are plotted in Fig.~\ref{fig:f5}b.
558:
559: It is important to note that, like a few other publications (e.g., \citealt{kmg01}; \citealt{tea02}), the work of \cite{tch05}
560: only contains a discussion of the SLOs near the core/envelope interface. Unlike them, we have decided
561: to address the question whether the SLOs die out at greater depth or not. In order to shorten our computation time
562: (i.e. in order to allow longer time steps)
563: when solving the PDE (\ref{eq:ampdf2}) in the bulk of radiative core we have bounded the IGW frequency
564: by the values $0.6\,\mu\mbox{Hz}\leq\omega\leq 3\,\mu$Hz, i.e. we have cut off the IGWs that produce
565: the SLOs very close to the interface, like those shown in Fig.~\ref{fig:f5}b. In spite of this, our minimum frequency
566: still approximately equals the lowest frequency $0.5\,\mu$Hz used by \cite{cht05}.
567: It is important to note that the IGW spectra we use are still normalized by equation
568: (\ref{eq:neteflux}) over the whole frequency and spherical degree intervals:
569: $0.3\,\mu\mbox{Hz}\leq\omega\leq 3\,\mu\mbox{Hz}$, and $1\leq l\leq l_{\rm c}=30$.
570: Because of its rapid decline with an increase of $\omega$ a spectrum normalized with
571: $\omega_{\rm min} > 0.3\,\mu$Hz would have more energetic high-frequency waves than ours,
572: hence it would produce even stronger oscillations of $\Delta\Omega$ than those obtained by us.
573:
574: We have solved equation (\ref{eq:ampdf2}) in the $k$\,$=$\,1-zoomed depth interval $0\leq\hat{z}\leq\hat{z}_{\rm b}=10^k\times(x_{\rm c}-x_{\rm b})$,
575: where $x_{\rm c}=0.713$ and $x_{\rm b}=x_{\rm c}-0.6$,
576: for time periods less than 1\,Gyr. Hoping to relate, later on, the angular momentum transport and element mixing by IGWs
577: in the Sun to canonical extra mixing in low-mass red giants, we have taken the parameter $f_{\rm v}=20$ in
578: the expression (\ref{eq:nuv}) for the shear-induced viscosity because with about that value \cite{dea06} succeeded in reproducing
579: evolutionary abundance variations of Li and C in the atmospheres of cluster and field red giants.
580: We have found that, even after having been enhanced by this large factor, the shear-induced viscosity
581: fails to extinguish large scale SLOs deeper in the solar radiative core. The blue curve in Fig.~\ref{fig:f6}a represents
582: an envelope of oscillation amplitudes of $\Delta\Omega$. The farther inward from the core/envelope interface,
583: the higher the $\Delta\Omega$ oscillation amplitude
584: and the longer its characteristic time are.
585: In the outer half of the radiative core the maximum amplitudes by far exceed the deviations of
586: the helioseismic data (red squares with error bars) from the uniform rotation profile. A better agreement with
587: the experimental data is obtained if we choose $f_{\rm v}=10^3$ (purple curve in panel a).
588: We have tried this value as well because the 3D hydrodynamic simulations by
589: \citet{bh01} of turbulent mixing induced by the shear instability have shown
590: that equation (\ref{eq:nuv0}) may underestimate the coefficient of turbulent
591: diffusion by three orders of magnitude. Of course, it is not clear
592: whether the results of \cite{bh01} can be applied directly to real
593: stars, given that they were obtained assuming plane-parallel geometry and
594: without taking into account ``the effects of rotation, nuclear reactions, and
595: variations in radiative processes''.
596:
597: However, before making any conclusions from the results of these computations
598: we have to check if the viscosities induced by the shear flows outlined in panel a
599: are high enough for the turbulent fluid motions producing them not to be broken down by
600: the molecular viscosity. For this to be true, the flow Reynolds number $Re=\nu_{\rm v}/\nu_{\rm mol}$
601: must exceed the critical Reynolds number $Re_{\rm c}\approx 40$ (\citealt{schea00}).
602: In panel b, we have plotted a time averaged $\nu_{\rm v}$ for the cases of
603: $f_{\rm v}=20$ (blue curve) and $f_{\rm v}=10^3$ (purple curve). In the same plot,
604: green curve depicts $\nu_{\rm mol}$ while red curve presents the quantity $Re_{\rm c}\times\nu_{\rm mol}$.
605: Comparing blue, purple, and red curves in panel b, we conclude that the shear-induced
606: turbulence can only be sustained near the base of the convection zone
607: where both the shear $q$ is sufficiently strong thanks to the very short lengthscales of
608: the SLOs and the quantity proportional to $\nu_{\rm v}f_{\rm v}^{-1}(\Omega q)^{-2}$
609: (eq. \ref{eq:nuv0} and dashed curve in Fig.~\ref{fig:f3})
610: steeply increases with $r$. This raises the following important question:
611: what alternative shear dissipating mechanism works in the Sun's outer radiative core
612: that successfully competes (as follows from the helioseismic data) with the disturbing
613: action of IGWs?
614:
615: \subsection{Molecular Viscosity and Ohmic Diffusivity}
616:
617: Fig.~\ref{fig:f6} shows that the viscosity needed to counteract the distortion of
618: the solar rotation profile by IGWs should not necessarily be too high.
619: Taking into account that the blue and purple curves in Fig.~\ref{fig:f6}b
620: represent the time averaged $\nu_{\rm v}$, whose real values change with time following
621: the oscillations of $\Delta\Omega$, it seems worth testing if the molecular viscosity
622: can smooth out the large scale SLOs alone.
623: The shear-induced turbulent viscosity can only be used down to a depth $z\approx 0.04$\,--\,0.05
624: because below this region the ratio $\nu_{\rm v}/\nu_{\rm mol}$ becomes smaller
625: than the critical Reynolds number (Fig.~\ref{fig:f6}b). It is interesting that the size of this region
626: approximately coincides with the thickness of the solar tachocline in which the latitudinal
627: differential rotation of the convective envelope is transformed into the quasi-solid body rotation
628: of the radiative core (e.g., \citealt{sz92,chea98}).
629: We want to find out if the viscous force in our computations fails to reduce
630: amplitudes of the SLOs in the outer
631: radiative core to values consistent with the helioseismic data simply because we use
632: the IGW spectrum (\ref{eq:zea97}) instead of (\ref{eq:tea02}).
633:
634: It is possible that this negative result is a function of the assumed IGW spectrum.
635: In order to respond to these questions,
636: we have employed the spectrum (\ref{eq:tea02}) and a combined
637: viscosity $\nu = \nu_{\rm v}\,(f_{\rm v})+f_{\rm mol}\times\nu_{\rm mol}$, where
638: $\nu_{\rm v}\,(f_{\rm v})$ is a substitute of equation (\ref{eq:nuv}) in which we set $f_{\rm v}=20$ for
639: $0\leq z\leq 0.04$ and $f_{\rm v}=0$ for $0.04 < z\leq z_{\rm b}$.
640: We have also investigated models with more vigorous IGWs.
641: Furthermore, as \cite{cht05} claimed, the total energy luminosity of IGWs produced by
642: fluctuating Reynolds stresses in the convective envelope of their solar model was
643: $8.5\times 10^{29}$\,erg\,s$^{-1}$. This is about 2.7 times as large as our estimated value
644: of $L_E(r_{\rm c})\approx 3.2\times 10^{29}$\,erg\,s$^{-1}$. Therefore, we have increased our turbulent Mach number
645: ${\cal M}_{\rm t}=8.3\times 10^{-4}$ by this factor and renormalized
646: the spectrum (\ref{eq:tea02}) respectively. Results of these computations obtained with
647: the factor $f_{\rm mol}=2$ are plotted with purple curves in Fig.~\ref{fig:f7}.
648: For test purposes, we have also repeated these computations with an extended set of the spherical
649: degree $l=1,2,\ldots,7,8$ (blue curves). Fig.~\ref{fig:f7} shows that
650: the molecular viscosity (even after it has been doubled) cannot compete with the IGWs
651: that have been shown by \cite{cht05} to be powerful enough to shape the Sun's solid body rotation.
652:
653: On the other hand, it turns out that a viscosity proportional to the ohmic diffusivity
654: $\nu = f_{\rm mag}\times\eta_{\rm mag}$ can extinguish the IGW-induced SLOs everywhere in the solar radiative
655: core except the tachocline region for a value of $f_{\rm mag}\ga 10$
656: (Fig.~\ref{fig:f8}). This seemingly pure academic exercise has some sense.
657: Let us assume that differential rotation in the solar radiative core has been suppressed by magnetic processes,
658: e.g. like those proposed by \cite{chmg93}, \cite{mlm06}, or \cite{s99} (for more details on the latter, see next section).
659: However, in order that magnetic fields generated by these processes not to decay too quickly through the ohmic dissipation,
660: an effective magnetic diffusivity $\eta_{\rm e}$ associated with them must exceed $\eta_{\rm mag}$.
661:
662: \subsection{Effective Magnetic Viscosity}
663: \label{sec:effmag}
664:
665: \cite{s99} has proposed a magnetohydrodynamic mode of angular momentum
666: transport in radiative zones of
667: differentially rotating stars. Fluid elements experience large-scale horizontal
668: displacements caused by an unstable configuration of
669: the toroidal magnetic field (one consisting of stacks of loops concentric with the rotation axis).
670: Small-scale vertical displacements of fluid elements are coupled to the horizontal motions,
671: which can cause both mild mixing and much more effective angular momentum transport.
672: Spruit's key idea
673: is that no initial toroidal magnetic field is actually needed
674: to drive the instability and mixing because the unstable field configuration
675: can be generated and maintained by differential rotation in a process similar to convective dynamo.
676: The Spruit dynamo cycle consists of two consecutive steps: first,
677: a poloidal field is generated by the vertical displacements of the unstable toroidal field; second,
678: the new poloidal field is stretched into a toroidal field by differential rotation.
679:
680: The Spruit mechanism produces a huge magnetic viscosity $\nu_{\rm e}\ga 10^9$\,cm$^2$\,s$^{-1}$
681: (\citealt{dp07}) that could indeed prevent
682: IGWs from disturbing the solar uniform rotation. However, it could produce that uniform rotation itself,
683: without being assisted by IGWs (\citealt{eea05}).
684:
685: The original prescription for the effective magnetic diffusivity and viscosity in the model of Spruit's dynamo
686: has recently been criticized by \cite{dp07} (see also \citealt{zea07}). The principal critical argument is that
687: \cite{s99} has overestimated the horizontal length scale of the Tayler instability that causes the concentric
688: magnetic loops to slip sideways. Spruit assumed the length was of order a local stellar radius. \cite{dp07}
689: suggested that one has to account of the Coriolis force when estimating the instability's horizontal length scale.
690: This reduces the diffusivity by about three orders of magnitude and results in the following expression:
691: \bea
692: \eta_{\rm e}\approx 2\,\frac{K_6\,\Omega_{-6}^2}{(N_T)_{-3}^2}\,q^2,\ \ \mbox{cm}^2\,\mbox{s}^{-1}.
693: \label{eq:newetae}
694: \eea
695: There is no need to do any further computations to understand that, with this revised
696: diffusivity, magnetic fields generated by the Spruit dynamo in the solar radiative core would immediately be dissipated through
697: the ohmic diffusivity, hence the whole transport mechanism would not function.
698: Indeed, equation (\ref{eq:newetae}) gives $\eta_{\rm e}\approx 1.8\times 10^2$\,cm$^2$\,s$^{-1} < \eta_{\rm mag}$
699: near the base of the solar convection zone even if we take $q\approx 1$ which obviously exceeds the upper limit constrained by the helioseismic data.
700: The revised prescription may only work to reduce differential rotation in a model of the young Sun in which both $\Omega_{-6}$ and $q$
701: have much larger values (Denissenkov et al., in preparation).
702:
703: \section{Conclusion}
704:
705: Our numerical solutions of the angular momentum transport equation (\ref{eq:ampdf}, or \ref{eq:ampdf2}) have demonstrated that neither the molecular viscosity
706: nor the shear-induced turbulent viscosity can reduce the large scale oscillations of angular velocity in the solar outer radiative core
707: caused by selective damping of IGWs, provided that the net energy flux of these waves is
708: strong enough to shape the Sun's solid body rotation.
709: If waves were the sole mechanism, we would therefore expect to see large deviations from rigid rotation.
710: The amplitudes of these oscillations are found to be too large to agree with the helioseismic data.
711: Our result holds even when the molecular and shear-induced viscosity are multiplied by large factors.
712: We have proved that only a viscosity exceeding the ohmic diffusivity by a factor of $\ga$\,10 can smooth out the IGW-induced
713: oscillations of the rotation profile. This may be an indirect indication
714: that some magnetic processes are at work here.
715: To be more precise, our finding actually satisfies a necessary condition for such processes to work because
716: the effective magnetic diffusivity $\eta_{\rm e}$ associated with them must exceed the ohmic diffusivity $\eta_{\rm mag}$.
717: Otherwise magnetic fields generated by them will decay through the ohmic resistivity too quickly.
718: For example, we have found that
719: magnetic torques are strong enough to successfully compete with the action of IGWs
720: only if the original prescription for the Taylor-Spruit dynamo, in which $\eta_{\rm e}\gg\eta_{\rm mag}$, is used.
721: However, this particular case turns out to be irrelevant to our problem because the Tayler-Spruit mechanism can
722: shape the solar solid body rotation alone (\citealt{eea05}), without being assisted by IGWs.
723:
724: The helioseismic data suggest that either there is an efficient angular momentum transport mechanism
725: in addition to IGWs that smooths out the SLOs produced by the waves or
726: the spectral energy distribution of IGWs is different (lower) from those used by us. Although the latter assumption
727: leads us outside the scope of our formulated problem we will comment on it.
728: It is possible that strong toroidal magnetic fields in the solar tachocline filter out
729: the IGWs with the doppler-shifted frequency $\sigma$ above the Alfv\'{e}n frequency (\citealt{kea99,kmg03}).
730: For the minimum frequency $0.6\,\mu$Hz used in our
731: computations of the large scale SLOs in the solar outer radiative core the magnetic field strength
732: required to prevent the inward wave propagation is about $(3\times 10^5)/l$ Gauss (\citealt{kea99}).
733: For $l=1$, this corresponds to quite a strong field. If it is present in the tachocline then
734: the waves with high spherical degrees will be trapped there. However, we have only considered
735: IGWs with $\omega \geq 0.6\,\mu$Hz and low l values.
736: Most of them are likely to propagate below the tachocline.
737: In order to find out if an enhanced viscosity
738: in the tachocline can hinder the propagation of low-degree high-frequency waves into the solar radiative core
739: we have done test computations in which the shear-induced viscosity was increased by a factor of $10^3$.
740: Their results plotted with black curves in Fig.~\ref{fig:f7} show that this does not help to solve the problem.
741:
742: Another, more radical possibility is
743: that the form of the IGW spectra employed by us is completely wrong. For instance, the IGW spectrum
744: estimated in the 2D hydrodynamic simulations by \cite{rg05} has a flat energy distribution
745: which goes three orders of magnitude below the peak luminosity in our Fig.~\ref{fig:f1}.
746: Apparently, if we applied that spectrum in our computations then even the molecular viscosity
747: could easily smooth out the SLOs produced by such IGWs. However, it is evident as well that IGWs with
748: this energy distribution could not produce the uniform rotation of the Sun by its present age
749: (multiply the ages of the rotation profiles in Fig.~1 from \citealt{cht05} by a thousand).
750:
751: To summarize, we do not see what microscopic or pure hydrodynamic processes could smooth out the large scale SLOs
752: induced by IGWs in the solar outer radiative core.
753: Therefore, we agree with the conclusion made by \cite{gmci98} about ``the inevitability of
754: a magnetic field in the Sun's radiative interior''.
755: Indeed, if IGWs are as strong as described by our employed spectra then this magnetic field
756: is required to trigger magnetic processes that will counteract the disturbing action of IGWs on the
757: solar rotation profile. On the contrary, if IGWs are weak then we are in need of such magnetic processes
758: to extract an excess angular momentum from the solar interior on the early MS.
759:
760: During the preparation of this work, results of a preliminary analysis of the GOLF data on solar g-mode
761: oscillations have been published by \cite{gea07} suggesting that the solar core at $r\la 0.15\,R_\odot$
762: may rotate three to five times as fast as the rest of the radiative core.
763: If these results prove to be correct, they will seem to rule out
764: the angular momentum redistribution in the Sun by IGWs, as proposed by \cite{cht05}, because in that case it
765: would have been the Sun's inner core to be spun down first. If these results are confirmed, it will mean that the strong $\mu$-gradient
766: in the Sun's central region has prevented any angular momentum transport mechanism from operating there.
767: There would also be strong implications for magnetic angular momentum transport, ruling out prescriptions
768: (such as \citealt{s02}) that predict a weak sensivity to $\mu$ gradients.
769:
770: \acknowledgements
771: We acknowledge support from the NASA grant NNG05 GG20G.
772: PAD thanks Tamara Rogers for usefull discussions and the HAO staff for the warm hospitality.
773: The National Center for Atmospheric Research is sponsored by the National Science Foundation.
774:
775: \appendix
776:
777: \section{Solution of the Angular Momentum Transport Equation}
778: \label{sec:solution}
779:
780: In order to shrink the problem's parameter space,
781: we consider a contribution to the transport of angular momentum only from the waves possessing
782: the maximum possible azimuthal number $|m|=l$, i.e. from those carrying the maximum angular momentum, both positive
783: and negative, at given values of $l$ and $\omega$.
784: To solve the main PDE (\ref{eq:ampdf}), we use a numerical method based on the ideas implemented
785: by \cite{s90} in his model of the wind flow in the Earth's stratosphere influenced by tropospheric IGWs.
786:
787: Taking into account the assumed axial symmetry of the IGW spectrum, i.e. that
788: $S_E(r_{\rm c},\,l,\,m,\,\omega) = S_E(r_{\rm c},\,l,\,-m,\,\omega)$,
789: and the fact that the optical depth
790: in the wave attenuation factor only depends on even powers of the doppler-shifted frequency $\sigma$ (eq. \ref{eq:tau}),
791: we recast the net angular momentum flux for $m=\pm l$ as
792: \bea
793: F_J(r)\equiv \frac{L_J(r)}{4\pi r_{\rm c}^2} = \sum_{l=1}^{l_{\rm max}}\,l\int_{-\omega_{\rm max}}^{\,\omega_{\rm max}}
794: S_J(l,\omega)\exp\{-\tau(r,l,l,\omega)\}\,d\omega,
795: \label{eq:netjflux2}
796: \eea
797: where
798: \bea
799: S_J(l,\omega) \equiv S_E(r_{\rm c},l,l,|\omega|)\,\omega^{-1}.
800: \label{eq:s}
801: \eea
802: In the interval $-\omega_{\rm c} < \omega < \omega_{\rm c}$, we set $S_J(l,\omega)\equiv 0$.
803: Note that $S_J(l,-\omega) = -S_J(l,\omega)$.
804:
805: From the computational standpoint, we find it convenient to convert eq. (\ref{eq:ampdf}) to
806: the following dimensionless form:
807: \bea
808: \frac{\partial u}{\partial\hat{t}} = \frac{1}{\rho x^4}
809: \frac{\partial}{\partial\hat{z}}\left(\rho x^4\nu_n\,\frac{\partial u}{\partial\hat{z}}\right) -
810: \frac{a_{n,k}}{\rho x^4}\left(\frac{L}{L_\odot}\right)\frac{\partial F}{\partial\hat{z}},
811: \label{eq:ampdf2}
812: \eea
813: where $x = r/R_\odot$, and
814: \bea
815: F(\hat{z},u) = \frac{10^3}{\rho_{\rm c}v_{\rm c}^3}\,
816: \sum_{l=1}^{l_{\rm max}}\,l\int_{-\omega_{\rm max}}^{\,\omega_{\rm max}}S_J(l,\omega)\,
817: \exp\{-\int_0^{\hat{z}}G(\hat{z}',\,l,\,\omega -lu)\,d\hat{z}'\}\,d\omega.
818: \label{eq:dfdz}
819: \eea
820: In the last equation, $G(\hat{z},\,l,\,\omega -lu) = [l(l+1)]^{3/2}\,\gamma^{-1}(\hat{z})\,(\omega -lu)^{-4}$,
821: where $\gamma^{-1}(\hat{z}) = 10^{-k}\times 0.2065\times K_6\,N_{-3}\,(N_T)_{-3}^2\,(x_{\rm c} - 10^{-k}\hat{z})^{-3}$
822: with $x_{\rm c} = r_{\rm c}/R_\odot$. We neglect the square root $\sqrt{1-(\sigma/N)^2}$ in the integrand's
823: denominator in eq. (\ref{eq:tau}) because we will only consider $\omega_{\rm max}\ll N_{\rm c}$
824: in which case $|\sigma|\equiv |\omega - lu| < 2\omega_{\rm max}\ll N$.
825:
826: When deriving eqs. (\ref{eq:ampdf2}\,--\,\ref{eq:dfdz}),
827: we have normalized our basic variables, which are assumed to be initially expressed in cgs units, as follows:
828: $u = \Delta\Omega/10^{-6}$, $\nu_n = \nu/10^n$, $\hat{z} = 10^k\times (x_{\rm c}-x)$,
829: $\hat{t} = (10^{n+2k}/R_\odot^2)\,t$, $K_6 = K/10^6$, and $N_{-3} = N/10^{-3}$. The quantity $k$ is a sort of
830: zooming parameter. Taking $k > 0$ allows us to look with the scrutiny at results of IGW damping taking place close to the base of
831: convective envelope, where variations of $u$ may occur on a very short lengthscale of order $10^{-3}$\,--\,$10^{-2}$\,$R_\odot$.
832: Having done these transformations, it turns out that $a_{n,k} = 1.362\times 10^{8-n-k}$, $\hat{z} = 0$ corresponds
833: to the core/envelope interface, and $t$ is measured in units of $1.535\times 10^{14-n-2k}$\, yr.
834: The factor $10^3$ in eq. (\ref{eq:dfdz}) comes about from a combination of
835: the factor $0.1$ that estimates the ratio of the convective flux to the total flux from the star
836: and the factor $10^4$ that represents the reciprocal to the normalization constant for
837: the turbulent Mach number that has been included into the coefficient $a_{n,k}$.
838: Given that $\omega$ is measured in $\mu$Hz, $a_{n,k}$ has additionally been multiplied by the factor $10^6$.
839: This is necessary to do because the wave frequency that we actually use is $\omega \equiv \omega/10^{-6}$, therefore,
840: when substituted into eq. (\ref{eq:dfdz}), relation (\ref{eq:s}) should be taken in the form
841: $S_J = 10^6\times S_E\,(\omega/10^{-6})^{-1}$.
842:
843: In equations (\ref{eq:ampdf2}\,--\,\ref{eq:dfdz}), all integrals are replaced with series of trapezoids while all derivatives
844: are approximated by finite differences. The zoomed depth interval $0\leq\hat{z}\leq\hat{z_{\rm b}}$ (here, $\hat{z_{\rm b}}=10^k(r_{\rm c}-r_{\rm b})/R_\odot$)
845: is divided into $M$ equal subintervals by $M+1$ mesh points while the axisymmetric frequency intervals
846: $-\omega_{\rm max}\leq\omega\leq-\omega_{\rm c}$ and $\omega_{\rm c}\leq\omega\leq\omega_{\rm max}$, representing
847: the retrograde and prograde waves, respectively, are divided into $L$ subintervals each.
848: The resulting system of $M+1$ nonlinear algebraic equations is linearized assuming that after every integration time step $\Delta\,\hat{t}$
849: the ratio $\max\{|\Delta u_j/u_j|\}_{j=2}^M\ll 1$. In order to speed up the computations, we follow the idea of \cite{s90} to interpolate
850: functions containing $G(\hat{z_j},l,\omega_i-lu_j)$ in $\hat{z_j}$, $l$, and the combination $(\omega_i-lu_j)$ using
851: initially prepared and stored tables.\footnote{We take advantage of the fact that $|\omega_i-lu_j|<2\omega_{\rm max}$.}
852:
853: \section{Viscosity Prescriptions}
854: \label{sec:visc}
855:
856: The viscosity $\nu$ in eq. (\ref{eq:ampdf}) plays a very important role because, depending on its value,
857: the viscous friction either succeeds or not in smoothing out
858: oscillations of the rotation profile in the radiative core growing in response to the local deposit of angular momentum that
859: accompanies the absorption of IGWs. In this work, we employ 5 different prescriptions for $\nu$ as well as some of their
860: combinations. These are a constant viscosity $\nu_0$,
861: the molecular viscosity $\nu_{\rm mol}$ (it dominates over the radiative viscosity in the solar-type MS stars),
862: a viscosity proportional to
863: the magnetic (ohmic) diffusivity $\eta_{\rm mag}$, a viscosity $\nu_{\rm v}$ associated with vertical turbulence
864: produced by the secular shear instability induced by differential rotation, and an effective
865: viscosity $\nu_{\rm e}$ related to magnetic torques generated by the Tayler-Spruit dynamo (\citealt{s99,s02}).
866:
867: For the viscosity due to the shear-induced vertical turbulence, we use the expression derived by \cite{mm96}
868: multiplying it by a free parameter $f_{\rm v}$
869: \bea
870: \nu_{\rm v} = f_{\rm v}\times\frac{8}{5}\,Ri_{\rm c}\frac{K}{N_T^2}\,\Omega^2q^2,
871: \label{eq:nuv0}
872: \eea
873: where $Ri_{\rm c} = \frac{1}{4}$ is the critical Richardson number, and the shear $q=(\partial\ln\Omega/\partial\ln r)$.
874: The parameter $f_{\rm v}$ takes into account the fact that, according to hydrodynamic simulations by \cite{bh01},
875: the original prescription may underestimate the viscosity by the factor of $\sim 10^3$. Alternatively,
876: \cite{c02} has supposed that $Ri_{\rm c}$ should be at least four times as large as its classical
877: value.
878: After the same normalization used in Appendix~A, we have
879: \bea
880: (\nu_{\rm v})_n = f_{\rm v}\times\frac{2}{5}\,10^{2k-n}x^2\frac{K_6}{(N_T)_{-3}^2}\left(\frac{\partial u}{\partial\hat{z}}\right)^2.
881: \label{eq:nuv}
882: \eea
883:
884: To compute the effective magnetic viscosity, we use both original Spruit's
885: equations (see our \sect{sec:effmag}) and the equations revised by \cite{dp07}
886: \bea
887: \nu_{\rm e} = \left(\frac{r^2\Omega\eta_{\rm e}^2}{q^2}\right)^{1/3},
888: \label{eq:nue0}
889: \eea
890: where the effective magnetic diffusivity is
891: \bea
892: \eta_{\rm e} \approx 2\frac{K}{N_T^2}\,\Omega^2q^2.
893: \label{eq:etae}
894: \eea
895: Combining the last two equations and normalizing the variables, we find
896: \bea
897: (\nu_{\rm e})_n = 2.686\times 10^5\times 10^{\frac{2}{3}k-n}\,\Omega_{-6}\left[x^2\frac{K_6}{(N_T)_{-3}^2}\right]^{2/3}
898: \left(\frac{\partial u}{\partial\hat{z}}\right)^{2/3}.
899: \label{eq:nue}
900: \eea
901:
902: It should be noted that we have implicitly assumed that $N_\mu^2=0$ in eqs. (\ref{eq:nuv0}) and (\ref{eq:etae}).
903: This approximation may be valid in the outer part of radiative core if we neglect the $\mu$-gradients
904: produced by the gravitational settling and radiative levitation of chemical elements.
905: Under this assumption, our choice of $\nu_{\rm v}$ is equivalent to that made by \cite{tch05}.
906: Indeed, although they have used a prescription proposed by \cite{tz97}
907: \bea
908: \tilde{\nu}_{\rm v} = \frac{8}{5}\,Ri_{\rm c}\frac{K}{N_T^2}\,\Omega^2q^2
909: \frac{(1+D_{\rm h}/K)}{1+(N_\mu^2/N_T^2)(1+K/D_{\rm h})}
910: \label{eq:nuvtz}
911: \eea
912: that takes into account a reduction of
913: the stable thermal stratification in the radiative core by strong horizontal turbulence described with
914: a diffusion coefficient $D_{\rm h}\gg\tilde{\nu}_{\rm v}$, putting $N_\mu^2=0$ in eq. (\ref{eq:nuvtz})
915: and noticing that in evolved solar-type MS stars
916: $D_{\rm h}\ll K$ (e.g., see Fig.~14 in the paper of \citealt{tch05}) transforms $\tilde{\nu}_{\rm v}$
917: into our $\nu_{\rm v}$.
918:
919: %\newpage
920:
921: \begin{thebibliography}{}
922:
923: \bibitem[Barnes(2003)]{b03}
924: Barnes, S.~A.~2003, ApJ, 586, 464
925:
926: \bibitem[Bouvier et al.(1997)]{bea97}
927: Bouvier, J., Forestini, M., \& Allain, S.~1997, A\&A, 326, 1023
928:
929: \bibitem[Br\"{u}ggen \& Hillebrandt(2001)]{bh01}
930: Br\"{u}ggen, M., \& Hillebrandt, W.~2001, MNRAS, 320, 73
931:
932: \bibitem[Canuto(2002)]{c02}
933: Canuto, V.~M.~2002, A\&A, 384, 1119
934:
935: \bibitem[Cattaneo et al.(1991)]{cea91}
936: Cattaneo, F., Brummell, N.~H., Toomre, J., Malagoli, A., \& Hurlburt, N.~E., 1991, ApJ,
937: 370, 282
938:
939: \bibitem[Charbonneau \& MacGregor(1993)]{chmg93}
940: Charbonneau, P., \& MacGregor, K.~B.~1993, ApJ, 417, 762
941:
942: \bibitem[Charbonneau et al.(1998)]{chea98}
943: Charbonneau, P., Tomczyk, S., Schou, J., \& Thompson, M.~J.~1998, ApJ, 496, 1015
944:
945: \bibitem[Charbonnel \& Talon(2005)]{cht05}
946: Charbonnel, C., \& Talon, S.~2005, Science, 309, 2189
947:
948: \bibitem[Couvidat et al.(2003)]{cea03}
949: Couvidat, S., Garc\'{\i}a, R.~A., Turck-Chi\`{e}ze, Corbard, T., Henney, C.~J.,
950: \& Jim\'{e}nez-Reyes, S.~2003, ApJ, 597, L77
951:
952: \bibitem[Denissenkov \& VandenBerg(2003)]{dvb03}
953: Denissenkov, P.~A., \& VandenBerg, D.~A.~2003, ApJ, 593, 509
954:
955: \bibitem[Denissenkov et al.(2006)]{dea06}
956: Denissenkov, P.~A., Chaboyer, B., \& Li, K.~2006, ApJ, 641, 1087
957:
958: \bibitem[Denissenkov \& Pinsonneault(2007)]{dp07}
959: Denissenkov, P.~A., \& Pinsonneault, M.~2007, ApJ, 655, 1157
960:
961: \bibitem[Eggenberger et al.(2005)]{eea05}
962: Eggenberger, P., Maeder, A., \& Meynet, G.~2005, A\&A, 440, L9
963:
964: \bibitem[Garc\'{\i}a L\'{o}pez \& Spruit(1991)]{gls91}
965: Garc\'{\i}a L\'{o}pez, R.~J., \& Spruit, H.~C.~1991, ApJ, 377, 268
966:
967: \bibitem[Garc\'{\i}a et al.(2007)]{gea07}
968: Garc\'{\i}a, R.~A., Turck-Chi\`{e}ze, S., Jim\'{e}nez-Reyes, S.~J.,
969: Ballot, J., Pall\'{e}, P.~L., Eff-Darwich, A., Mathur, S., \& Provost, J.~2007, Sience, 316, 1591
970:
971: \bibitem[Goldreich et al.(1994)]{gea94}
972: Goldreich, P., Murray, N., \& Kumar, P.~1994, ApJ, 424, 466
973:
974: \bibitem[Gough \& Mcintyre(1998)]{gmci98}
975: Gough, D., \& Mcintyre, M.~E.~1998, Nature, 394, 755
976:
977: \bibitem[Grevesse \& Noels(1993)]{gn93}
978: Grevesse, N., \& Noels, A.~1993, in Origin and Evolution of the Elements,
979: ed.. N. Prantzos, E. Vangioni-Flam, \& M. Casse (Cambridge: Cambridge Univ. Press), 15
980:
981: \bibitem[Kawaler(1988)]{k88}
982: Kawaler, S.~D.~1988, ApJ, 333, 236
983:
984: \bibitem[Keppens et al.(1995)]{kea95}
985: Keppens, R., MacGregor, K.~B., \& Charbonneau, P.~1995, A\&A, 294, 469
986:
987: \bibitem[Kim \& MacGregor(2001)]{kmg01}
988: Kim, E.-J., \& MacGregor, K.~B.~2001, ApJ, 556, L117
989:
990: \bibitem[Kim \& MacGregor(2003)]{kmg03}
991: Kim, E.-J., \& MacGregor, K.~B.~2003, ApJ, 588, 645
992:
993: \bibitem[Kiraga et al.(2005)]{kea05}
994: Kiraga, M., Stepien, K., \& Jahn, K.~2005, AcA, 55, 205
995:
996: \bibitem[Krishnamurthi et al.(1997)]{kea97}
997: Krishamurthi, A., Pinsonneault, M.~H., Barnes, S., \& Sofia, S.~1997, ApJ, 480, 303
998:
999: \bibitem[Kumar \& Quataert(1997)]{kq97}
1000: Kumar, P., \& Quataert, E.~J.~1997, ApJ, 475, L143
1001:
1002: \bibitem[Kumar et al.(1999)]{kea99}
1003: Kumar, P., Talon, S., \& Zahn, J.-P.~1999, ApJ, 520, 859
1004:
1005: \bibitem[Lighthill(1978)]{l78}
1006: Lighthill, J.~1978, Waves in Fluids, (Cambridge: Cambridge Univ. Press), Chap. 5
1007:
1008: \bibitem[Lindzen \& Holton(1968)]{lh68}
1009: Lindzen, R.~S., \& Holton, J.~R.~1968, J. Atmos. Sci., 25, 1095
1010:
1011: \bibitem[Maeder \& Meynet(1996)]{mm96}
1012: Maeder, A., \& Meynet, G.~1996, A\&A, 313, 140
1013:
1014: \bibitem[Matt \& Pudritz(2008)]{mp08}
1015: Matt, S., \& Pudritz, R.~E.~2008, arXiv:0801.0436v2 [astro-ph]
1016:
1017: \bibitem[Menou \& Le Mer(2006)]{mlm06}
1018: Menou, K., \& Le Mer, J.~2006, ApJ, 650, 1208
1019:
1020: \bibitem[Montalb\'{a}n \& Schatzman(2000)]{ms00}
1021: Montalb\'{a}n, J., \& Schatzman, E.~2000, A\&A, 354, 943
1022:
1023: \bibitem[Plumb(1977)]{p77}
1024: Plumb, R.~A.~1977, J. Atmos. Sci., 34, 1847
1025:
1026: \bibitem[Press(1981)]{p81}
1027: Press, W.~H.~1981, ApJ, 245, 286
1028:
1029: \bibitem[Rieutord \& Zahn(1995)]{rz95}
1030: Rieutord, M., \& Zahn, J.-P.~1995, A\&A, 296, 127
1031:
1032: \bibitem[Ringot(1998)]{r98}
1033: Ringot, O.~1998, A\&A, 335, L89
1034:
1035: \bibitem[Rogers \& Glatzmaier(2005)]{rg05}
1036: Rogers, T.~M., \& Glatzmaier, G.~A.~2005, MNRAS, 364, 1135
1037:
1038: \bibitem[Rogers \& Glatzmaier(2006)]{rg06}
1039: Rogers, T.~M., \& Glatzmaier, G.~A.~2006, ApJ, 653, 756
1040:
1041: \bibitem[Saravanan(1990)]{s90}
1042: Saravanan, R.~1990, J. Atmos. Sci., 47, 2465
1043:
1044: \bibitem[Schatzman et al.(2000)]{schea00}
1045: Schatzman, E., Zahn, J.-P., \& Morel, P.~2000, A\&A, 364, 876
1046:
1047: \bibitem[Sestito \& Randich(2005)]{sr05}
1048: Sestito, P., \& Randich, S.~2005, A\&A, 442, 615
1049:
1050: \bibitem[Sills \& Pinsonneault(2000)]{sp00}
1051: Sills, A., \& Pinsonneault, M.~H.~2000, ApJ, 540, 489
1052:
1053: \bibitem[Spiegel \& Zahn(1992)]{sz92}
1054: Spiegel, E.~A., \& Zahn, J.-P.~1992, A\&A, 265, 106
1055:
1056: \bibitem[Spruit(1999)]{s99}
1057: Spruit, H.~C.~1999, A\&A, 349, 189
1058:
1059: \bibitem[Spruit(2002)]{s02}
1060: Spruit, H.~C.~2002, A\&A, 381, 923
1061:
1062: \bibitem[Stauffer \& Hartmann(1987)]{sh87}
1063: Stauffer, J.~R., \& Hartmann, L.~W.~1987, ApJ, 318, 337
1064:
1065: \bibitem[Stein \& Nordlund(1989)]{sn89}
1066: Stein, R.~F., \& Nordlund, A.~1989, ApJ, 342, L95
1067:
1068: \bibitem[Talon \& Zahn(1997)]{tz97}
1069: Talon, S., \& Zahn, J.-P.~1997, A\&A, 317, 749
1070:
1071: \bibitem[Talon et al.(2002)]{tea02}
1072: Talon, S., Kumar, P., \& Zahn, J.-P.~2002, ApJ, 574, L175
1073:
1074: \bibitem[Talon \& Charbonnel(2003)]{tch03}
1075: Talon, S., \& Charbonnel, C.~2003, A\&A, 405, 1025
1076:
1077: \bibitem[Talon \& Charbonnel(2005)]{tch05}
1078: Talon, S., \& Charbonnel, C.~2005, A\&A, 440, 981
1079:
1080: \bibitem[Yoden \& Holton(1988)]{yh88}
1081: Yoden, S., \& Holton, J.~R.~1988, J. Atmos. Sci., 45, 2703
1082:
1083: \bibitem[Zahn et al.(1997)]{zea97}
1084: Zahn, J.-P., Talon, S., \& Matias, J.~1997, A\&A, 322, 320
1085:
1086: \bibitem[Zahn et al.(2007)]{zea07}
1087: Zahn, J.-P., Brun, A.~S., \& Mathis, S.~2007, A\&A, 474, 145
1088:
1089: \end{thebibliography}
1090:
1091: %-----------------------------------------------------------------
1092:
1093: \clearpage
1094: \begin{figure}
1095: \plotone{f1.eps}
1096: \caption{Logarithms of the energy luminosity of IGWs at the Sun's
1097: core/envelope interface summed over all available azimuthal
1098: numbers ($m=-l,\ldots,l$) as functions of $l$ and $\omega$.
1099: Solid curve corresponds to spectrum (\ref{eq:zea97}),
1100: dashed curves --- to spectrum (\ref{eq:tea02}) for $l=1,\,2,\,3$,
1101: and 4 (from the lower to upper curve).
1102: }
1103: \label{fig:f1}
1104: \end{figure}
1105:
1106: %-----------------------------------------------------------------
1107:
1108: %-----------------------------------------------------------------
1109:
1110: \clearpage
1111: \begin{figure}
1112: \plotone{f2.eps}
1113: \caption{The convective overturn frequency $\omega_{\rm c}$ (dashed curve),
1114: the buoyancy frequency $N$ (solid curve), and
1115: the ratio of the convective flux $\rho v_{\rm c}^3$
1116: to the total flux $F_\odot$ (dotted curve) as functions of the relative depth
1117: (expressed in units of the pressure scale height) on the both sides of
1118: the Sun's core/envelope interface.
1119: }
1120: \label{fig:f2}
1121: \end{figure}
1122:
1123: %-----------------------------------------------------------------
1124:
1125: %-----------------------------------------------------------------
1126:
1127: \clearpage
1128: \begin{figure}
1129: \plotone{f3.eps}
1130: \caption{The frequency $\omega_{\tau=1}$ for which
1131: the effective optical depth (eq. \ref{eq:tau}) in the IGW attenuation
1132: factor $\exp(-\tau)$ equals to one at a given radius in the case of
1133: uniform rotation (from the lower to upper solid curve:
1134: $l$ increases from 1 to 4, and $\omega_{\tau=1}\propto [l(l+1)]^{3/8}$).
1135: Dashed curve is the normalized stellar structure parameter in expression
1136: (\ref{eq:nuv0}) for the shear-induced viscosity.
1137: }
1138: \label{fig:f3}
1139: \end{figure}
1140:
1141: %-----------------------------------------------------------------
1142:
1143: %-----------------------------------------------------------------
1144:
1145: \clearpage
1146: \begin{figure}
1147: %\epsfxsize=11cm
1148: %\epsffile [0 50 400 725] {f4.mod.eps}
1149: %\epsffile [0 100 450 740] {f4.eps}
1150: \epsscale{0.75}
1151: \plotone{f4.eps}
1152: \caption{The shear-layer oscillations (SLOs) of $\Delta\Omega = \Omega(r)-\Omega_{\rm c}$
1153: at the top of solar radiative core computed (eq. \ref{eq:ampdf2})
1154: with the constant viscosity $\nu_{0,8}\equiv \nu_0/10^8$ using
1155: the IGW spectra (\ref{eq:zea97}) (panel a) and (\ref{eq:tea02})
1156: (panel b) for the following sets of parameters:
1157: $0.3\,\mu\mbox{Hz}\leq\omega\leq 0.6\,\mu\mbox{Hz}$;
1158: $l=1,\,2,\,3,\,4$; $m=\pm l$; and $u_{\rm max}=10^{-4}\,\mu$Hz in the boundary and initial
1159: conditions (\ref{eq:init}\,--\,\ref{eq:bound}). Panel a:
1160: black curves --- $\nu_{0,8}=5\times 10^{-4}$, the time interval between
1161: consecutive curves is $\Delta t=10^3$ yr;
1162: blue --- $\nu_{0,8}=10^{-4}$, $\Delta t= 5\times 10^3$ yr;
1163: green --- $\nu_{0,8}=5\times 10^{-5}$, $\Delta t= 5\times 10^3$ yr;
1164: red --- $\nu_{0,8}=3.8\times 10^{-5}$, $\Delta t= 10^4$ yr. Panel b:
1165: black --- $\nu_{0,8}=8\times 10^{-4}$, $\Delta t= 10^2$ yr;
1166: blue --- $\nu_{0,8}= 10^{-4}$, $\Delta t= 10^3$ yr;
1167: green --- $\nu_{0,8}= 7\times 10^{-5}$, $\Delta t= 10^3$ yr;
1168: red --- $\nu_{0,8}= 4.9\times 10^{-5}$, $\Delta t= 2\times 10^3$ yr.
1169: }
1170: \label{fig:f4}
1171: \end{figure}
1172:
1173: %-----------------------------------------------------------------
1174:
1175: %-----------------------------------------------------------------
1176:
1177: \clearpage
1178: \begin{figure}
1179: %\epsfxsize=11cm
1180: %\epsffile [0 100 450 740] {f5.eps}
1181: %\epsffile [0 50 400 725] {f5.mod.eps}
1182: \epsscale{0.75}
1183: \plotone{f5.eps}
1184: \caption{The SLOs at the top of solar radiative core computed
1185: with the constant (panel a) and shear-induced (panel b)
1186: viscosity using the IGW spectrum (\ref{eq:zea97})
1187: and the same sets of basic parameters as in the previous figure.
1188: Exceptions are the enhanced value of $u_{\rm max} = 0.1\,\mu$Hz
1189: for green and red curves, and the extended set of
1190: $l=1,\,2,\ldots,7,\,8$ for red curve. In panel a,
1191: all curves have
1192: $\nu_{0,8}= 10^{-4}$, and $\Delta t= 5\times 10^3$ yr.
1193: Curves in panel b have been computed using the parameter
1194: $f_{\rm v}= 1$ in eq. (\ref{eq:nuv}), and $\Delta t= 5\times 10^3$ yr.
1195: }
1196: \label{fig:f5}
1197: \end{figure}
1198:
1199: %-----------------------------------------------------------------
1200:
1201: %-----------------------------------------------------------------
1202:
1203: \clearpage
1204: \begin{figure}
1205: %\epsfxsize=11cm
1206: %\epsffile [0 100 450 740] {f6.eps}
1207: %\epsffile [25 200 425 650] {f6.mod.eps}
1208: \epsscale{0.75}
1209: \plotone{f6.eps}
1210: \caption{Panel a: blue and purple curves are the envelopes of amplitudes of
1211: the SLOs deep in the solar radiative core computed
1212: with the shear-induced viscosity (eq. \ref{eq:nuv} with
1213: $f_{\rm v}=20$ --- blue curve, and with $f_{\rm v}=10^3$ --- purple curve)
1214: using the IGW spectrum (\ref{eq:zea97}) and
1215: the same sets of basic parameters as in Fig.~\ref{fig:f4},
1216: except for the cut-off frequency interval
1217: $0.6\,\mu\mbox{Hz}\leq\omega\leq 3\,\mu$Hz.
1218: Red squares with error bars in panel a are the helioseismic
1219: data from \cite{cea03}. Panel b: green curve --- the molecular
1220: viscosity $\nu_{\rm mol}$; red --- the quantity $Re_{\rm c}\times\nu_{\rm mol}$,
1221: where the critical Reynolds number $Re_{\rm c}=40$ (\citealt{schea00});
1222: blue and purple curves
1223: --- the time averaged shear-induced viscosities produced by
1224: the SLOs whose envelopes are plotted in panel a with the same color.
1225: }
1226: \label{fig:f6}
1227: \end{figure}
1228:
1229: %-----------------------------------------------------------------
1230:
1231: %-----------------------------------------------------------------
1232:
1233: \clearpage
1234: \begin{figure}
1235: %\epsfxsize=11cm
1236: %\epsfxsize=12cm
1237: %\epsffile [0 100 450 740] {f7.eps}
1238: %\epsffile [25 200 425 650] {f7.mod.eps}
1239: \epsscale{0.75}
1240: \plotone{f7.eps}
1241: \caption{Same as in the previous figure but for spectrum (\ref{eq:tea02})
1242: multiplied by the factor 2.7 (see text)
1243: and for the combined viscosity $\nu = \nu_{\rm v}\,(f_{\rm v}) +
1244: 2\nu_{\rm mol}$. Exceptions are blue curves
1245: that were computed for $l=1,\,2,\ldots,7,\,8$.
1246: In the expression (\ref{eq:nuv}) for $\nu_{\rm v}$,
1247: the following parameters have been used:
1248: $f_{\rm v}=20$ (blue and purple curves) and
1249: $f_{\rm v}=10^3$ (black curve) for $0\leq (r-r_{\rm c})/R_\odot\leq 0.04$,
1250: while $f_{\rm v}=0$ (all curves) for $(r-r_{\rm c})/R_\odot > 0.04$.
1251: For comparison, red curves are computed using spectrum (\ref{eq:zea97}).
1252: }
1253: \label{fig:f7}
1254: \end{figure}
1255:
1256: %-----------------------------------------------------------------
1257:
1258: %-----------------------------------------------------------------
1259:
1260: \clearpage
1261: \begin{figure}
1262: %\epsfxsize=11cm
1263: %\epsfxsize=12cm
1264: %\epsffile [0 100 450 740] {f8.eps}
1265: %\epsffile [25 200 425 650] {f8.mod.eps}
1266: \epsscale{0.75}
1267: \plotone{f8.eps}
1268: \caption{Same as in the previous figure but for spectrum (\ref{eq:zea97})
1269: and for the combined viscosity $\nu = \nu_{\rm v}\,(f_{\rm v}) +
1270: f_{\rm mag}\times\eta_{\rm mag}$. In both cases, the parameter
1271: $f_{\rm v}=20$ has been used in the expression (\ref{eq:nuv}) for $0\leq (r-r_{\rm c})/R_\odot\leq 0.04$.
1272: Blue curve is computed with $f_{\rm mag} = 2$, purple --- with
1273: $f_{\rm mag}=10$. Black curve shows the magnetic diffusivity $\eta_{\rm mag}$.
1274: }
1275: \label{fig:f8}
1276: \end{figure}
1277:
1278: %-----------------------------------------------------------------
1279:
1280: \end{document}
1281:
1282:
1283: