A&A 423, L37-L40 (2004)
DOI: 10.1051/0004-6361:200400034
K. H. Chan 1 - X. Liao 2 - K. Zhang 1 - C. A. Jones 1
1 - Department of Mathematical Sciences, Laver Building,
University of Exeter, EX4 4QE, UK
2 -
Shanghai Astronomical Observatory,
Chinese Academy of Sciences, Shanghai 200030, PR China
Received 4 June 2004 / Accepted 12 July 2004
Abstract
A fully three-dimensional, nonlinear, time-dependent spherical
interface dynamo with a solar-like internal differential rotation
and a single-cell meridional circulation is investigated. There
are four regions in our interface dynamo model: a uniformly
rotating core, a thin differentially rotating tachocline, a
turbulent convection zone and a nearly insulating exterior. The
different regions are coupled by the electric and magnetic
matching conditions at the interfaces. For the first time,
non-axisymmetric spherical nonlinear interface dynamos are found.
When the magnetic Reynolds number based on the differential
rotation is small or when the effect of the meridional circulation
is substantial, we show that the spherical interface dynamos are
non-axisymmetric even though the -distribution,
differential rotation and meridional circulation are all
axisymmetric. Implication of the results for the solar dynamo is
discussed.
Key words: magnetohydrodynamics - Sun: interior - Sun: magnetic field
A highly differentially rotating transition zone between the convection and radiative regions of the Sun, the solar tachocline, plays an essential role in generating large-scale solar magnetic activities such as the 11 yr sunspot cycle (e.g., Spiegel & Zahn 1992; Parker 1993; Tobias et al. 1995). It is the tachocline that offers an ideal location for both the generation and storage of the Sun's strong azimuthal magnetic fields, leading to the first cartesian interface dynamo model proposed by Parker (1993). A key idea in an interface dynamo is that the production of a weak poloidal magnetic field and a strong azimuthal magnetic field occurs in separate fluid regions with discontinuous magnetic diffusivities across their interface.
There exist a number of spherical interface dynamo models (e.g.
Charbonneau & MacGregor 1997; Markiel & Thomas
1999; Dikpati & Charbonneau 1999; Covas
et al. 2001). Two important assumptions have been
employed in nearly all the existing spherical interface dynamos.
They assumed that the generated magnetic field
is
axisymmetric, usually in the form of
![]() |
(1) |
The recent solar observations suggest that the solar dynamo is strongly non-axisymmetric, characterized by magnetic activities in persistent different longitudes and the total number of active regions per rotation per hemisphere varied between zero and seven (e.g., De Toma et al. 2000). Non-axisymmetry of the solar magnetic fields may be attributable to the joint instability of latitudinal differential rotation and concentrated magnetic field in the tachocline, which can produce non-axisymmetric magnetic fields with longitudinal wavenumners up to m=5 (e.g., Gilman & Fox 1997; Gilman & Dikpati 2000; Zhang et al. 2003). De Toma et al. (2000) argued that the number of persistent bands of magnetically active nests in the observed solar magnetic fields corresponds to the longitudinal wavenumber m in the magnetic instability problem.
The primary objective of this letter is to report a new interface dynamo phenomenon discovered in the numerical simulation of a fully three-dimensional finite-element spherical interface dynamo. We found that the spherical interface dynamos with an axisymmetric alpha-effect and an axisymmetric flow (the differential rotation and meridional circulation) are non-axisymmetric when the magnetic Reynolds number based on the differential rotation is small or when the effect of the meridional circulation is substantial.
Our spherical interface dynamo model consists of four different
zones. The inner radiative sphere,
,
has a very small
magnetic diffusivity and rotates uniformly with the angular
velocity
.
We take the reference of frame that rotates
with
.
On the top of the radiative core, a strong
differential rotation with a large radial shear is confined in
the tachocline
with magnetic diffusivity
.
Note that magnetic fields cannot be generated by the
differential rotation even in the presence of radially
discontinuous magnetic diffusivities (e.g., Zhang & Liao
2003). However, the weak poloidal magnetic field diffuses
down into the tachocline where it is strongly sheared, resulting
in a strong magnetic field
.
This
process is described by the following equations
![]() |
(2) |
![]() |
(3) |
![]() |
(4) |
![]() |
(5) |
![]() |
(6) |
Three non-dimensional quantities characterize our spherical
interface dynamo model: the magnetic alpha Reynolds number
and the magnetic Reynolds number
based on the
differential rotation, the magnetic Reynolds number
based
on the meridional circulation, which are defined by
When
and
,
i.e., the generation of the
magnetic field without the effects of both the differential
rotation and the meridional circulation, the resulting dynamo
corresponds to a conventional
dynamo driven by an
axisymmetric
in the convection zone. It should be noted
that our simulation does not impose any azimuthal or equatorial
symmetries. By performing simulations at various values of
,
it is found that the dynamo action first occurs at
about
.
In Fig. 1 (upper panel), we show
the magnetic energy
of the generated magnetic field as a
function of time for two different
.
The structure of
the generated magnetic field for the stationary dynamo at
is depicted in Fig. 2. An important new
feature of the spherical dynamo is that the conventional nonlinear
dynamo is always non-axisymmetric with the azimuthal wavenumber m=4, regardless of the spatial properties of the initial
condition. The non-axisymmetry of the spherical dynamo can be
explained by the effect of the nearly perfectly conducting core
which blocks the effective communication of the generated magnetic
field between the different regions of longitudes, leading to the
preference of a non-axisymmetric dynamo mode.
![]() |
Figure 1:
Magnetic energy ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 2:
Contours of ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
When
is non-zero but
,
i.e., the generation of
the magnetic field is affected by the differential rotation, the
resulting dynamo corresponds to an
dynamo driven
by an axisymmetric
and influenced by the differential
rotation. We have performed a number of simulations for different
values of
,
which are shown in Fig. 1 (lower panel).
It is found that the spherical dynamos are still non-axisymmetric
with the azimuthal wavenumber m=4 when
is insufficiently
large,
.
Figure 3 shows the structure of the
generated magnetic field for
with
on a
spherical surface. When
,
however, the
effective communication of the generated magnetic field between
the different regions of longitudes is re-established by the
differential rotation, leading to a periodic axisymmetric
interface dynamo. In this case, strong azimuthal field is
generated in the tachocline and the magnetic field propagates
radially toward the surface of the convection zone, regardless of
whether
or
.
The typical structure
of the radially propagating magnetic field for a positive
is depicted in Fig. 4 in a meridional
plane.
![]() |
Figure 3:
Contours of ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 4:
Contours of ![]() ![]() |
Open with DEXTER |
When both
and
are non-zero, the resulting dynamo
corresponds to an
dynamo driven by an
axisymmetric
and affected by both the differential
rotation and meridional circulation. An important effect of the
large-scale meridional circulation is to reduce the effective
communication of the generated magnetic field between the
different regions of longitudes. The interface dynamos become
non-axisymmetic, in the form of azimuthally propagating dynamo
waves, when
is sufficiently large. At any fixed meridional
plane, as shown in the butterfly diagram in Fig. 5, the
dynamo wave appears to propagate toward the equator. However, this
only reflects an azimuthally propagating non-axisymmetric dynamo
wave with the wavenumber m=1. The azimuthal structure of the
generated toroidal field for
and
as a function
of time is displayed in Fig. 6 at the interface between
the tachocline and convection zone where the toroidal field is the
strongest. In a real system, a strong toroidal field stored in
the tachocline in a particular longitudinal region would be
susceptible to magnetic buoyancy leading to an eruption or other
magnetic activities in the longitudinal region near the outer
surface of the convection zone (Ferriz-Mas et al.
1994).
![]() |
Figure 5:
A butterfly diagram for the dynamo solution at a fixed
longitude with
![]() ![]() |
Open with DEXTER |
![]() |
Figure 6:
Contours of ![]() ![]() |
Open with DEXTER |
This paper presents the first non-axisymmetric spherical interface dynamos found in a fully three-dimensional, nonlinear dynamo model using a finite element method with three-dimensional tedrahedralization in the spherical system. An important finding is that non-axisymmetric dynamo modes can be preferred in a spherical interface dynamo even though both the alpha-profile and large-scale flow (the differential rotation and meridional circulation) are axisymmetric. This provides a possible explanation of non-axisymmetry of the observed solar magnetic fields (De Toma et al. 2000).
Our interface dynamo model follows the classical picture of Parker
(1955) in that the -effect is assumed to be
operated throughout the convection zone. Alternatively, an
interface dynamo may restrict the
-effect near the surface
of the Sun (Babcock 1961) or near the base of the
convection zone resulting from the hydromagnetic or hydrodynamic
instabilities taking place in the tachocline (e.g., Dikpati &
Gilman 2001). An investigation into the effect of
different locations of the alpha-profile on a nonlinear spherical
three-dimensional interface dynamo is currently under way.
Acknowledgements
This work is supported by UK PPARC and NERC grants. X.L. is supported by the Royal Society and NSFC grants. K.Z. would like to thank Dr. S. Tobias and Professor M. Proctor for many helpful discussions about the solar dynamo problem.