Hydromagnetic Waves Doc 2
Hydromagnetic Waves Doc 2
1. Introduction
Although the problem of the propagation of linear magnetohydrodynamic
waves in an inhomogeneous medium is of great interest in solar physics (e.g. [1]),
,
it has not been yet investigated in sufficient detail. Moreover, the propagation of
MHD waves has been studied mostly in the case where the Alfvén waves decouple
from magnetosonic ones (see e.g. [2, 3]).
Few analytical calculations relating both the Alfvén and the magnetosonic
waves have been attempted. There are some numerical simulations made by Holl-
weg et al. [4]. They conclude that shear Alfven wave in a solar magnetic flux
tube can drive sound waves which eventually dissipate into shocks. It is quali-
tatively suggested that Alfven waves may heat the corona undirectly by driving
the slow mode, with some of the properties of spicules. Sakai and Sonneup [5]
have derived model equations which describe the long dispersive Alfven wave
driving sound wave. On the other hand the equation governing the evolution of `
(335)
336 K. Murawski
the fast wave envelope modulated by a slow wave driven by the ponderomotive
force has been derived for a sausage wave travelling along magnetic photospheric
slab with rigid walls by Sahyouni et al. [6]. Model equations describing dispersive
Alfven-magnetosonic waves interaction have been derived by Shukla et al. [7].
Coupling between magnetosonic waves and tearing mode has been described in
terms of model equations by Sakai and Washimí [8]. Coupled nonlinear Schrödinger
equations governing the interaction of sausage and kink surface waves in a plasma
slab have been derived by Vladimirov et al. [9].
In all above mentioned cases the derived equations are some generalizations
of the Zakharov equations originally derived for the Langmuir wave and slow
density plasma response. A recent source of references on these equations can
be found in Murawski et al. [10]. The Zakharov equations do not take, however, ..
lead to creation of new waves (see Wentzel [11]). Actually, the Alfvén wave can
drive both the slow and fast mode. Due to the small value of the sound speed in
comparison to the Alfven one, we should expect weak coupling in the Alfvén—slow
mode interaction and a strong one for the Alfvén—fast wave interaction. This pro-
cess of driving of the fast wave can be even enhanced by the phase mixing effect
caused by inhomogeneities in the embient magnetic field.
The paper is organized as follows. The next Section presents fundamental set:
of equations for the Alfvén and magnetosonic waves which are shortly described
in Section 3. A dispersion relation for the viscous non-dispersive Alfvén wave is
derived in Section 4. Burgers-type equations describing the Alfvén waves propa-
gation are presented in Section 5. A case of linear polarization is also discussed.
Equations which govern the Alfvén-magnetosonic waves interaction are derived in
the next Section. Numerical results for the fast mode propagating perpendicular
to the ambient magnetic field are shown in the subsection 6.1.2. The final part of
the paper contains a short summary and conclusions.
2. Fundamental equations
The driving force for the Alfven wave is the magnetic tension alone. The
dispersion relation is (for the inviscid plasma)
ω = kVA cos Θ.
So, it can not propagate in a direction perpendicular (Θ = π/2) to the magnetic
field. The Alfvén wave is transverse. The velocity perturbation is normal both
to the applied magnetic field and the propagation direction. The magnetic field
perturbation is perpendicular to Β 0 . There are no pressure or density changes
associated with the wave. The energy flows along the field at the Alfven speed. .
For the dissipation free medium we have got the following dispersion relation:
The sign + corresponds to the fast wave. Its velocity becomes the faster of either
VA or cs for θ = 0: for cs » VA ω/k ≈ cs — acoustical in character (longitudinal
,
The energy flow is confined to near the magnetic field direction. The wave is driven
by tension and pressure forces. The gas and magnetic pressure variation are out
of phase. In the corona the slow mode is more acoustic than magnetic.
Note that essentially this expression is similar to the one for the inviscid and
dispersive Alfvén wave (with the Hall term in the induction equation (2.3) in-
cluded), see e.g. [16]. The important difference is that here stands imaginary unit
i corresponding to a dissipation.
where now ε describes the weakness of nonlinearity [18] and allow the density
to be varied in time. The nonlinear coefficient ε can be defined as the ratio of a
characteristic wave amplitude to a transverse length scale. Collection of terms at
ε (e.g. [19]) as a compatibility condition gives us coupled twodimensional Burgers
equations:
where
Note that B0/VA is a constant. Similar equations have been derived by Ruderman
[16] and Mjolhus and Wyller [17]. They, however, have taken resistivity instead of
viscosity into account.
Equations (5.3-5.5) are difficult to solve analytically. Because there is a
lack of corresponding discussion in the literature, we will simply consider some
one-dimensional cases.
where f Ξ u or υ, and the nonlinear β and dissipative α coefficients are defined as: .
The first term in this equation describes energy. So, we see that energy decreased
in time. .
and integrating (5.3.2) over ζ we get the following ordinary differential equation:
which is very convenient for the phase analysis (e.g. [20]). For α > 0 and β < 0
there are no physical solutions. In the corona, however, both coefficients are greater
than zero. In this case the phase analysis leads to the conclusions that we should
expect finite solutions for -lm < 1 ≤ 1m, where lm Ξ (2s/3) √ s/3β. Under this
constraint for each value of the free parameter 1 there are two shocks characterized
by:
1) expansion shock: αn < V < b and increases with ς,
2) compression shock: b < V < c and decreases with ς,
where α < b < c are the roots of the equation made from the right hand side
(r. h. s.) of Eq. (5.3.4). For 1 = l m , b = c and the expansion shock solution is
given by
The ΑΙfvén wave drives both the slow and fast waves. So, let us consider
a mix of these waves. Because in the corona the sound speed is very small the
Αlfvén—fast mode interaction is very strong. In other words, because in the corona
Alfven-Magneosonic Waves Interaction 341
vfVA, the process of driving of the fast waves is more efficient than the cor-
responding one for the slow mode. Due to a phase mixing we can expect that in
the upper corona the fast waves can propagate obliquely to the magnetic field.
Other directions of the propagation are also allowed but due to the inhomogeneity
in the x-direction waves are mostly damped. In the following sections we consider
both short and long Alfven waves limits. The short Alfven wave interacts itself
(selfmodulation) and also is modulated by the magnetosonic wave response.
where f0 (x) is the undisturbed state, fi and f2 describe the long slear Alfvén and
magnetosonic waves, respectively. Because the magnetosonic waves are driven by
the Alfven wave we should expect that the former ones are in lower magnitnde :
Other flrst order quantities are taken to be zero. Due to the dispersion relation for
the Alfvén wave we use the same variables stretching as in Sec. 5, (see Eq. (5.1)).
From Eqs. (2.1-2.5) collecting terms at ε, we get
Note that the Alfvén wave variables are coupled to the vertical magnetic field,
velocity, density and pressure. The process of interaction is thus much more com-
plicated than in the case of the Zakharov equations.
The magnetosonic waves are described by equations which are obtained at ε2:
342 K. Murawski
where
Note the ponderomotive terms at the r. h. s. of (6.1.5) and (6.1.7) and lack of any
one in (6.1.6). Thus we see that the flow is driven due to the existence of gradients
in this direction described by the first terms of the r. h. s. These equations contain
also the pseudo-damping term c2Axu2x connected with a phase mixing. We can
prove it by considering the equation:
Note that this equation is adiabatic. So at the level of these calculation8 there
is no coronal heating. It is usefull, however, to provide approximate solution for
the propagating vertically slow wave even for ths case. The full problem does not
seem to be solvable in an analytical way.
Let us assume that the Alfven wave is linearly polarized in the x-direction
(υ 1 = 0) and look for the stationary solutions ( ξ = z - ct). Then, we get
Choosing other values of n we can find other solutions. This problem, however, is
not discussed in this paper.
6.1.3. Fast mode lmit
Burgers equation in the limit of long wavelength waves. For this aim we expand
the quantities into the serie8:
and due to the spatial inhomogeneity introduce the following coordinate stretching:
Substitution of the expansion (6.1.35) and (6.1.36) into Eqs. (2.1-2.5) leads at ε 2
to the inhomogeneous Burgers equation:
and the ponderomotive term due to the Alfvén wave has been added phenomeno-
logically to the right hand side: be sin e t. Note that the pseudo-damping term
1 cA)
/2(ln x u is a consequence of the spatial inhomogeneity taken into consideration.
Equation (6.1.40) shows that if viscosity is dropped, the linear terms on
the left hand side give u 2 cΑ = const. implying that wave energy flux density is
conserved. The term proportional to u is however a pseudo-damping one because
the energy f' u 2 dx is not conserved if CA is a function of the coordinate x,
CA = cΑ(x). The other way of proving it is to get a linear dispersion relation from
(6.1.40). After Fourier analysis of the left hand side of it, we get
So, we see that both utt and u terms are responsible for damping of the wave. The
second derivative term is responsible for viscous damping because α is dependent
346 K. Murawski
on the viscosity coefficients η 0 and η1. The second damping term is connected with
the inhomogeneity and thus is called pseudo-damping term.
For the homogeneous Burgers equation, both the viscous α and the nonlinear
β coefficients are positive. So, there are only shock solutions for which u < 0,
where the phasor ξ = t - x.
6.1.5. Numerical results
The homogeneous Burgers equation possesses an expansion (in ξ) shock wave
solution written as
Note a similarity of these expressions to the corresponding ones for the long Alfvén
waves. Now, however, we have got a phase velocity which is usually different from
VA. An expression for the group velocity λ is found from the equations at ε 2
Finally, from ε 3 , we obtain (for the Alfvén wave) complex coefficients nonlinear
Schrödinger equations:
where
The ponderomotive terms are derived by a method of averaging over the fast
variables (x, z, t) in the Alfvén wave. Note that the r. h. s. of Eq. (6.2.9) disappears
for the homogeneous field. Thus, they components of the magnetosonic waves are
driven only in the case of the inhomogeneous field. Additionally, we must include
equations for b 2 and p2:
7. Conclusions
Short and long Alfvén waves propagate according to Burgers-type and com-
, plex-coefficients-nonlinear Schrödinger equations, respectively. For the derivation
of the complex coefficient nonlinear Schrodinger equation see also [24]. Its solu-
tions have been discussed recently by Stenflo [25] and Stenflo et al. [26]. The long
Alfvén wave is more weakly damped due to the phase mixing effect (which is de-
scribed by the —α(x)u xx term) than the short wave. The Alfvén wave drives the
slow and fast magnetoacoustic waves because of gradients in the z and x direction.
These waves are damped both by the viscosity and the phase mixing effect. Thus,
they cascade their energies into lower scales. The x component of the fast wave
is additionally damped due to inhomogeneity (-c2 Axu2x ) cascading its energy to
lower scales. The y component of the fast mode is driven only by the short Alfven
wave because of inhomogeneity. The short Alfvén wave is not coupled with the
pressure p2 as the long Alfvén wave is.
The numerical calculations performed for the fast wave propagating perpen-
dicular to the ambient inhomogeneous magnetic field have shown that due to the
350 K. Murawski
inhomogeneity the shock wave is attenuated and increases its velocity. The peri-
odic ponderomotive force has also been taken into account to show that waves go
into lower and lower scales. This mechanism gives us a little more insight into the
process of damping of the fast mode due to the inhomogeneous magnetic field and
probably into a process of corona heating.
One wouhd also have thought that an interesting problem would be to show
how far the waves must propagate to create shocks. This would be a calculation
relevant to coronal heating. Work in .this direction is in a progress and will be
published elsewhere.
Acknowledgements
The author wishes to express his sincere thanks to the Solar Theory Group
at St Andrews, especially Dr Bernard Roberts and Prof. Eric Priest, for their
encouragement and help, to Dr Luigi Nocera for correspondence, and to the SERC
for financial support.
References
[1] Β. Roberts, in: Proc 4th European Meeting on Solar Physics. The Hydromagnetics
of the Sun, Noordwijkerhout, ESA SP-220, The Netherlands, 1984, p. 137.
[2] G. Mann, J. Plasma Phys. 40, 281 (1988).
[3] B.K. Som, B. Dasgupta, V.L. Patel, M.R. Gupta, Phys. Fluids B 1, 2340 (1989).
[4] J.V. Hollweg, S. Jackson, D. Galloway, Solar Phys. 75, 35 (1982).
[5] J.-I. Sakai, B.U. Ö Sonnerup, J. Geophys. Res. 88, 9069 (1983).
[6] W. Sahyouni, I. Zhelyazkov, P. Nenovski, Sol. Phys. 115, 17 (1988).
[7] P.K. Shukla, G. Feix, L. Stenflo, Astrophys. Space Sci. 147, 383 (1988).
[8] J.-I. Sakai, H. Washimi, Astrophys. J. 258, 823 (1982).
[9] S.V. Vladimirov, L. Stenflo, M.Y. Yu, Phys. Lett. 153, 144 (1991).
[10] K. Murawski, E. Infeld, J. Ziemkiewicz, Wave Motion 13, 337 (1991).
[ii] D.G. Wentzel, Sol. Phys. 39, 129 (1974).
[12] E. Priest, Solar Magnetohydrodynamics, Reidel, Dordrecht, 1982.
[13] M. Ruderman, 5ol. Phys. 131 , 11 (1991).
[14] L. Nocera, E.R. Priest, Geophys. Astrophys. Fluid Dyn. 37, 193 (1986).
[15] L. Noces, E.R. Priest, J.V. Hollweg, Geophys. Astrophys. Fluid Dyn. 35, 111
(1986).
[16]M. Ruderman, Fluid Dyn. 22, 299 (1987).
[17] E. Mjolhus, J. Wyller, Phys. Scr. 33, 442 (1986).
[18] K. Murawski, Z. Natforsch. A 41, 1186 (1986).
[19] Α. Jeffrey, T. Kawahara, Asymptotic Methods in Nonlinear Wave Theory, Pitman,
London 1982.
[20] K. Murawski, Aust. f. Phys. 40, 593 (1987).
[21] T. Taniuti, C.C. Wei, f. Phys. Soc. Jpn. 24, 941 (1968).
Alfven-Magneosonic Waves Interaction 351
[22] J.P. Boris, D.L. Book, J. Compt. Phys. 20, 397 (1976).
[23] D. Schnack, J. Killeen, J. Comput. Phys. 35, 110 (1980).
[24] Ν. Asano, J. Phys. Soc. Jpn 36, 861 (1974).
[25]L. Stenflo, J. Phys. Α, Math. Gen. 21, L499 (1988).
[26] L. Stenflo, M.Y. Yu, Ρ.K. Shukla, Phys. Scr. 40, 257 (1989).