Académique Documents
Professionnel Documents
Culture Documents
E-mail: nariyuki@edu.u-toyama.ac.jp
Received January 2, 2016; Accepted March 7, 2016; Published April 27, 2016
...............................................................................
...............................................................................
Subject Index
1.
Introduction
It is well known that magnetohydrodynamic (MHD) turbulence in a solar wind plasma often shows
a clear correlation between magnetic field and plasma bulk velocity (Alfvenic correlation) [13].
Although phenomenology using the incompressible MHD system successfully reproduces powerlaw indexes of anisotropic MHD turbulence [46], observational studies indicate that the condition
of zero cross-helicity and zero residual energy, which is the precondition of the phenomenology, is
usually not valid in the solar wind [3,7,8]. This obviously corresponds to the existence of fluctuations
with Alfvenic fluctuations, which are composed of unidirectional MHD waves. On the other hand,
it is also well known that the cross-helicity decreases with increasing heliocentric distance [3,9].
Thus, the Alfvenic fluctuations are damped through some processes in the interplanetary space and
MHD turbulence approaches the fully developed state, which is assumed in the phenomenology. This
scenario is also supported by observational results of the radial dependence of the anisotropy of the
correlation length [10,11].
Since MHD waves in the solar wind have finite amplitude, damping processes through wave
wave coupling (parametric instabilities [1217]) can be relevant to the decrease of the cross-helicity.
Nonlinear evolution of transverse MHD waves (quasi-parallel propagating fast and intermediate
modes) has been discussed by a lot of authors using theoretical models [13,14,1824] and direct
numerical simulations [2535]. It is noteworthy that although computer simulations play significant
roles in the investigation of such nonlinear phenomena nowadays, classical theoretical models are still
valuable for the understanding of phenomena and concept formation from numerical/observational
The Author(s) 2016. Published by Oxford University Press on behalf of the Physical Society of Japan.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/),
which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
A nonlinear evolution equation of envelope-modulated Alfven waves including both ion kinetic
effects and ion acoustic waves (KTDNLS: kinetically modified triple-degenerated derivative
nonlinear Schrdinger equation) is discussed. The consistency with the kinetic model of the ponderomotive density fluctuation (KDNLS: kinetically modified derivative nonlinear Schrdinger
equation) is set as the constraint to the resultant model. To close the system, modeling of the
second-order specific heat ratio is demonstrated. The transverse stability analysis of Alfven
waves using the KTDNLS is carried out. The resultant nonlinear evolution equation is similar to the one obtained from the KDNLS, which shows that the ion kinetic effects enhance the
transverse instabilities.
Y. Nariyuki
2.
In a similar way to previous studies [42,44], we start from the Hall-MHD equation:
= (u),
t
u
|b|2
= (uu) p +
+ (b )b,
t
2
b
1
= (u b)
(b) b ,
t
0
(2)
b = 0,
(4)
(1)
(3)
where the magnetic field b is normalized to the ambient magnetic field B0 , the mass density is normalized to the background mass density 0 , time t is normalized to the proton gyrofrequency p ,
(5)
u = u0 + u1 + 2 u2 + ,
(6)
b = b0 + b1 + 2 b2 + ,
(7)
p = p0 + p1 + 2 p2 + ,
(8)
2/12
results. The derivative nonlinear Schrdinger equation (DNLS) is known as the simplest theoretical
model for unidirectional nonlinear transverse MHD waves [14,15], in which wavewave coupling
in envelope-modulated waves is described. Since the envelope-modulated waves accompany the
longitudinal component, the waves can resonate with ions through Landau resonance. The incorporation of the effects of the ion Landau resonance (KDNLS: the derivation of the kinetically modified
DNLS) [12,3641] has also been carried out. On the other hand, the longitudinal component of
MHD waves (slow modes) is excluded in ordering of the DNLS. To incorporate slow modes into the
DNLS, a nonlinear evolution equation with different ordering (TDNLS: triple-degenerate DNLS)
was derived by Hada [42]. From a similar point of view, a novel model that has the nonlinear Hall
term was discussed by Emoto and Yoshida [43]. Nariyuki and Hada discussed the kinetically modified TDNLS (KTDNLS) to be consistent with KDNLS with a simple Landau fluid closure derived
by Medvedev and Diamond [40]. Although the simple KDNLS model [19,20,23] can qualitatively
reproduce important characteristics of nonlinear Alfven waves, the dependence of coefficients on the
ion beta ratio and ion-to-electron temperature ratio is not consistent with that obtained by the fullVlasov theory [37,38]. The purpose of this paper is to construct the KTDNLS that agrees with the
KDNLS derived from the full-Vlasov theory through rescaling. We here use the consistency between
KDNLS and KTDNLS as a constraint that the resultant KTDNLS should satisfy, as discussed in
Sect. 2. In Sect. 3, the specific heat ratio that is consistent with the KTDNLS is phenomenologically discussed. In Sect. 4, the transverse stability of Alfven waves, which is relevant to the radial
dependence of the spectral anisotropy of the solar wind MHD turbulence, is carried out by using the
wave-packet limit. A summary of the results and a description of future issues are presented in Sect. 5.
Y. Nariyuki
(9)
= 2 y,
(10)
= 2 z,
(11)
= 2 t,
(12)
where b0 = (1, 0, 0), u0 = (0, 0, 0). We here introduce the complex transverse magnetic field
b = b y + ibz and bulk velocity u = u y + iu z . From the equations at the order of 2 and 3 , we
find [42]
b1 1 b1
2 b1
+
1 + i 2 = 0,
(13)
1 (Te 1 + pi1 ) = 0,
(14)
|b1 |2
pi2 + Te 2 2 + 12 +
+ bx2
2
= 0,
(15)
at the order of 3 .
From Eqs. (14) and (15), we can find the evolution equation for the longitudinal waves [42]:
1
+
12
|b1 |2
+ bx2 = 0,
+ 1 +
2
(16)
where
bx2 b y1 bz1
+
+
= 0,
(17)
= 1 + ( 1)(1 Te )/2,
(18)
= f 1 = ( Ti + Te ) 1,
(19)
where Ti = pi /, respectively. The equation set (13), (16), (18) is the TDNLS first derived
by Hada [42]. If the waves are strictly 1D, bx2 = 0. By rescaling b b/ 1/2 and O(1),
/ 1/2 , and / 1/2 , we find the static approximation [45]:
1 =
|b1 |2 + 2bx1
.
2 1 f
(20)
Let us here turn to the kinetic modification of the discussion above. By using linear solutions of
nonlinear Landau damping to close the equation set, we have the kinetically modified version of the
3/12
where = (/) + i (/ ).
In previous studies [23,42], the total pressure p obeys the adiabatic relation pi = const.,
where is the real constant. We here consider the isothermal electrons and the adiabatic ions
( pi = const.). Then we find
Y. Nariyuki
(21)
where H () is the Hilbert transformation. The derivation of coefficients N1 and N2 using a simple
model is discussed in the next section. From Eqs. (13) and (21), we find the KDNLS [37,38]:
b1
2 b1
+
(22)
(N1 + N2 H ) |b1 |2 b1 (N1 + N2 H ) |b1 |2 + i 2 = 0.
12
N1
1
r ,
2
2 N1 + N22
(24)
1
N2
i ,
2
2 N1 + N22
(25)
|b1 |2
+ bx2 = 0,
+ (r + i H )1 +
2
(26)
at the order of 2 . Equation (21) is recovered from Eq. (26) in a similar way to the derivation of
Eq. (20) from Eq. (15). When i = 0, Eq. (26) recovers the same form as Eq. (16).
Equations (24) and (25) are the conditions that the resultant KTDNLS should satisfy, while the
definition of the specific heat ratio is not given by these. Actually, for instance, ion acoustic waves
in the kinetic system have a number of branches. This indicates that the specific heat ratio cannot be
unified in the kinetic system, while Eqs. (18) and (19) explicitly include . In the next section, we
discuss the procedure to construct the specific heat ratio of the phenomenological model.
3.
The coefficients of Eq. (21) (N1 and N2 ) can be determined by using the linear solutions of Landau
and transit time damping [3739,41,46]. Nariyuki et al. [46] discussed a simple model that uses the
longitudinal velocity distribution function:
g(t, x, vx ) = f t, x, vx , v y , vz dv y dvz ,
(27)
where f (t, x, vx , v y , vz ) is the velocity distribution function (VDF) of ions. When
g = g0 + g1 + 2 g2 + ,
1 = g1 dvx .
4/12
(28)
We here consider the kinetically modified TDNLS (KTDNLS) that can recover the KDNLS (22).
Nariyuki and Hada [23] derived a 1D-KTDNLS using a simple Landau-fluid model. Their model
can reproduce the KDNLS of Medvedev and Diamond [40] through the rescaling discussed above.
We here take a direction similar to Nariyuki and Hada [23]; the KTDNLS is derived to be consistent
with the KDNLS.
From Eq. (21), we have
N1
N2
2
2
H 1 ,
(23)
|b1 | + 2bx1 =
N12 + N22
N1 + N22
Y. Nariyuki
(29)
where
vx g0
p b + p e ,
(30)
Ti (vx 1)
vx g0
g0
G TD =
p b
p b ,
(31)
Ti (vx 1)
vx 1
g0
G bx =
bx1 ,
(32)
vx 1
2 /v , v 2 = 2T , and h indicates the Fourier transwhere pb = |b1 |2 /2, g0 (vx ) = exp vx2 /vth
th th
i
formed value of h. G LD , G TD , and G bx correspond to the contributions of the nonlinear Landau
damping, the nonlinear transit time damping of envelope-modulated Alfven waves, and linear transit
time damping of obliquely propagating waves, respectively. Equation (29) is obtained by adding the
phenomenological term G TD to the linear solution of the previous study [46]. By using Eq. (29),
G LD =
1 C (1 Te A) Te B D
,
2 (1 Te A)2 + B 2 Te2
1 (Te BC + D (1 Te A))
,
2 (1 Te A)2 + B 2 Te2
2
ZR
A = 2 1+
,
vth
vth
N2 =
(33)
(34)
(35)
12
2 e vth
B=
,
3
vth
(36)
ZR
,
vth
(37)
v12
e th
D=
,
vth
(38)
C =
1
where Z 0 vth
is the plasma dispersion function [47], Z R = Re[Z 0 ], and Eqs. (33) and (34) agree
with N x1 and N x2 , respectively [46]. N1 and N2 in Eqs. (33) and (34) also agree with those in the
full-Vlasov theory [38] with the massless electrons. The advantage of the formulation (29) is to make
it easy to discuss the macro variables such as the density and the pressure, since the integration of
the perpendicular direction has already been carried out.
When we consider the same ordering as TDNLS, |b1 |2 |b1 |2 and bx1 0. Thus,
vx g0
g 1 =
p b + p e .
(39)
Ti (vx 1)
In a similar way to previous studies [28,29,46], we find
where p1 =
p 1 = i Ti 1 ,
Te 2
i =
Z0 1 ,
Ti
1
vx2 g1 dvx , Z 0 = 2 (1 + Z 0 ), and = vth
.
5/12
(40)
(41)
N1 =
Y. Nariyuki
(42)
(43)
i I =
(44)
(45)
which corresponds to the kinetic dispersion relation of ion sound waves [28,29,46,48]. From Eqs. (41)
and (45), we have the constant specific ratio
1 Te
.
Ti
(46)
p2 = i R Ti 2 + 12 + r 1 + i I H (2 )
(48)
where is the parameter that describes the deviation from the fluid description, and = (i R 1)
(1 Te )/2. Let us briefly discuss the physical meaning of . If we consider the expansion Re[i ] =
i R + i R1 + 2 i R2 + , we find i R1 Ti = (1 Te i R Ti ). Namely, we here close the system by using the ansatz of the second-order specific heat ratio, in which the nonadiabatic effects are
incorporated. Then we have another condition:
r = (1 + )(i R Ti + Te 1) =
We here set > 1.
6/12
N1
1
.
2 N12 + N22
(49)
i =
Y. Nariyuki
(a)
(b)
Figure 1 shows the dependence of (a) N1 /2 N12 + N22 (black line) and i R on , and
(b) N2 /2 N12 + N22 (black line) and i I on with Tr = Ti /Te = 0.8. The gray lines indicate i R
and i I , defined by Eqs. (43) and (44). In this case, the black and gray lines have no intersection
(condition (47)). The gray dots of Fig. 2 show one of the intersections (the lowest value of ) between
N2 / N12 + N22 and i I (44). This shows that the intersection appears with < 0.72. The smallest
value of ( 0.48) indicates that becomes close to 1 (Fig. 2(b)). For comparison, let us discuss
the modified version:
i R =
2Te
Te 2
Z R,
2 + 1
Ti
Ti
(50)
i I =
2Te v2
e th .
Ti
(51)
The dashed lines of Figs. 1 and 2 show the cases using Eqs. (50) and (51). As seen in Fig. 2, the
resultant relations (dashed lines) are different from those using Eqs. (43) and (44).
While these models are artificial, the result suggests that, if we can use the empirical relation
between Tr and obtained from the real solar wind plasma or the kinetic simulation data, the specific
heat ratio can be modeled by using N1 and N2 . Namely, from the lines as shown in Fig. 2(a), which
are directly obtained from observations or numerical simulations, we can have empirical formulas.
For instance, if we consider functional forms such as
i R =
2Te a
Te 2
Z R,
2 + 1
Ti
Ti
7/12
(52)
Fig. 1. The dependence of (a) N1 /2 N12 + N22 (black line),
1 Te i R Ti using (43) (gray line), and
1 Te i R Ti using (50) (dashed line); (b) N2 /2 N12 + N22 (black line), i I Ti using (44) (gray line), and
i I Ti using (51) (dashed line) on with Tr = Ti /Te = 0.8.
Y. Nariyuki
(a)
(b)
2Te b v2
e th ,
Ti
1
i I =
(53)
where a and b are real constants, we can use a, b, and as fitting parameters. Although Eqs. (52)
and (53) are not justified by kinetic theories, the procedure above can give the kinetic specific heat
ratio of a plasma of the Alfvenic state.
4.
Finally we discuss the transverse stability of the multidimensional KTDNLS of Eqs. (13), (17),
and (26). The transverse instabilities of parallel propagating Alfven waves such as filamentational
instabilities have been discussed by several authors [5054]. We here discuss the transverse instability
by using the wave-packet limit in a similar way to previous studies [50,54].
For simplicity, let us replace the variables of the KTDNLS of Eqs. (13), (18), (26) as follows:
b b
2b
+
+ i 2 = 0,
t
x
x
b y
bz
bx
+
+
= 0,
x
y
z
|b|2
+
2 + (r + i H ) +
+ bx = 0.
t
x
2
In the following discussion, we use the 2D coordinate (bz /z = 0).
8/12
(54)
(55)
(56)
Fig. 2. (a) The intersections between N2 /2 N12 + N22 and i I Ti using (44) (gray dots) and i I Ti using (51)
(black circles); (b) , which satisfies the condition (49), with (43) (gray dots) and with (50).
Y. Nariyuki
(57)
= y,
(58)
= 2 t,
(59)
(1)
(2)
(2)
(3)
(3)
b = 1 ei + 2 2 e2i + 0 + 3 1 ei + 3 ei3 + ,
(2)
(2)
(2)
(3)
(3)
bx = 2 A1 ei + A1 ei + A0 + 3 A2 e2i + A2 e2i + ,
(2)
(2)
(2)
(3)
(3)
= 2 1 ei + 1 ei + 0 + 3 2 e2i + 2 e2i + ,
(60)
(61)
(62)
where = 0 t k0 x. By taking the contribution of the terms with eim in Eq. (54), we find [50]
(63)
vg = 2k0
(64)
1
(1)
(2)
2 1
(2) (1)
ik0 0 1 1 + i
=0
(65)
(m)
isgn(mk0 )eim . Then, from Eq. (56), we find
(2)
(2k0 r ) 0 =
(2)
1
(2) 2
1
2
(2)
+ A0 ,
(66)
(k0 r )2 + i2
(1)
i 1
2k0
(1)
(r + ii ) i
1
,
(67)
i 1
,
=
2k0
which is obtained from the terms with (, m) = (2, 1) in Eq. (55), is used.
On the other hand, from Eqs. (55) and (66), we find
(1)
(2)
2 A0
2 1 |2
1
(2)
2k0
+ A0 .
=
2
2
(2k0 r ) 2
(2)
A1
(68)
(69)
= cos ,
= sin ,
9/12
(70)
(71)
0 = k02
Y. Nariyuki
1
where
(1)
2 1
(1) 2 (1)
i 1 1 + (i + )
= 0,
2
k0
sin
1 +
,
=
2 (2k0 r )
2 (r 2k0 ) 2k0 (r 2k0 ) cos2 + sin2
(72)
(73)
= cos2 + r sin2 , and = i sin2 . Notice that Passot and Sulem [54] also derived a
nonlinear Schrdinger equation from KDNLS. As shown by Passot and Sulem [54], through the
(1)
stability analysis to the parent wave 1 = 0 ei0 + + ei+ + ei , where 0 = |0 |2 t,
= t K , = 0 , we find
= i K 2 2 K 4 + 202 K 2 .
(74)
5.
In the present study, the KTDNLS that corresponds to the KDNLS is discussed. By using the consistency between the KDNLS and the KTDNLS as a constraint, a scheme to calculate the second-order
specific heat ratio is obtained. Although the present models of the specific heat ratio are ad hoc, the
procedure presented can be utilized with the data of the spacecraft observation and kinetic simulations; from the empirical relation between ion and electron temperatures, we can find a functional
form of the specific heat ratio and calculate . Of course, the systematic derivation of the specific
heat ratio from kinetic models is still an open question. The important point is that the secondorder correction of the specific heat ratio () may include effects of the finite amplitude of Alfven
waves Actually, the dispersion relation can be modified by the presence of the finite-amplitude Alfven
waves [55].
When we consider one-fluid modeling of space plasmas, which plays an important role in describing global-scale phenomena as a phenomenology, such an empirical relation of the thermodynamical
parameters is useful. Some previous studies discussed the reproduction of the linear Vlasov theory
from a fluid system with a specific heat ratio [5658]. For global-scale phenomena, it is better to truncate the terms including the Hilbert transformation into the usual dissipation term in a similar way to
Sect. 4 and the previous study [54]. As is well known through observational studies [59], solar wind
plasmas are composed of multicomponents of ions, which significantly affects the characteristics of
the solar wind MHD turbulence [8]. These effects are incorporated into the equation of state in onefluid modeling [60]. The pseudo-temperature anisotropy [61] can also appear in one-fluid modeling
[60] as a consequence of the integration of the fluctuations, the scale of which is smaller than the
coarse-grained scale of one-fluid modeling. These suggest that the thermodynamical characteristics
of each order in the expansion as shown in Sect. 2 can be different.
As discussed in Sect. 4, the transverse instabilities can be enhanced by the ion kinetic effect. This
agrees with the previous study [54] and the longitudinal instabilities [23,62,63]. The phase between
the compressional component of the magnetic field and the density is also shifted by the ion kinetic
effect. The transverse instability can play an important role in causing the anisotropy of the magnetic
10/12
Since < 0 (i > 0 and i < 0), the imaginary part of (growth rate) is enhanced by the ion
kinetic effects. Equations (67) and (68) show that the phase between bx and is modified by the ion
kinetic effects, as predicted by the static approximation in the KDNLS (21).
Y. Nariyuki
spectra (correlation length) in the solar wind; observational studies [10,11] suggest that quasi-parallel
propagating Alfvenic fluctuations decrease with increasing heliocentric distance, as mentioned in
Sect. 1. Although the parameter regime in which the linear growth of the transverse instabilities is
larger than that of the longitudinal instabilities is limited [51,52], the existence of seed fluctuations and/or a parent wave with broadband spectra can nonlinearly drive the instabilities, whose
linear growth rate is not larger than the others [30,35]. Direct numerical simulation of the resultant
KTDNLS to analyze the energy transfer among the wave modes is an important future issue.
Acknowledgements
This work was supported by MEXT/JSPS under a Grant-in-Aid for Young Scientists (B) No. 15K17770.
The dispersion relation of the full-Vlasov theory was numerically solved by using the EMDISP code, developed
by Prof. T. Hada and the JST astro/space simulation group.
References
11/12
Y. Nariyuki
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
12/12