A&A 420, 1107-1115 (2004)
DOI: 10.1051/0004-6361:20040119
A. N. Kryshtal - S. V. Gerasimenko
Department of Cosmic Plasma Physics, Main Astronomical Observatory of the National Academy of Sciences of Ukraine, 03680 Zabolotnogo Str., 27, Kiev 127, Ukraine
Received 12 February 2003 / Accepted 16 February 2004
Abstract
We investigate the stability to the development of
plasma waves in the preflare situation of a loop structure at the
chromospheric part of a current circuit of a loop. We investigate
the conditions under which low-frequency plasma instabilities can
develop, assuming the absence of beam instabilities.The
large-scale quasi-static electric field in the loop circuit is
assumed to be "subdreicer" and weak. Thus the percentage of
"runaway" electrons is very small and their influence on the
process of instability development is negligible. The pair Coulomb
collisions are described by a BGK-model integral. We consider the
situation when the plasma at the surface layer of a loop has a
spatial gradient of density. In accordance with
Heyvaerts-Priest-Rust theory, such a preflare situation would
typically exist when the amplitude of the weak electric field
in the circuit of an "old"
loop in an active region begins to increase when "new" magnetic
flux emerges from under the photosphere. We have found that two
types of waves are generated in such a plasma due to the growth
of instabilities: the "kinetic Alfven-like" waves and new type of
waves, in the range of magnetoacoustic ones. The instability of
these latter waves has a clear threshold and it can be considered
as an "indicator" of the development of a preflare situation in an
active region.
Key words: plasmas - Sun: flares - Sun: chromosphere
Plasma instabilities are traditionally assumed to be one
of the main sources of waves in the flare atmosphere (Priest 1982;
Somov 1994; Zaitsev et al. 1994). Their hierarchy is usually
considered as an important element in the theory of cyclotron
maser emission (CME) (Melrose 1989; Zaitsev et al. 1994; Mel'nikov
et al. 2002) as well as in the dynamics of current sheets which
form in flaring arcades (Podgorny & Podgorny 2001), namely where
flares most frequently occur (De Jager 1959; Priest 1982). Thus
the loops of arcades are "post-flare" loops of a previous flare
and the preflare of the next one. The investigation of the
preflare plasma state on the basis of Heyvaerts-Priest-Rust (HPR)
theory (Heyvaerts et al. 1977; Podgorny & Podgorny 2001) is
current practice. In this theory a flare is considered to be the
result of the interaction of a "new" magnetic flux, which emerges
from under the photosphere and an "old" one, which passes through
a current circuit of a loop in an arcade. Such interaction results
in an adiabatic slow increase of the amplitude of a weak
quasi-static electric ("DC") field in the circuit of the "old"
loop. From general physical considerations (De Jager 1959;
Kadomtsev & Pogutse 1967) it is clear that, in such circumstances,
different low-energetic wave instabilities of the "stream" type
will grow in plasma. If the growth time of the large-scale
electric field amplitude
is much larger than the time of instability development, - this
situation occurs most frequently (Poletto & Kopp 1986), - then
the mechanism of "direct initiation" of instability by the
electric field will take place. The corresponding instabilities
will have a clearly expressed threshold character (Kryshtal &
Kucherenko 1995). The value of the amplitude
,
expressed in units of the local Dreicer field
,
will be the threshold value in this case. We
assume that in preflare situation there are only very few
"super-energetic" charged particles (i.e. plasma "without beams").
When the electric field is weak, i.e. when
In connection with the "perpetual" problem of solar physics, the
problem of coronal heating, Aschwanden (2001) has formulated
interesting conclusions based on the recent soft X-ray and EUV data from space observations with the YOHKOH, SOHO and TRACE satellites. We consider that the inclusion of the chromosphere is
very important not only "in conventional AC and DC models"
(Aschwanden 2001) but in the investigation of the dynamics of the
preflare plasma state in a loop in general. Preferential footpoint
heating (Aschwanden 2001) can be explained from our point of view
in the framework of the well-known Ionson model (Ionson 1978). The
most important necessary condition is the generation of the
kinetic Alfven waves (KAW) on the "chromospheric floor" of the
loop current circuit. These waves propagate almost perpendicular
to the direction of the magnetic field
of the
loop in its surface layer (Ionson 1978). Second, the observed
overdensity of order
(Aschwanden 2001),
which, of course, is not less in the chromosphere, seems to be a
good reason to assume that strong gradients of the plasma density
can have a larger effect than the similar values of temperature and
magnetic field amplitude
.
Here
we denote by
the inverse of the gradient scale
,
is the reduced spatial gradient of
density. The value of
is proportional to the inverse
value of the mean scale of the density inhomogeneity of the
electrons and ions (
). Analogous values for the
temperature and the magnetic field can be introduced in a similar
way. Since the end of the 1950s (De Jager 1959), it is well known
that overdensity is the most distinguishing property of post-flare
loops. When the effect of the reduced spatial gradient of density
dominates, it is possible to neglect the shear influence on the
instability development. This needs some special condition
regarding the plasma state and the characteristics of the
perturbation. The increasing number of detections of flarelike
events at temperatures of
MK in EUV (with
SOHO/ETT and TRACE) (Aschwanden 2001) allows us to assume that
the early stage of the preflare process can correspond to a
temperature of
0, 5 MK. The plasma parameters should be
obtained from the well-known semiempirical model of chromospheric
flare regions (Machado et al. 1980).
We investigate the physical conditions for the growth of
low-frequency plasma wave instabilities in the case of the
long-wave perturbations propagating almost perpendicular to the
direction of the loop magnetic field
.
We use a
local rectangular Cartesian coordinate system with the Z-axis
directed along the field vectors
and
(we assume that
). Taking into account that we consider a plasma in the "chromospheric floor" near the footpoint of a loop, this means that the XY-plane is actually parallel to
the surface of the photosphere. For intermediate calculations we
have used a cylindrical coordinate system in velocity space
(
)
(Alexandrov et al. 1988).
We assume that the plasma has a one-dimensional spatial gradient
of its density along the X-axis. The thickness of the surface
layer of the loop is the mean spatial scale of inhomogeneity of
the plasma density. For the local solutions of the dispersion
relation for the low-frequency plasma waves (Michailovsky 1963) we
consider that the origin of the Cartesian coordinates is placed
near the inner border of the surface layer. We assume
(Michailovsky 1963), that the wave frequency
satisfies the inequality
In the present investigation, we use the same mechanism of "direct initiation" of instability and the same plasma model as in our previous work (Kryshtal & Kucherenko 1995; Kryshtal 2000), the most important properties of which are the following:
On the other hand there is observational evidence for a
correlation between the time of maximum E0 (t) and the
moment at which the first energy release occurs in a flare
(Poletto & Kopp 1986; Zaitsev et al. 1994). The left-hand side
of inequality (3) approximately equals the value
,
where
is the time for the amplitude E0 (t) to grow to its maximum value.
We assume that the electric field under consideration is extremely
weak and that the threshold value
,
at which the instability, with growth rate
and growth time
,
starts developing, does not much exceed the equilibrium value
.
This value corresponds to the
origin of the preflare process when interaction between the "old"
and "new" magnetic fluxes is absent. In this situation
equals the time of the short-period prediction of a flare and the instability appearance can be considered in a sense as a
"forerunner" of a flare process.
For such magnetic field amplitudes and with the parameters given in Eq. (26) and extremely small values of
,
the conditions (24) and (25) can be easily satisfied in the case of a weak inhomogeneity. In the opposite case of a strong inhomogeneity, for
extremely small values of
,
when instability can develop on the background of turbulence and the value of
can sharply increase, the conditions (24) and (25) practically determine an upper limit for
.
In the case of weak inhomogeneity, when the ion-electron collisions dominate, we can estimate for the parameters (26) the equilibrium value
,
which corresponds to the steady state in the loop circuit (without interaction of the "new" and "old" magnetic fluxes). If we assume that the loop under consideration is a "semitorus" with small radius R0, it can be shown that
The starting point for obtaining the dispersion relation (DR) for this plasma model is the expression for the scalar dielectric permeability of a hot magnetoactive plasma with spatial inhomogeneity. (Alexandrov et al. 1988).
When the electric field in the low-frequency wave
in the inhomogeneous medium can be considered as quasi-potential,
the equation for this permeability in the inhomogeneous plasma,
Equation (33) can be solved by the standard Euler method, but for extremely small values of
(actually as
)
it becomes impossible to
obtain the solution in an analytical form. So, we had to make use
of the numerical calculations based on exact formulae (Mishina & Proskuryakov 1962). We have used the roots of the resolvent equation. For all roots to be real and positive, and thus to
obtain all four roots of the MDR (33) as real, the condition
The obtained real roots of the MDR (33) have to be substituted into formula (32) to analyse the expressions for the growth rates. Only if
For the set of parameters (26) and
we have
The calculations show that four roots of the MDR (33) can
be formally split into two pairs. The roots, which we call
"
-" and "
-waves", have in the ranges
(61)
(64) very close (but not the same)
values of
and
,
but opposite signs (
is positive
and
is negative). The
-wave can be
easily interpreted as the "right" (i.e. with kz > 0) kinetic
Alfven wave, modified by the pair collisions and large-scale
subdreicer electric field. Unfortunately, we cannot interpret the
-wave in the same way as the inverse kinetic
Alfven wave. In the framework of the present model we have used in calculations
only
and
as variable parameters to
obtain the maximum effect. We do not know the details of the
polarization of perturbation, so, we cannot say whether the
negative values of
are the result of
incompleteness of our description of perturbation, or whether they
demonstrate the well-known "parasitic" effect, when negative
frequencies appear in DR (Michailovsky 1963; Alexandrov et al.
1988). For the same reason in the other pair of roots we consider
only the
-wave (with positive values) and suppose
that only this root corresponds to a "real" wave. It is hard
enough to determine exactly the dispersion law for this wave. We
can only approximately describe it as a wave from the range of the
slow magneto-acoustic ones, because in our model the dispersion
law for these SMA-waves has the approximate form
.
This wave is definitely modified by the weak
drift motions, pair Coulomb collisions and subdreicer electric
field. It practically does not depend on
(more exactly,
this dependence is very weak) and depends on
in an unusual way in comparison to KAW. This last fact
does not permit us to consider this
-wave as the
"exact" SMA-wave, even taking into account the corrections to the
dispersion law due to the collisions, subdreicer field and drift
motions. Here we meet an interesting phenomenon: at
,
t = 1 and
("Michailovsky's"-case, 1963), without the collisions and electric field, the DR (33) becomes a polynomial of third order. The roots of
this DR are the two Alfven (in our case - the kinetic Alfven)
waves and slow drift-Alfven wave (which contains in its dispersion
law the drift frequency (41)
(42) as the
factor, and because of this is very slow) (Michailovsky 1963).
When
,
t > 1 and
,
the KAWs are, of course, modified, but remain KAWs. At the same
time the slow drift Alfven wave vanishes, and instead of this the
-waves appear, which contain the drift frequency (41)
(42) in their dispersion laws not as the small factor, but as the small correction. Figures 1 and 2
show the behaviour of the functions
and
at
![]() |
Figure 1:
The KAW-like root of MDR
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 2:
The SMAW-like root of MDR
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 3:
The growth rate of KAW-like instability
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 4:
The growth rate of SMAW-like instability
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 5:
The inverse KAW-like root of MDR
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 6:
The inverse SMAW-like root of MDR
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 7:
The growth rate of the inverse KAW-like instability
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 8:
The growth rate of inverse SMAW-like instability
![]() ![]() ![]() |
Open with DEXTER |
An interesting situation occurs for kz < 0. The
- and
-waves become the physical meaningful ones, and
-waves in a sense lose their physical meaning. Figures 5 and 6 show the behaviour of the inverse KAW-like
-wave and
the inverse SMAW-like
-wave.Their growth rates
and
are shown in Figs. 7 and 8. In this case the
-wave plays the role of a forerunner
of a flare. But it has one very important defect: at the same value of
it has too high value of
.
In the framework of our model this is important.
The role of magneto-acoustic waves in the flare models (Priest 1982; Somov 1994; Zaitsev et al. 1994; Zaitsev & Stepanov 1982), especially in the laboratory modelling of this phenomenon (Somov 1994; Zaitsev et al. 1994), is well-known. There is a considerable number of investigations (Rosenraukh & Stepanov 1988; Terekhov et al. 2002; Zaitsev & Stepanov 1989) of the pulsations of the flare emission as well as investigations of the transverse waves in the loop structures (Podgorny & Podgorny 2001; Mel'nikov et al. 2002; Schrijver et al. 2002). From our point of view, the most interesting peculiarities of the present plasma model and of the results that we have obtained are the following:
Acknowledgements
The authors thank Dr. K. V. Alikajeva and Prof. S. I. Gopasyuk for useful discussions of the present work.