FORMATION OF CORONAL MHD SHOCK WAVES

II. The Pressure Pulse Mechanism
B. VRŠNAK and S. LULI
´
C
Hvar Observatory, Faculty of Geodesy, Kaˇ ci´ ceva 26, HR-10000 Zagreb, Croatia
(Received 12 February 2000; accepted 11 April 2000)
Abstract. The ignition of coronal shock waves by flares is investigated. It is assumed that an explo-
sive expansion of the source region caused by impulsive heating generates a fast-mode MHD blast
wave which subsequently transforms into a shock wave. The solutions of 1-D MHD equations for the
flaring region and for the external region are matched at their boundary. The obtained results show
under what conditions flares can ignite shock waves that excite the metric type II bursts. The heat
input rate per unit mass has to be sufficiently high and the preflare value of the plasma parameter β
in the flaring region has to be larger than β
crit
0
. The critical values depend on the flare dimensions
and impulsiveness. Larger and more impulsive flares are more effective in generating type II bursts.
Shock waves of a higher Mach number require a higher preflare value of β and a more powerful
heating per unit mass. The results demonstrate why only a small fraction of flares is associated with
type II bursts and why the association rate increases with the flare importance.
1. Introduction
Metric type II bursts (Nelson and Melrose, 1985) reveal the propagation of fast-
mode MHD shock waves in the solar corona (Uchida, 1974). The question whether
these shocks are caused by flares or fast material ejecta is the subject of numer-
ous studies (for a review, see Cliver, Webb, and Howard, 1999). Let us briefly
summarize some of the observational results regarding this problem.
The majority of metric type II bursts starts at frequencies close to or below
100 MHz, several minutes after the peak of the associated microwave burst (Har-
vey, 1965; Švestka and Fritzová-Švestková, 1974). The radio emission usually
ceases at frequencies higher than 20 MHz (Nelson and Melrose, 1985). Occasion-
ally, type II bursts start in the decimetric wavelength range, and their beginning can
precede the peak of the associated microwave burst (Vršnak et al., 1995).
It is well known that there is an association between metric type II bursts and
flares and that the association rate increases with the flare importance (Cliver,
Webb, and Howard, 1999). Drago and Tagliaferri (1967) found a correlation be-
tween the rise times of the Hα flare emission and the time delays of the associated
type II bursts. More recently, Pearson et al. (1989) have shown that a weak cor-
relation exists between the impulsiveness of hard X-ray bursts and the onset fre-
quencies and time delays of the type II bursts. Vršnak et al. (1995) have shown that
microwave bursts associated with type II bursts occurring in the dm–m wavelength
range have in average about two times higher peak fluxes than flares of the same
Solar Physics 196: 181–197, 2000.
© 2000 Kluwer Academic Publishers. Printed in the Netherlands.
182 B. VRŠNAK AND S. LULI
´
C
importance not associated with type II bursts. Vršnak (2000) has found a relation
between the time delay of type II bursts and the impulsiveness of the associated
microwave bursts.
Recently, Klassen et al. (1999) showed that the occurrence of a type II burst
is often preceded by a type II burst precursor, consisting of numerous impulsive
fast-drifting bursts in the decimetric wavelength range. This feature is usually
characterized by a slowly drifting high frequency edge and is associated with an
impulsive microwave and hard X-ray burst (Klassen et al., 1999).
Observations indicate that type II burst shock waves have velocities in the order
of 1000 km s
−1
and low Mach numbers, usually between 1.2 and 1.7 (Nelson
and Melrose, 1985). Similar velocities are observed in the case of chromospheric
Moreton waves spreading out from the flare site (Moreton, 1960; Smith and Har-
vey, 1971). Uchida (1974) has shown that both phenomena are probably caused
by a common shock wave ignited by a flare. Vršnak and Luli´ c (2000, hereafter
Paper I) have shown that such shock waves can be generated by an abrupt, but not
necessarily superalfvénic, expansion of the source region.
In Paper I the process driving the expansion of the source region was not spec-
ified. However, considering the time/distance scales it was inferred that the ‘high-
frequency’ type II bursts starting in the dm–m wavelength range are most prob-
ably caused by solar flares, since even the most abrupt ejections – flare sprays
(Tandberg-Hanssen, Martin, and Hansen, 1980) – are not impulsive enough to cre-
ate such a type II burst. In this paper an explosive expansion of the flaring volume
will be considered as a source of the shock wave. Signatures of such a flare-ignited
process will be discussed and confronted with observations. Furthermore, an order
of magnitude analysis will be presented, to show under what conditions flares can
ignite shock waves that excite metric type II bursts. The results can explain why
only a small fraction of flares is associated with type II bursts.
2. Expansion of the Flaring Volume
2.1. THE MODEL
In the following, a 1-D model will be considered, where all quantities are uniform
in the y- and z-directions and the magnetic field is oriented in the y-direction (Fig-
ure 1). The magnetic field, plasma density, temperature and pressure are taken to be
uniform initially, having the values B
0
, ρ
0
, T
0
, and p
0
, respectively. Furthermore,
it will be assumed that the plasma is perfectly conducting (magnetic diffusivity
η = 0) implying that the ‘frozen-in’ condition is satisfied (Priest, 1982). The
initial value of the ratio of the plasma and the magnetic pressure is taken to be
β
0
= p
0
/p
B
0
1, and it is assumed that the plasma is at rest.
Let us assume that an impulsive heating of the source region begins at t = 0 and
that the heat input rate per unit volume is described by the function q(x, t ). The
FORMATION OF CORONAL MHD SHOCK WAVES, II 183
Figure 1. Definition of the flaring region (shaded) and the external region: (a) before the onset of
heating; (b) after the heating had started. The magnetic field B
i
and the density ρ
i
in the internal
region decrease due to the expansion, whereas in the external region B
e
and ρ
e
increase.
behaviour of the system is governed by the MHD equations (see Paper I) which
can be written in a 1-D situation as:
ρ
_
∂u
∂t
+u
∂u
∂x
_
= −
∂p
tot
∂x
, (1)
ρ
_
∂e
∂t
+u
∂e
∂x
_
+p
∂u
∂x
= q , (2)
∂B
∂t
+
∂(Bu)
∂x
= 0 , (3)
∂ρ
∂t
+
∂(ρu)
∂x
= 0 , (4)
p =
ρ
m
p
kT = (γ −1)ρe . (5)
Here p
tot
= p +B
2
/2µ
0
= p +p
B
is the total pressure, m
p
is the proton mass and
γ = (s +2)/s is the ratio of specific heats determined by the number of degrees of
freedom s. Furthermore, according to Figure 1, the magnetic field and the plasma
flow velocity are denoted as B = B
y
, u = u
x
, respectively.
It will be assumed that the heat is deposited only into the region between x = 0
and the marginal magnetic field line initially located at x = x
0
, i.e. to the same
plasma element (Figure 1). The boundary at x = 0 is considered fixed, whereas
the other boundary (further on simply the boundary) can move freely and its co-
ordinate and velocity will be denoted as x
t
and u
t
:
u
t
=
∂x
t
∂t
, x
t
= x
0
+x
t
= x
0
+
t
_
0
u
t
dt . (6)
The velocity u
t
will be called the expansion velocity.
184 B. VRŠNAK AND S. LULI
´
C
At any moment, the heat is released only within 0 < x < x
t
(further on flaring
volume or i-region), and all the quantities describing a physical state of this volume
will be denoted by the subscript ‘i’. The expansion of the flaring volume causes a
compression of the plasma in the ambient region (further on external region or
e-region – the related quantities will be denoted by the subscript ‘e’). A large am-
plitude MHD perturbation is created, spreading through the e-region. The leading
edge of the perturbation forms during the time interval 0 < t < t
m
in which the
expansion velocity increases from u = 0 to the maximum value u
m
. The spatial
profile of the leading edge subsequently steepens and a shock wave forms after a
time/distance determined by the expansion velocity time profile (see Paper I).
2.2. THE BEHAVIOUR OF THE INTERNAL REGION
For a provisional function q(x, t ) the evolution of the system can be determined
only numerically. In this paper a particular family of q(x, t ) functions will be
considered, allowing for an analytical description of the heated volume expansion.
In order to obtain the explicit expressions relating the heat input rate and the ex-
pansion velocity, the problem will be inverted. Instead of specifying the function
q(x, t ) and searching for the response of the system, it will be demanded that the
flaring volume expands ‘uniformly’, meaning that the plasma density within the
flaring volume remains uniform during the expansion. Then, the solution of the
system of Equations (1)–(5) will be found for a prescribed function u
t
, i.e. the
evolution of the system will be completely determined imposing the kinematics of
the flaring volume boundary.
Since the frozen-in condition is satisfied (η = 0), in the 1-D situation the
conservation of magnetic flux and mass (Equations (3) and (4)), together with the
imposed ‘uniform’ expansion constraint (∂ρ
i
/∂x = 0), imply that the magnetic
field B
i
also remains uniform during the expansion (∂B
i
/∂x = 0) and one can
write:
B
i
B
0
=
ρ
i
ρ
0
=
x
0
x
t
. (7)
So, within the flaring volume the magnetic pressure p
B
i
= B
2
i
/2µ
0
, normalized
with respect to the initial value p
B
0
= B
2
0
/2µ
0
= ρ
0
v
2
A
0
/2, can be expressed as
P
B
i

p
B
i
p
B
0
=
_
B
i
B
0
_
2
=
_
x
0
x
t
_
2
. (8)
Going back to Equation (4) and taking into account ∂ρ
i
/∂x = 0 one finds:
1
ρ
i
∂ρ
i
∂t
= −
∂u
∂x
. (9)
Since ρ
i
does not depend on x one can write ∂u/∂x = C, where C does not depend
on x. This implies that the plasma flow velocity within the flaring volume (x < x
t
)
can be expressed as
FORMATION OF CORONAL MHD SHOCK WAVES, II 185
u =
x
x
t
u
t
, (10)
since it is assumed that u = 0 at x = 0.
Using Equations (6) and (10), Equation (1) can be transformed to obtain
ρ
i
∂u
t
∂t
x
x
t
= −
∂p
i
∂x
, (11)
where it was taken into account that ∂B
i
/∂x = 0 so that ∂p
B
i
/∂x = 0. Since
ρ
i
∂u
t
/∂t does not depend on x, the integration of Equation (11) with respect to x
implies that the gas pressure in the i-region is of the form p
i
= C
1
− C
2
(x/x
i
)
2
,
where C
1
and C
2
do not depend on x. Denoting the gas pressure at x = 0, at the
moment t , as p(x = 0, t ) ≡ p
0
t
and the difference of the gas pressure at x = 0 and
at x = x
t
as p
t
one finds:
p
i
= p
0
t
−p
t
_
x
x
t
_
2
. (12)
The superscript ‘0’ will be used further on to denote quantities at x = 0.
The plasma pressure p
i
at the flaring volume boundary at the moment t is
p
i
(x = x
t
, t ) ≡ p
b
t
= p
0
t
− p
t
. The superscript ‘b’ will be used further on to
denote quantities at the flaring volume boundary. Substituting Equation (12) into
Equation (11) one finds
ρ
i
2
∂u
t
∂t
=
p
t
x
t
, (13)
which determines the acceleration of the boundary:
∂u
t
∂t
=
2p
t
x
0
ρ
0
. (14)
Here, Equation (7) was used to eliminate ρ
i
.
Using Equations (6), (7), and (10), as well as Equation (5) with the ratio of
specific heat capacities γ =
5
3
, i.e., using e = 3p/2ρ, Equation (2) can be written
as
q =
3
2
∂p
i
∂t
+
3
2
p
i
x
t
u
t
+
3
2
∂p
i
∂x
x
x
t
u
t
+
p
i
x
t
u
t
. (15)
Taking into account Equation (12) one finds for 0 < x < x
t
q(x, t ) = q
0
t
−q
t
_
x
x
t
_
2
, (16)
where
q
0
t
=
3
2
∂p
0
t
∂t
+
5
2
p
0
t
x
t
u
t
, (17)
186 B. VRŠNAK AND S. LULI
´
C
and
q
t
=
3
2
∂p
t
∂t
+
5
2
p
t
x
t
u
t
(18)
are the rate at which the heat is deposited at x = 0 and the difference between
the heat released at x = 0 and x = x
t
, respectively. The expansion velocity u
t
is
related to p
t
by Equation (14), and x
t
is defined by Equation (6). Equation (16)
shows that the heat deposited at the boundary q(x = x
t
, t ) ≡ q
b
t
= q
0
t
− q
t
can
be expressed as
q
b
t
=
3
2
∂p
b
t
∂t
+
5
2
p
b
t
x
t
u
t
. (19)
3. Results
3.1. ‘MATCHING’ OF THE INTERNAL AND EXTERNAL REGION
Fast expansion of the flaring region causes a compression of the magneto-plasma
ahead the boundary (Figure 1(b)). In Paper I it was shown that starting with a
β
0
0, the gas pressure in the e-region remains much smaller than the magnetic
pressure (β
e
0). Furthermore, it was shown that the factor of compression ≡
ρ
e

0
= B
e
/B
0
depends on the velocity of the boundary:

t
=
_
1 +
U
t
2
_
2
, (20)
where U
t
= u
t
/v
A
0
. The highest value of compression
m
= (U
m
) is related to
the final Mach number of the shock wave as
M =
_
5 +
m
8 −2
m

m
(21)
(see Paper I).
Since the total pressure p
tot
= p+p
B
must be a continuous function, p
tot
i
(x
t
) =
p
tot
e
(x
t
) must hold at the flaring volume boundary, and one can write
B
2
i

0
+p
b
t
= p
tot
e
(x
t
) ≈
B
2
e

0
, (22)
since β
e
1 is assumed. Normalizing Equation (22) with respect to the initial
magnetic pressure p
B
0
= B
2
0
/2µ
0
and using the relations B
i
/B
0
= x
0
/x
t
and
B
e
/B
0
= one finds
P
b
τ
=
2
−X
−2
τ
=
_
1 +
U
τ
2
_
4
−X
−2
τ
, (23)
FORMATION OF CORONAL MHD SHOCK WAVES, II 187
where the normalized time τ = t /t
m
and the normalized co-ordinate of the bound-
ary X
τ
= x
t
/x
0
were introduced.
Equation (6) can be written in the normalized form as
U
τ
=
∂X
τ
∂τ
, X
τ
= 1 +
1

_
τ
0
U
τ
dτ , (24)
where = t
A
/t
m
, and t
A
= x
0
/v
A
0
represents the Alfvén travel time across the
flaring region. Similarly, Equation (14) can be normalized such that
P
τ
=
∂U
τ
∂τ
, (25)
where P
τ
= p
t
/p
B
0
. One can also write
T
0
τ
T
0
= X
τ
P
0
τ
β
0
, (26)
where T
0
τ
is the plasma temperature at x = 0 (further on the central temperature)
and P
0
τ
= p
0
t
/p
B
0
.
Prescribing the function U
τ
Equations (23), (24), and (25) determine P
b
τ
, X
τ
,
and P
τ
, respectively. Furthermore, P
b
τ
and P
τ
determine P
0
τ
= P
b
τ
+ P
τ
(Equation (12)). Normalizing Equations (17) and (19) with respect to the initial
magnetic field energy density (Q
0
τ
= q
0
t
/p
B
0
and Q
b
τ
= q
b
t
/p
B
0
) one can write
Q
0
τ
=
ϕ
0
τ
t
A
=
φ
0
τ
t
m
; Q
b
τ
=
ϕ
b
τ
t
A
=
φ
b
τ
t
m
, (27)
where the dimensionless functions φ
0
τ
and φ
b
τ
read
φ
0
τ
=
3
2
∂P
0
τ
∂τ
+
5
2
P
0
τ
U
τ
X
τ
, (28)
and
φ
b
τ
=
3
2
∂P
b
τ
∂τ
+
5
2
P
b
τ
U
τ
X
τ
. (29)
The functions ϕ can be expressed as ϕ = φ. The functions ϕ
0
τ
, ϕ
b
τ
, φ
0
τ
and φ
b
τ
depend on the parameter = t
A
/t
m
and on the highest value of the expansion
velocity U
m
. Large flares (large t
A
= x
0
/v
A
0
) and short-duration flares (short t
m
)
have a larger . Bearing in mind that the energy release process caused by the
magnetic field reconnection develops on a time scale larger than the Alfvén time
scale (Priest and Forbes, 1986), i.e., t
A
< t
m
, it will be assumed that < 1. Taking
as an example a very large and fast developing event characterized by x
0
10
5
km
and t
m
≈ 100 s, and using for the Alfvén velocity v
A
0
≈ 1000 km s
−1
, one finds
1.
Knowing Q
0
τ
and Q
b
τ
one can evaluate Q
τ
= Q
0
τ
− Q
b
τ
, determining the
heating in the flaring volume:
188 B. VRŠNAK AND S. LULI
´
C
Q(x, τ) = Q
0
τ
−Q
τ
_
x
x
τ
_
2
(30)
(see Equation (16)). In this way the solution of the system of Equations (1)–(5) is
completed. Integrating Equation (28) over the interval 0 < x < x
τ
one finds the
rate at which the energy is released in the entire flaring region:
Q
tot
(τ) = (
2
3
Q
0
τ
+
1
3
Q
b
τ
)x
t
. (31)
Using the relations x
t
= X
τ
x
0
and v
A
0
= x
0
/t
A
one finally finds
Q
tot
(τ) = v
A
0
(
2
3
ϕ
0
τ
+
1
3
ϕ
b
τ
)X
τ
≡ v
A
0

(τ) , (32)
where the introduced dimensionless function (τ) depends on the parameter , as
well as on U
m
.
3.2. AN EXAMPLE
Let us consider as an example the flare whose evolution is governed by the gener-
ating function denoted as F1 in Paper I. It is defined as
U = U
m
sin
2
_
π
2
τ
_
(33)
and its behaviour in the time interval 0 < τ < 1 determines the spatial profile of
the leading edge of the associated blast wave.
In Figure 2 the evolution of the heating of the flaring region is shown for several
values of the parameters and U
m
, illustrating properties of flares developing on
different spatial/time scales and characterized by different expansion velocities. A
faster expansion (larger value of U
m
) implies a higher Mach number of the shock
wave generated by the flare (see Paper I). Spatially large flares (large x
0
, i.e., large
t
A
= x
0
/v
A
0
) and fast developing flares (short t
m
) are characterized by a larger .
The dimensionless function (τ) defined by Equation (32) is shown in Fig-
ure 2(a). The rate at which the heat is released in the flaring volume can be evalu-
ated as q
tot
(τ) = p
B
0
v
A
0
(τ). Figure 2(a) indicates that the maximum value
max
and the time when it is achieved (τ
max
) primarily depend on the applied value of
U
m
. The values of
max
do not depend significantly on the value of the parameter
. So, it can concluded that a higher value of U
m
(and thus the higher Mach num-
ber of the associated shock wave) requires more powerful heating. Furthermore,
a stronger preflare magnetic field and a higher Alfvén velocity require a more
powerful heating q
tot
(τ) for a given U
m
.
The behaviour of the central temperature is shown in Figure 2(b), where β
0
=
0.1 was used to evaluate Equation (26). A lower value of β
0
would give higher
temperatures. One finds that the central temperature is higher in the case of a small
and a large U
m
. This implies that small flares have to be hotter than large flares in
order to generate an expansion of the same U
m
. Such behaviour is consistent with
FORMATION OF CORONAL MHD SHOCK WAVES, II 189
Figure 2. Heating of the flaring region shown for different values of U
m
and = t
A
/t
m
. (a) The
dimensionless function (τ) shown for U
m
= 0.8 (thin) and U
m
= 0.4 (thick) obtained using
= 0.2 (dotted), = 0.1 (full) and = 0.01(dashed). (b) The ‘central’ temperature ratio T
0
τ
/T
0
obtained using U
m
= 0.8 (thick) and U
m
= 0.4 (thin) for = 0.05 (dotted), = 0.1 (full)
and = 0.2 (dashed). (c) The dimensionless function φ
0
(τ) shown for U
m
= 0.8 and = 0.1
(thick-full); U
m
= 0.4 and = 0.1 (dotted); U
m
= 0.4 and = 0.05 (dashed). The function φ
b
(τ)
is shown for U
m
= 0.8 and = 0.1 by the thin-full line. (d) The difference φ = φ
0
− φ
b
for the
same values of parameters as in (c).
the results shown in Figure 2(a). For a given U
m
, the total heating rate Q
tot
does
not depend significantly on the dimensions of the flare. So, the rate at which the
heat is released per unit volume is higher in small flares, resulting in higher plasma
temperatures. Figure 2(b) shows that the temperature maximum is achieved after
τ = 1.
The dimensionless function φ
0
(τ) defined by Equation (28) is shown in Fig-
ure 2(c). It describes the heat input rate at X = 0. The heating rate per unit volume
can be evaluated according to Equation (27) as q(τ) = (p
B
0
/t
m
)φ(τ). For the
matter of illustration, the function φ
b
(τ) (heating rate at the boundary) is also
shown for U
m
= 0.8 and = 0.1 (thin full line). In Figure 2(d) the difference
between the heating rate at X = 0 and at the boundary φ(τ) = φ
0
(τ) − φ
b
(τ)
is presented for the same values of parameters as used in Figure 2(c). One finds
that in the case of the uniform expansion considered here, the heat is released
almost uniformly across the flaring volume (see the thick and thin full lines in
Figure 2(c)); comparing Figures 2(c) and 2(d) one finds that the differences φ
are much smaller than the peak value of φ. Furthermore, it is evident that the heat
release spreads from the central region towards the boundary during the process.
190 B. VRŠNAK AND S. LULI
´
C
Figure 3. (a) The highest value of the heating rate of the flaring region represented by
max
as a
function of U
m
for = 1 (shaded), = 0.5 (dashed), = 0.2 (dotted) and = 0.1 (full). (b)

max
as a function of for U
m
= 0.4 (dotted), U
m
= 0.8 (full) and U
m
= 1.2 (dashed). (c) τ
max
as a function of for the same values of U
m
as used in (b). (d) The impulsiveness of the heating
imp =
max
/t
max
shown as a function of U
m
/t
m
for t
m
= 100 s and for the same values of as in
(a).
In the beginning the heating is more powerful at X = 0, whereas later it is more
powerful at the boundary.
Figure 2(c) shows that a higher value of U
m
requires a more powerful heating
per unit volume q(τ) for a given B
0
and t
m
. Furthermore, q(τ) depends on the
parameter . The maximum value of q needed to achieve a given U
m
is higher in
spatially small flares (small t
A
, i.e., small ).
In Figure 3 the dependence of the highest value of the total heating rate of the
flaring volume and the impulsiveness of the heating process on the parameters U
m
and is depicted.
max
is presented as a function of U
m
and in Figures 3(a) and
3(b), respectively. One finds that the value of the highest heat input rate Q
max
=
v
A
0

max
depends primarily on U
m
and v
A
0
.
Figure 3(c) shows the dependence of the time of the maximum heat input rate
τ
max
on the parameter for different values of U
m
. Figure 3(d) depicts the impul-
siveness of heating imp =
max

max
as a function of U
m
/t
m
. One finds that imp is
a monotonic function of U
m
/t
m
that does not significantly depend on in the range
of values < 0.2. This indicates that the starting frequency and the time delay of
FORMATION OF CORONAL MHD SHOCK WAVES, II 191
the type II burst excited by a flare are related primarily to the impulsiveness of the
heat input rate in the flaring volume (see Figure 7 in Paper I).
3.3. CONDITIONS FOR TYPE II BURST FORMATION
3.3.1. The Plasma Parameter β
Using Equations (5) and (7), Equation (23) can be rewritten in the form
_
x
0
x
t
_
2

0
x
0
x
t
T
i
T
0
=
2
, (34)
which gives
T
i
T
0
=
1
β
0
_
x
t
x
0

2

x
0
x
t
_
, (35)
In the following, the quantity x
t
m
= u
m
t
m
/2 will be used as an order of
magnitude estimate of the increase of the flaring volume due to its expansion in
the interval 0 < t < t
m
. In the normalized form it can be expressed as X
τ=1

X
m
= U
m
/2 where X
m
= x
t
m
/x
0
.
Let us now consider in more detail two extreme types of events. The first type of
events are those satisfying the condition x
t
m
/x
0
1 (i.e., X
m
≡ x
t
m
/x
0
1).
We will call such an event the small flare since the stated condition is easily
achieved for a small value of x
0
, i.e., small value of . The other extreme is an
event satisfying x
t
m
/x
0
1 (i.e., x
t
m
/x
0
≈ 1). Since X
m
= U
m
/2 one
finds that such events have to be large and fast developing (small t
m
and large x
0
,
implying large ). We will call such an event simply the large flare.
In the small-flare case, Equation (35) can be written in an order of magnitude
form as
β
0

2
m
x
t
m
x
0
T
0
T
i
. (36)
In the large-flare case, Equation (35) reduces to
β
0
≈ (
2
m
−1)
T
0
T
i
. (37)
The compression
m
is related to U
m
by Equation (20) and to the Mach number M
of the associated shock wave by Equation (21).
Equations (36) and (37) can be combined with one constraint imposed by ob-
servations. In the hottest parts of a flare the temperature usually ranges between
10
7
and 10
8
K, and is rarely exceeding 10
8
K (Dennis and Schwartz, 1989; Kosugi,
1994). So, taking T
i
< 2×10
8
K, one can use the limit T
i
/T
0
< 100. In Figure 4(a)
β
0
= β
crit
0
is shown as a function of the Mach number of the associated shock wave
using T
i
/T
0
= 100. A flare can generate a type II burst of a given Mach number
only if it occurs in a coronal region having β
0
above β
crit
0
(M). The large-flare case,
192 B. VRŠNAK AND S. LULI
´
C
depicted by the thick line, is obtained straightforwardly using Equations (20), (21),
and (37). In the case of a small flare one can approximately take x
t
m
≈ x
t
m
since
x
t
m
x
0
. Estimating x
t
m
approximately as x
t
m
≈ u
m
t
m
/2, Equation (36) can
be transformed into an order of magnitude form which reads
β
0

2
U
m
2
T
0
T
i
. (38)
The results obtained using Equations (38) and taking T
i
/T
0
= 100 are shown in
Figure 4(a) by thin lines for different values of . Figure 4(a) shows that for a given
value of a higher Mach number requires a higher value of β
0
. Furthermore, it
can be concluded that for a given β
0
and M, small flares (small t
A
) have to develop
faster (shorter t
m
). Finally, for the same M, a less impulsive flare must occur in a
higher β
0
region.
Let us compare an extremely small flare characterized by x
0
= 1000 km, with
a larger flare characterized by x
0
= 10 000 km. Assuming that the impulsive heat
release in both flares is lasting for t
m
= 100 s, and taking v
0
= 1000 km s
−1
one finds values of as 0.01 and 0.1, respectively. Observations of type II bursts
indicate that usually M ≥ 1.2 holds (Nelson and Melrose, 1985) which corre-
sponds to U
m
≥ 0.25 and ≥ 1.26. Using these values and taking for the ratio
of temperatures T
i
/T
0
= 100, Equation (38) gives the values β
crit
0
= 0.2 for the
smaller flare and β
crit
0
= 0.02 for the larger one. In the low corona, where flares
occur, β 1 is usually valid (Dulk and McLean, 1978). This implies that very
small flares can produce a type II burst only under exceptional circumstances when
coronal conditions deviate from β 1.
Equations (37) and (38) show that the flares producing type II bursts have to be
very hot and impulsive. Figure 4(a) demonstrates that large flares can produce type
II bursts more easily, i.e., the critical values are more convenient for the coronal
conditions. Furthermore, one finds that larger and more impulsive flares can gen-
erate shock waves of higher Mach numbers, and thus the type II bursts of higher
starting frequencies and shorter onset time delays.
3.3.2. Shear of the Preflare Magnetic Field
Let us consider another aspect of Equation (23). The heat release rate per unit
mass can be related to the heat release rate per unit volume q as q = ρ
i
.
Applying Equation (7) this relation can be written as q = ρ
0
x
0
/x
t
. Substituting
this expression into Equation (19) one finds

b
ρ
0
=
3
2
∂p
b
t
∂t
x
t
x
0
+
5
2
p
b
t
x
0
u
t
. (39)
In the small-flare case (x
t
m
≈ x
t
m
) Equation (39) can be written in an order-
of-magnitude form expressing the average heat release rate per unit mass in the
interval 0 < t < t
m
as

b

p
b
m
u
m
ρ
0
x
0
, (40)
FORMATION OF CORONAL MHD SHOCK WAVES, II 193
Figure 4. The lower limit of (a) the preflare value of the plasma parameter β
0
; (b) the total heat
released per unit mass; (c) the critical value of the shear ϑ. The large-flare case is depicted by the
thick line and the small-flare cases are shown by thin lines denoted by the applied values of . The
two dotted thick vertical lines represent the Mach number range inferred from observations and the
thick vertical line shows the Mach number corresponding to U
m
= 1, i.e., u
m
= v
A
0
.
where p
b
m
is the highest pressure attained at the flaring volume boundary. Equa-
tion (40) was obtained replacing x
t
, p
b
t
, and u
t
in Equation (39) by the average
values x
t
, p
b
t
, and u
t
, respectively. Then, we used the approximations ∂p
b
t
/∂t ≈
p
b
m
/t
m
, x
t
≈ x
t
m
/2, x
t
m
≈ u
t
t
m
, u
t
≈ u
m
/2, and p
b
t
≈ p
b
m
/2.
When the condition x
t
m
≈ x
t
m
is satisfied, Equation (23) can be written
approximately as
P
b
m

2
, (41)
where P
b
m
= p
b
m
/p
B
0
. Multiplying Equation (40) by t
m
ρ
0
/p
B
0
and substituting P
b
m
from Equation (41), the total heat (E
b
=
b
t
m
) liberated per unit mass at the flaring
volume boundary during t
m
can be expressed in the normalized form as
194 B. VRŠNAK AND S. LULI
´
C
E
b
E
B
0

2
U
m

, (42)
where E
B
0
= p
B
0

0
is the magnetic energy associated with B
0
that is contained
in an unit plasma mass element. The dependence of E
b
/E
B
0
on U
m
is shown in
Figure 4(b) for different values of .
Let us presume that the heat is liberated in situ as a part of the energy release
process causing the flare. This means that it is provided by the preflare magnetic
field component associated with electric currents, i.e., with the stored free energy.
Denoting this component of the magnetic field as B
z
, the ratio B
z
/B
0
can be related
to the total heat released per unit mass during the time t
m
as (B
z
/B
0
)
2
= E
b
/E
B
0
.
On the other hand, one can express the ratio B
z
/B
0
introducing the shear angle ϑ
as B
z
/B
0
= tan ϑ. Equation (42) then implies
ϑ > arctan
_

_
U
m

_
, (43)
since Equation (16) implies that E
0
> E
b
.
The dependence of the lower limit of the shear angle ϑ on the Mach number of
the shock wave as determined by Equation (43) for the small-flare case is presented
in Figure 4(c) by the thin lines. Assuming that the shear angle ϑ is related to the
shear of the photospheric magnetic field, Figure 4c indicates that for M ≥ 1.2
small flares must occur in regions of extremely large shear. Bearing in mind that
there is also a non-thermal component of the energy release, the critical values are
even higher.
When x
t
m
/x
0
1 (large flare) Equation (19) can be written in an order of
magnitude form as

b

3
2
p
b
m
t
m
. (44)
For x
t
m
/x
0
1 Equation (23) can be approximately written as
P
b
m

2
−1 . (45)
Following the same procedure as in deriving Equation (43) one finds:
ϑ > arctan
_
1.5(
2
−1) . (46)
Equation (46) gives a lower limit of the shear ranging from ϑ > 45

for M = 1.2
to ϑ > 70

for M = 2. The result given by Equation (46) is shown in Figure 4(c)
by the thick line. Figure 4(c) indicates that larger flares are more effective in
generating perturbations that cause type II radio bursts than the smaller ones.
FORMATION OF CORONAL MHD SHOCK WAVES, II 195
4. Discussion and Conclusion
The energy released in solar flares is provided by free magnetic energy accumu-
lated in the non-potential preflare magnetic field structure. It can be presumed that
the energy is liberated by the fast reconnection of the magnetic field, proceeding
at some 10% of the Alfvén velocity (Priest and Forbes, 1986). The heat needed
for the flaring volume expansion that generates the blast can be released by two
mechanisms. It can be liberated in situ by the reconnection process itself (Vrš-
nak, 1989), simultaneously with the non-thermal component. This is sometimes
observed as a very hot, hard X-ray emitting plasma, at the summits of flaring
loops (Kosugi, 1994). The other possible mechanism is the thermalization of the
accelerated particles. In this case the blast wave ignition should take place in lower
layers of the solar atmosphere (Karlický and Odstrˇ cil, 1994). In both cases one can
expect that the heat release (q
tot
(t )) is similar to the hard X-ray or microwave burst.
The back-extrapolations of the type II emission lanes indicate that the blast is
ignited close to the peak of an impulsive microwave and hard X-ray burst (Vrš-
nak et al., 1995). Simultaneously, numerous fast-drifting bursts occur in the dm-
m wavelength range, forming a type II burst precursor (Klassen et al., 1999).
The analysis presented in this paper shows that the highest heat input rate oc-
curs somewhat earlier than the maximum expansion velocity is achieved (Fig-
ure 3(c)). Inspecting the examples shown in Figure 6 in Paper I, one finds that
back-extrapolations of the synthesized type II burst harmonic lanes point to the
segment of the ‘precursor’ corresponding to the time of the most powerful heating.
Assuming that the hard X-ray and microwave bursts reveal processes that also pro-
vide the plasma heating, one finds a good correspondence between the observations
and the model.
Taking as an example the generating function defined by Equation (33) it was
shown that the average acceleration U
m
/t
m
is higher for a higher impulsiveness
of the heat input rate Q
max

max
(Figure 3(d)). This means that the starting fre-
quencies and time delays of type II bursts should depend on the energy release
impulsiveness (see Figure 7 in Paper I). Such a dependence was found by Pearson
et al. (1989) for hard X-ray bursts and by Vršnak (2000) for microwave bursts.
Two conditions for the type II burst onset were inferred in Section 3.3. Firstly,
the preflare value of the parameter β
0
has to be above some critical value, as shown
by Equations (37) and (38). The critical value is higher for smaller and less im-
pulsive flares. For example a flare characterized by = 0.1 can generate a shock
wave of the Mach number M = 1.2 only if it occurs in a β
0
> 0.02 region. Such a
value of the parameter is a characteristic of e.g., a flare in a v
A
0
= 1000 km s
−1
environment, developing on the time scale of t
m
= 100 s and having the length
scale of x
0
= 10
4
km. A lower value of (e.g., a longer time scale or a smaller
length scale) implies a higher necessary value of β
0
. Similarly, a higher value of M
requires a higher β
0
or a larger .
196 B. VRŠNAK AND S. LULI
´
C
Secondly, the heat released per unit mass (E) must be sufficiently high (Fig-
ure 4(b)). It has to be at least several times higher than the magnetic field energy
contained in an unit mass plasma element. The ratio is higher for smaller and less
impulsive flares (Figure 4(b)). This also implies that such flares have to be hotter
considering the same value of B
0
(Figure 2(b)). Furthermore, if the heat is produced
in situ, a higher value of E requires that a larger amount of the free magnetic field
energy is stored in an unit mass. So, the shear of the magnetic field has to be larger
in smaller and less impulsive flares.
The inferred conditions for the type II burst occurrence can be confronted with
observations analysing well observed events like in Aurass et al. (1999). The com-
parison of the two events presented there shows that the smaller 1N/C4.7 flare
(named E2) was associated with a type II burst, whereas the larger 2B/M4.4 flare
(named E1) was not. The event E1 took place in a region characterized by a modest
shear of only 40

–50

, whereas the event E2 occurred in a region of a strong shear
of 70

–80

. This is consistent with the conditions given by Equations (43) and
(46) and the results presented in Figure 4(c). A much stronger shear in the case of
the event E2 provided a larger amount of the stored energy density, i.e., a higher
value of E was provided, although the magnetic field was somewhat weaker than
in the event E1. Furthermore, it was estimated that although the flare E1 released
about ten times more energy during the non-thermal energy release phase, the rate
at which it was released per unit mass was more than four times higher in the flare
E2 since it was released in a smaller volume and in a shorter time.
Regarding the condition on β
0
let us note that in some cases homologous flaring
can be essential for type II burst formation. The first flare in the sequence causes
an increase of the coronal density due to the ‘evaporation’ process, so that the next
flare occurs in a higher β environment (Strong et al., 1984). So, it is possible that
the first of the two successive flares in the same active region is not associated with
a type II burst, whereas the second one is. Such an effect can presumably explain
the apparent contradictions found by Cliver, Webb, and Howard (1999) in a critical
re-examination of the ‘Alfvén velocity condition’. It was demonstrated there that
some of flares occurring successively in the same active region produced type II
bursts and some did not. They concluded that since the Alfvén velocity cannot
change significantly within few hours, the type II burst excitation by a flare can not
depend critically on the value of the Alfvén velocity. However, the value of β
0
can
change easily due to the ‘evaporation’ process, changing the conditions for type II
burst occurrence.
Finally, let us summarize the conditions favourable for the flare-ignited shock
wave formation:
– a relatively high preflare plasma β environment, characterized by low Alfvén
velocity;
– a highly sheared preflare magnetic field structure;
FORMATION OF CORONAL MHD SHOCK WAVES, II 197
– a preflare state that provides a fast developing instability which can drive the
energy release process on a time scale comparable with the Alfvén travel time and
which is efficient in releasing the heat;
– a large spatial extent of the unstable magnetic structure.
The variety of necessary conditions for the shock wave formation may explain
why only a small fraction of flares is associated with type II bursts. Even large
flares are not always associated with type II bursts since some of the necessary
conditions may not be satisfied. The estimates presented in Section 3.3. explain
the tendency that the association rate of type II bursts increases with the flare
importance: the criteria are less restrictive for larger flares. Finally the presented
analysis demonstrates why type II bursts are characterized by low Mach numbers.
A large Mach number would require an extremely powerful heating, causing an
unreasonably large temperature increase.
References
Aurass, H., Vršnak, B., Hofmann, A., and Ruždjak, V.: 1999, Solar Phys. 190, 267.
Cliver, E. W., Webb, D. F., and Howard, R. A.: 1999, Solar Phys. 187, 89.
Dennis, B. R. and Schwartz, R. A.: 1989, Solar Phys. 121, 75.
Drago, F. G. and Tagliaferri, G. L.: 1967, Osservatorio Astrofisico Arcetri, Contributo
Dulk, G. A. and McLean, D. J.: 1978, Solar Phys. 57, 279.
Harvey, G. A.: 1965, J. Geophys. Res. 70, 13.
Karlický, M. and Odstrˇ cil, D.: 1994, Solar Phys. 155, 171.
Klassen, A., Aurass, H., Klein, K.-L., Hofmann, A., and Mann, G. : 1999, Astron. Astrophys. 343,
287.
Kosugi, T.: 1994, in S. Enome and T. Hirayama (eds.) Nobeyama Radio. Obs. Report 360, 11.
Moreton, G. E.: 1960, Astrophys. J. 65, 494.
Nelson, G. J. and Melrose, D. B.: 1985, in D. J. McLean and N. R. Labrum(eds.), Solar Radiophysics,
Cambridge University Press, Cambridge, p. 333.
Pearson, D. H., Nelson, R., Kojoian, G., and Seal, J.: 1989, Astrophys.J. 336, 1050.
Priest, E. R.: 1982, Solar Magnetohydrodynamics, D. Reidel Publ. Co., Dordrecht, The Netherlands.
Priest, E. R. and Forbes, T. G.: 1986, J. Geophys. Res. 91, 5579.
Smith, S. F. and Harvey, K. L.: 1971, in C. J. Macris (ed.), Physics of Solar Corona, D. Reidel Publ.
Co., Dordrecht, The Netherlands, p. 156.
Strong, K. T., Benz, A. O., Dennis, B. R., Leibacher, J. W., Mewe, R., Poland, A. I., Schrijver, J.,
Simnett, G., Smith, J. B. and Sylwester, J.: 1984, Solar Phys. 91, 325.
Švestka, Z. and Fritzová-Švestková L.: 1974, Solar Phys. 36, 417.
Tandberg-Hanssen, E., Martin, S. F., and Hansen, R. T.: 1980, Solar Phys. 65, 357.
Uchida, Y.: 1974, Solar Phys. 39, 431.
Vršnak, B.: 1989, Solar Phys. 120, 79.
Vršnak, B.: 2000, in A. Hanslmeier and M. Messerotti (eds.), The Dynamic Sun, Astrophysics and
Space Science Library, Kluwer Academic Publishers, Dordrecht, The Netherlands (in press)
Vršnak, B. and Luli´ c, S.: 2000, Solar Phys. 196, 157 (this issue).
Vršnak, B., Ruždjak, V., Zlobec, P., and Aurass, H.: 1995, Solar Phys. 158, 331.