Drift Shell Bifurcation Near The Dayside Magnetopause in Realistic Magnetospheric Magnetic Fields

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Drift shell bifurcation near the dayside magnetopause in realistic

magnetospheric magnetic fields


Yifei Wan,
1
Stanislav Sazykin,
1
Richard A. Wolf,
1
and M. Kaan ztrk
2
Received 24 February 2010; revised 23 April 2010; accepted 13 May 2010; published 2 October 2010.
[1] We study trapped energetic particles in the terrestrial magnetosphere undergoing
drift shell bifurcation in the magnetic field lacking northsouth and eastwest symmetry.
Drift shell bifurcation occurs near the dayside magnetopause, where, due to the
solar wind compression, the field strength has a local maximum near the equatorial
plane. As a result, a charged particle may become temporarily trapped in one of the
hemispheres while traversing the region. Although this phenomenon has been known
for a long time, only recently were the associated second invariant changes quantified
for the magnetic field with northsouth and eastwest symmetry. Here we show that
if the magnetic field lacks such symmetry, the effect is more significant. We calculate
changes to the second invariant of keV to MeV electrons in Tsyganenko magnetic
fields with nonzero interplanetary magnetic field (IMF) B
Y
component. The changes are
on the order of the invariant itself, and thus, this effect is much larger than for the case
of symmetric magnetic field (when the particle gyroradius is much less than the
magnetospheric scale length). We also quantify the effect for different values of the
solar wind dynamic pressure, IMF B
Z
component, and the Dst index with the
Tsyganenko magnetic field T02. We find that Dst has no noticeable role, while larger
solar wind ram pressure increases the second invariant changes. We verify our calculations
by numerical integration of the guiding center drift equations and discuss properties of
different versions of these equations.
Citation: Wan, Y., S. Sazykin, R. A. Wolf, and M. K. ztrk (2010), Drift shell bifurcation near the dayside magnetopause in
realistic magnetospheric magnetic fields, J. Geophys. Res., 115, A10205, doi:10.1029/2010JA015395.
1. Introduction
[2] In the typical magnetospheric magnetic field config-
uration, trapped particles of sufficiently high energies will
execute gradient and curvature drifts on closed orbits while
gyrating and bouncing between the hemispheres in the
converging magnetic field (mirror) geometry. Separation of
characteristic time scales of the three quasiperiodic motions
leads to the most common mathematical description (use
of adiabatic invariants) to study the dynamics of such par-
ticles [e.g., Northrop, 1963; Walt, 1994]. Processes that
then cause the changes of the adiabatic invariants must
be accounted for separately. In this paper, we analyze
one particular not very wellknown mechanism leading to
changes of the second and third adiabatic invariants for
particles drifting through the narrow region around the
dayside magnetopause, namely, drift shell bifurcation (DSB
for brevity).
[3] Trajectories of particles drifting in a static magnetic
field without electric field could be obtained by numerical
integration of the full relativistic NewtonLorentz equation,
d
dt
mv qv B: 1
[4] Here g = (1 v
2
/ c
2
)
1/2
is the relativistic factor, m
and q are particles mass and charge, v is the velocity vector,
and B is the local magnetic field. However, for reasons to be
explained in section 4, we adopt the socalled guiding center
approach in this work. The phenomenon of drift shell
bifurcation is illustrated later in the left column of Figure 4,
which shows the guiding center trajectories of an electron in
the Earths magnetic field described by the Tsyganenko T02
model (also commonly referred to as T01_01) [Tsyganenko,
2002a, 2002b]. The electron drifts around the Earth while
executing the familiar bouncing motion about the equatorial
plane (except on the dayside when it is trapped into one of
the hemispheres), which typically can be taken as a surface of
minimum magnetic field strength. The last fact is often stated
by saying that the curve of magnetic field strength versus the
distance along the magnetic field line has a Ushape during the
halfbounce period. The surface mapped out by the trajectory
is called a drift shell.
1
Physics and Astronomy Department, Rice University, Houston, Texas,
USA.
2
Department of Information Systems and Technologies, Yeditepe
University, Istanbul, Turkey.
Copyright 2010 by the American Geophysical Union.
01480227/10/2010JA015395
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, A10205, doi:10.1029/2010JA015395, 2010
A10205 1 of 17
[5] Let us denote the first and second adiabatic invariants
as m and J. If the energy is conserved as will be in our case,
the mirror field B
m
can be used in place of m,
B
m

mv
2

2
2
; 2
and the second invariant J can be replaced by the geometric
second invariant I,
I B
m

J
2p

_
sm2
sm1

1
B s
B
m

ds; 3
where p is the momentum, B(s) is the magnetic field strength
along a field line, and s
m1
and s
m2
are the locations of the
mirror points. In a limited region before noon, an electron (if
the mirror field B
m
is sufficiently small) undergoes a rapid
(over a couple of bounce periods) transition (bifurcation) and
becomes trapped in one of the hemispheres. There is another
point past noon where the electron returns to its bounce
motion about the equatorial plane. As can be seen from the
amplitude of the bounce motion, the second invariant I has
changed and the electron is now on a different drift shell.
[6] Trajectories of the type plotted in Figure 4 are often
called Shabansky orbits [Shabansky, 1971]. This phe-
nomenon has been known for a long time [e.g., Northrop and
Teller, 1960; Mead, 1964; Shabansky and Antonova, 1968;
Roederer, 1970; Delcourt and Sauvaud, 1999; ztrk and
Wolf, 2007] and is a result of the magnetic field having
a region on the dayside where, due to compression of the
magnetosphere by the dynamic pressure of the solar wind,
magnetic field strength profiles along field lines have a
W shape, as illustrated in Figure 1 for a typical T02 magnetic
field configuration. In other words, there are two minima that
are off the equatorial plane, and it is possible for particles to
become trapped and bounce about one of them.
[7] It can be argued [e.g., Shabansky, 1971; ztrk and
Wolf, 2007, and references therein] that the first adiabatic
invariant is conserved along the trajectory, but we later verify
in section 7.2 through fullparticle tracing that the first adi-
abatic invariants of the electrons we study are indeed con-
served. Of particular interest is the change in the second
invariant after a particle drifts through the region near the
dayside magnetopause with a mirror field weaker than the
local nearequatorial maximum(see Figure 1). Calculation of
such nonadiabatic change is the subject of this paper. Previ-
ously, ztrk and Wolf [2007] extensively analyzed drift
shell bifurcations and the changes of the second invariant I
for a magnetic field configuration that was a superposition of
two parallel dipoles. Their results were stated for magnetic
fields with northsouth and eastwest symmetry. In the
symmetric case, the change DI in the second invariant re-
sulting from one passage across the day side scales with the
gyroradius r and is typically a few rs (small compared
to initial I). When the result is expressed as a radial diffusion
in L with radiation belt particles in mind, the value of the
diffusion coefficient becomes comparable to that due to
magnetic field fluctuations for radiation belttype electrons of
energies only in excess of 10 MeV. In other words, the effect
DSB has on violation of the second invariant conservation in
the symmetric case is likely small.
Figure 1. Magnetic field strength along the field plotted for three magnetic local times for the T02
model with parameters solar wind ram pressure P
DYN
= 2 nPa, zero tilt angle, IMF B
Y
= 10 nT, IMF
B
Z
= 10 nT, Dst = 0. All three field lines exhibit typical W shapes. Curves labeled dawn and dusk
have their equatorial crossings at 10 MLT and 14 MLT, respectively. All three field lines cross the equa-
torial plane at the distance 10.1 R
E
.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
2 of 17
[8] At the same time, it was pointed out (but not analyzed)
by ztrk and Wolf [2007] that if the magnetic field has a
dawndusk asymmetry, one can expect a change in second
invariant simply due to the asymmetry in the magnetic field,
and that the effect may be of zeroth order in gyroradius and
independent of energy and bounce phase. It is the purpose of
this paper to extend the analysis to this case and characterize
and quantify the second invariant changes of particles under-
going DSB in realistic magnetic fields. One obvious reason
for such an asymmetry is penetration of the y component
(in the GSM coordinates) of the interplanetary magnetic field
into the magnetosphere.
2. Assumptions
[9] In this work, we assume that the external magnetic
field is given by the semiempirical magnetic field model
T02 [Tsyganenko, 2002a, 2002b] parameterized by the IMF
B
Y
and B
Z
values, solar wind dynamic pressure P
DYN
, Dst
index, and the value of the dipole tilt angle. We further
assume that the tilt angle is zero, and that the intrinsic
magnetic field of Earth is a dipole with no tilt. Unless stated
otherwise, we use a magnetic field with IMF B
Z
= 5 nT,
P
DYN
= 2 nPa, and Dst = 0. Constancy of magnetic field in
time allows us to separate the effects of nonzero B
Y
from the
radial diffusion due to field fluctuations. Electric field is also
assumed to be zero since it is not essential for this analysis.
One prerequisite of the study is that the first invariant m is
conserved. This places a restriction on the gyroradius, thus
excluding particles with very high energies. We treat elec-
trons only; for protons, the results will be the same. However,
the following facts should be considered: (1) they gradient
and curvature drift around the Earth in the opposite direc-
tion; (2) they have larger gyroradii, which leads to a stricter
restriction on the conservation of the first adiabatic invariant;
thus, protons subject to the investigation of the DSBeffect are
supposed to have lower energy ranges than electrons.
3. The nature of the Second Invariant Changes
in the Case of B
Y
0
[10] Magnetic field strength curves such as the dashed and
dashdotted ones in Figure 1 become asymmetric with
respect to their equatorial crossing points if B
Y
0. For
instance, if IMF B
Y
< 0 (directed dusk to dawn), then on the
dawn side, the local minimum field strength in the Northern
Hemisphere B
minN
is smaller than its counterpart B
minS
in
the Southern Hemisphere; on the dusk side, B
minN
> B
minS
.
The relations between the two local minima are just the
opposite for positive (dawntodusk) IMF B
Y
conditions.
[11] Figure 2 shows a sample calculation of profiles of the
magnetic field strength along an electron trajectory across
the bifurcation region. Shown are 5 points in time as the
electron drifts from dawn across noon onto the afternoon
local time sector under negative IMF B
Y
condition. When
the local maximum value (near the equatorial plane) of a
given magnetic field line exceeds the mirror field strength
B
m
, which is conserved along the path, let us define two
quantities
I
n
B
m

_
sn2
sn1

1
B s
B
m

ds 4
Figure 2. A sequence of snapshots of profiles of magnetic field strength along field lines at 5 points
along the trajectory (time goes left to right) of an electron drifting through the bifurcation region. Parti-
cles parameters are B
m
= 55 nT, initial second invariant I
i
= 3 R
E
. IMF B
Y
= 5 nT for this case.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
3 of 17
and
I
s
B
m

_
ss2
ss1

1
B s
B
m

ds; 5
representing the second adiabatic invariants in the Northern
and Southern hemispheres, respectively. Here s
n1
and s
n2
correspond to the mirror points in the Northern Hemisphere
while s
s1
and s
s2
are in the Southern Hemisphere. A smaller
B(s) contributes more to the integral for a given mirror field
B
m
, which should of course be larger than the local mini-
mum and smaller than the equatorial maximum to make the
integral meaningful. So on the dawn side, I
n
> I
s
for the
same field line, while on the dusk side, I
n
< I
s
. These rela-
tions could also be illustrated by the contour plots of I
n
and
I
s
projected to the equatorial plane (Figure 3). For negative
IMF B
Y
, I
n
has larger values on the dawn side while I
s
is
larger on the dusk side.
[12] In our study trapped energetic electrons drifting
around the Earth keep their second adiabatic invariants
conserved except when they come to a bifurcation field line
near the dayside magnetopause, where the equatorial maxi-
mum just exceeds the mirror field strength, and get trapped
into one of the hemispheres. It will cause the sudden decrease
of the second adiabatic invariant. Conserving I
n
or I
s
and
drifting past the local noon, the electrons will face the other
bifurcation field line and escape the confinement with a
sudden increase of the second adiabatic invariant. But due
to the existence of nonzero IMF B
Y
, the magnitudes of the
sudden changes will be different, which explains the net
change of the total second adiabatic invariant after drifting
across the bifurcation region. The change turns out to be of
the same order of the initial second adiabatic invariant, as we
will show below.
4. Guiding Center Tracing Results
[13] Since solving (1) numerically is computationally very
expensive and is susceptible to numerical errors, a simpler
approach to analyzing effects of DSB on the second invariant
changes in the case of nonzero B
Y
is to trace particle trajec-
tories numerically by solving a version of the guiding center
equations [e.g., Northrop, 1963; Walt, 1994; Brizard and
Chan, 1999; Tao et al., 2007]. Of course the conservation
of the first adiabatic invariant should first be ensured, which is
the case in our study. For reasons explained later in section 7,
we chose to solve the form of equations derived [Brizard and
Chan, 1999] from the Hamiltonian formalism,
dR
dt
v
k
B
*
B
*
k


qB
*
k
b rB; 6
dv
k
dt

2
mB
*
k
B
*
rB; 7
where B
*
= B+ gmv
||
r b / q, B
||
*
= B
*
b = B + gmv
||
b (r
b) / q, b is the unit vector of the magnetic field, v
||
is the
velocity component parallel to B, and R is the position of the
particles gyrocenter.
[14] In Figure 4, we show four qualitatively different
types of solutions that can be obtained by numerically
Figure 3. Contour plots of I
n
and I
s
(equations (4) and (5)) in the dayside equatorial plane under 15 nT
IMF B
Y
. The mirror field strength is 55 nT.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
4 of 17
Figure 4. Four cases of guiding center trajectories in the magnetic field with IMF B
Y
= 5 nT illustrating
four types of drift shell bifurcation (the differences are initial conditions as listed in Table 1). Electrons are
launched from the equatorial plane at 6 MLT with initial velocity V and initial pitch angle a. I
I
and I
F
are
initial and final second invariants, respectively. (left) Trajectories. (right) Second invariants along trajectories.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
5 of 17
integrating equations (6)(7) for the same magnetic field
using the embedded CashKarp RungeKutta integration
scheme [Press et al., 1992]. We mostly choose the relative
accuracy of 10
6
in the adaptive stepper. The associated
variations of the second invariant I differ due to different
initial conditions, which are listed in Table 1.
[15] The two obvious cases for the second invariant
changes are shown in (b) and (c), where the electron un-
dergoes bifurcation twice, on both sides of the local noon,
and becomes trapped in the southern (b) or northern (c)
hemisphere. During the electrons drift motion, its second
adiabatic invariant is conserved except near the bifurcation
field lines. Consider Figure 2 (obtained from a case similar
to the one of (b)). The electron with initial second adiabatic
invariant I
n1
+ I
s1
gets trapped into the Southern Hemi-
sphere at the dawn side bifurcation line. It then drifts across
the bifurcation region, keeping its invariant I
s1
, until it
comes to the dusk side bifurcation line and returns to the
Northern Hemisphere with second adiabatic invariant I
n2
+
I
s2
. Since I
s1
= I
s2
and I
n1
> I
n2
, the total second adiabatic
invariant will be changed by the amount of
DI I
n2
I
n1
: 8
Likewise, an electron trapped in the Northern Hemisphere
will experience a change in the second invariant by the
amount of
DI I
s2
I
s1
: 9
There are two more, less obvious cases, when the change of
the second invariant results from bifurcating only once ((a)
and (d)). These are two special cases when the electrons
initial second adiabatic invariant is small. The electron will
experience only one bifurcation point in the bifurcation
region and get its second adiabatic invariant changed only
once. For an electron with small initial second adiabatic
invariant (Figure 5), when it drifts to the location where the
magnetic field profile first becomes W shape, the local
maximum field magnitude is already larger than the mirror
field value (so there is no bifurcation point on the dawn
side) and the electron is trapped around the field minimum
in the Northern Hemisphere. It will thus only experience one
bifurcation point on the dusk side, increasing its second
adiabatic invariant by the amount of
DI I
s2
: 10
The other special case (Figure 6) happens when the electron
bifurcates on the dawn side with a small second invariant
I
s1
. If there is no corresponding bifurcation point (because
when I
s2
= I
s1
the local maximum field strength is larger
than the mirror field magnitude) on the dusk side, the electron
Table 1. Four Cases for Which Guiding Center Calculations are
Presented in Figure 4
a
Case Y
GSM
(R
E
) V(c) a () I
I
(R
E
) I
F
(R
E
) DI (R
E
) B
m
(nT) E (keV)
(a) 7.5 0.6 75.4 0.23 1.04 0.81 60.0 127.7
(b) 8.5 0.5 49.0 2.10 1.68 0.42 65.0 79.1
(c) 8.0 0.3 60.0 1.03 2.07 1.04 60.5 24.7
(d) 8.0 0.5 60.5 0.99 0.20 0.79 60.0 79.1
a
Electrons are launched from the equatorial plane at 6 MLT with initial
radius |Y
GSM
|, initial velocity V and initial pitch angle a. I
I
and I
F
are initial
and final second invariants, respectively. DI is the change of the second
invariant after drift shell bifurcation. B
m
and E are the mirror field
magnitude and energy, respectively.
Figure 5. Snapshots of profiles of magnetic field strength along field lines for the case with a bifurcation
point only on the dusk side. B
m
= 55 nT, initial second invariant I
i
= 0.3 R
E
. IMF B
Y
= 5 nT.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
6 of 17
will drift out of the bifurcation region with its second adia-
batic invariant changed by the amount of
DI I
n1
: 11
5. PhaseSpace Diagrams
[16] Another useful way to look at the dynamics of the
particle undergoing drift shell bifurcation and nonadiabatic
change of its second invariant is to invoke phasespace dia-
grams, or pq diagrams, widely used in nonlinear mechanics.
In Figure 7, we present pq diagrams for the case (c) (Figure 4
and Table 1). Closed curves represent quasiperiodic motion.
The area enclosed by each curve is proportional to the second
invariant I. It is easy to see how the character of motion
changes at the two bifurcation points and that the final area
enclosed in the curve after the particle drifts around the Earth
and comes back to the starting point has changed.
6. Analysis Based on Magnetic Field Geometry
[17] Numerically solving guiding center equations allows
for detailed analysis of particle dynamics. However, as was
pointed out by ztrk and Wolf [2007], under some addi-
tional assumptions listed in section 2, changes in the second
invariant DI can be estimated for a particle of given mirror
field strength B
m
and initial second invariant I by consid-
ering properties of the magnetic field only. The results then
can be compared with sample calculations obtained from
guiding center tracing to ensure consistency. In this section,
we describe the procedure for evaluating the second invariant
changes for any values of B
m
and I (defining drift shells) of
particles entering the bifurcation region and present our main
results. An obvious advantage of this procedure is its ease of
use, as no trajectory tracing is necessary.
6.1. Procedure to Compute the Second Invariant
Changes
[18] Before we continue, we want to clarify the definitions
of three terms that are used throughout the paper. The
bifurcation region is the area near the dayside magnetopause
where magnetic field strength along the field line has a local
maximum near the equator. This also implies that a local
minimum exists on each side of the equator. The bifurcation
field line is one on the particles drift path where the near
equator magnetic field strength B
peak
just equals the mirror
field value B
m
. The bifurcation point is defined as its equa-
torial crossing point. To sort out the bifurcation points, we
need to define two sets of contours. In Figure 8, we plot a
set of B
peak
contours C(B
peak
) at the equatorial crossings of
the magnetic field lines for 5 nT IMF B
Y
. As particles drift
closer to local noon, B
peak
exceeds B
m
, which leads to parti-
cles being trapped into one of the hemispheres. In addition,
we need contours C(I(B
peak
)), where B
m
in the definition
(3) of the total second invariant I is replaced by the local
maximumfield B
peak
. Figure 9 illustrates the I(B
peak
) contours
under the same IMF conditions. It is noteworthy that the
intersection of C(B
peak
) and C(I(B
peak
)) (Figure 10) only
defines the bifurcation point when the particle enters the
bifurcation region. At this critical point, the particle decides
whether to continue its drift in the northern or Southern
Hemisphere. After that, its second invariant is conserved until
it leaves the bifurcation region. I(B
peak
) should be replaced by
I
n
(B
peak
) or I
s
(B
peak
) when searching for the other bifurcation
point, depending on the hemisphere it was trapped in. It is
Figure 6. Snapshots of profiles of magnetic field strength along field lines for the case with a bifurcation
point only on the dawn side. B
m
= 55 nT, initial second invariant I
i
= 1.2 R
E
. IMF B
Y
= 5 nT.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
7 of 17
Figure 7. Electron phase space trajectories in one bounce period between mirror points at different mag-
netic local times along the drift trajectory in Figure 4c. Here p is the (canonical) parallel momentum, and
conjugate coordinate q is the distance along the magnetic field line. The area inside the curve is pro-
portional to the second invariant I. Particle traverses from left to right and top to bottom.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
8 of 17
then straightforward to evaluate the second invariant changes
based on the values at the corresponding bifurcation points.
6.2. Results
[19] When we apply the procedure outlined in section 6.1
for a number of values of the mirror field and the second
invariant, we obtain a set of curves shown in Figure 11. The
Figure 11 (top) is for our nominal magnetic field configu-
ration with IMF B
Y
= 5 nT. In general, for each I value
there are two possibilities for DI, corresponding to trapping in
the Northern Hemisphere (solid lines) or Southern Hemi-
sphere (dashed lines) depending on the initial conditions.
The most important conclusion that one should make from
Figure 11 is that typical changes in I are of the same order of
magnitude as the initial I, thus the effect is significant. To
make sure that this effect is not spurious and specific to this
particular value of IMF B
Y
, we present results in the other two
panels of Figure 11 for the cases of 10 and 15 nT IMF B
Y
.
These results indicate that the changes of the second invariant
are large for all the nonzero B
Y
cases we study.
6.3. Dependence on IMF B
Z
, Dst, and P
DYN
[20] To estimate how sensitive this effect is to different
Tsyganenko T02 model configurations, we set the default
values as solar wind dynamic pressure P
DYN
= 2 nPa, Dst = 0,
IMF B
Y
= 5 nT, and B
Z
= 5 nT (Figure 11, top) and change
only one parameter at a time. The effect of varying values of
IMF B
Y
was already shown in the previous subsection.
Generally, stronger negative B
Y
causes the absolute value of
DI to increase, but it is no longer true when B
m
becomes large.
Figure 8. Contours of the nearequator maximum mag-
netic field strength B
peak
under 5 nT IMF B
Y
.
Figure 9. Contour plot of I(B
peak
) under the same IMF con-
ditions as Figure 8.
Figure 10. Contour plots of B
peak
and I(B
peak
). The inter-
sections define the bifurcation points where particles enter
the bifurcation region.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
9 of 17
[21] At a fixed B
m
level, the absolute value of DI is larger
when P
DYN
(Figure 13, right) is larger but decreases when
southward (negative) IMF B
Z
(Figure 12) is stronger, while
Dst (Figure 13, left) only has a small influence on the
change of the second invariant. In these plots, B
m
values are
chosen only when drift shell bifurcation can happen at those
levels.
6.4. Consistency With Guiding Center Tracing Results
[22] Having at our disposal the program to trace guiding
center trajectories as described in section 4, we compare
results fromthe two approaches in Figure 14. Superimposed on
the curves obtained with the procedure outline in section 6.1
are four crosses that represent solutions obtained by solving
the guiding center equations from section 4. It can be seen
that, for the four cases of guiding center calculations, the
results are in excellent agreement with calculations in this
section. To get points from solving (6)(7) that cover a large
range of mirror fields and second invariants requires a trial
and error approach in choosing the initial conditions; there-
fore, we only list four representative cases, although in our
experience the agreement is always good.
7. Discussion
[23] Before summarizing the results, there are several points
that need to be addressed and are specific to this particular
situation.
7.1. Guiding Center Equations
[24] First, we would like to comment on the choice of the
guiding center approximation equations used in section 4. It is
known [e.g., Walt, 1994; ztrk and Wolf, 2007; Kim et al.,
2008] that the standard textbook guiding center approxi-
mation equations derived by Northrop [1963],
dR
dt
v
k
B
B


qB
2
B rB
mv
2
k
qB
2
B k; 12
dv
k
dt

2
mB
B rB; 13
where m = g
2
mv
?
2
/ (2B)(m / g is the relativistic magnetic
moment) and k = (b r)b, are not sufficiently accurate when
numerically tracing particles. Specifically, Walt [1994] states
that the second equation, (13), needs to be more accurate
for numerical orbit tracing, while ztrk and Wolf [2007] and
Kim et al. [2008] choose (6)(7) because these equations
include higherorder terms and thus are supposed to be more
accurate. Brizard and Chan [1999] [also see Tao et al., 2007]
state that equations (6)(7), derived from the Hamiltonian
approach [Littlejohn, 1983; see also review by Cary and
Brizard, 2009, and references therein], can be shown to
contain the original Northrop equations (12)(13) when
Figure 11. Changes in the second invariant for a number of
levels of B
m
(its value being labeled by the number at the
right end of each curve; in unit nT) and I computed for
(top) IMF B
Y
= 5 nT, (middle) IMF B
Y
= 10 nT, and (bot-
tom) IMF B
Y
= 15 nT. Solid lines, particle trapped in the
Northern Hemisphere (also labeled by N); dashed lines, in
the Southern Hemisphere (also labeled by S).
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
10 of 17
neglecting phasespace volumepreserving terms (such as the
difference between B
||
*
and B). Here we attempt to clarify the
situation.
[25] When numerically integrating the standard Northrop
equations with the Tsyganenko 02 model, we experienced
the nonconservation of the second invariant (even when
the electron was not at the bifurcation lines), which might
be expected if the equations are not sufficiently accurate.
However, we also found nonconservation of the mirror field
B
m
(as shown in Figure 15), which is an indication of the
nonconservation of energy when integrating the equations. It
should be noted that our numerical integration was carried
out with very high accuracy to minimize numerical errors.
Nonconservation of B
m
is more troublesome and makes
equations (12)(13) unsuitable for this study. By comparison,
here we show explicitly the two extra terms contained in the
second BrizardChan equation and how the problem could
be solved by adding these terms to the standard Northrop
equations.
[26] If we take (6)(7) and expand B
*
and B
||
*
in powers
of " = m/q (proportional to the gyroradius), equation (6)
becomes
dR
dt
v
k
B
*
B
*
k


qB
*
k
b rB
v
k
B
mv
k
q
r b
B
mv
k
q
b r b


q
1
B
mv
k
q
b r b
b rB

v
k
B
B
mv
k
q
r b
_ _
1
mv
k
qB
b r b O "
2
_ _
_ _


qB
1
mv
jj
qB
b r b O "
2
_ _
_ _
b rB

v
k
B
B
mv
2
k
qB
r b
v
k
B
B
mv
k
qB
b
r b

qB
b rB O "
2
_ _

v
k
B
B
mv
2
k
qB
r b b b r b

qB
b rB O "
2
_ _
:
It is equivalent to equation (12) after omitting O("
2
) terms
because
b k b b r b b r b b
r b b b r b :
Figure 12. Changes in the second invariant for a number of
levels of B
m
(its value being labeled by the number at the
right end of each curve; in unit nT) and I computed for
(top) IMF B
Z
= +5 nT, (middle) IMF B
Z
= 10 nT, and (bot-
tom) IMF B
Z
= 15 nT. Solid lines, particle trapped in the
Northern Hemisphere (also labeled by N); dashed lines,
in the Southern Hemisphere (also labeled by S). IMF
B
Y
= 5 nT for all three cases.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
11 of 17
Equation (7) could be rewritten as
dv
k
dt

2
mB
*
k
B
*
rB

2
m
1
B
mv
k
q
b r b
B
mv
k
q
r b
_ _
rB

2
mB
1
mv
k
qB
b r b O "
2
_ _
_ _
B
mv
k
q
r b
_ _
rB

2
mB
B rB
mv
k
q
r b rB
_

mv
k
q
b r b b rB
_
O "
2
_ _
:
It contains two more terms of the order of " when com-
pared to equation (13). These two extra terms could be
neglected when the magnetic field is curl free, such as a
dipole or doubledipole magnetic field. When r B = r
(Bb) = Br b + rB b = 0, r b = (rB b) / B, so both
terms vanish. Numerically, we did not encounter the problem
of energy nonconservation when tracing particles under
dipole or doubledipole models. However, for Tsyganenko
magnetic fields, the external magnetic field induced by dif-
ferent current systems breaks the curlfree condition. The
absence of these two terms leads to unphysical results.
Adding them to the standard Northrop equations, we got
the same simulation results (Figure 16) as those of the
BrizardChan equations, which were already shown pre-
viously (Figure 4c).
[27] It could also be shown that the BrizardChan equa-
tions are equivalent to the guiding center equations extended
to higher order by Northrop and Rome [1978] (specifically
equations (3) and (5) in that paper) if we neglect the relativ-
Figure 13. Changes in the second invariant for a number of levels of B
m
(its value being labeled by the num-
ber at the right end of each curve; in unit nT) and I: (left) for two levels of Dst, 25 nT and 50 nT; (right) for
two values of P
DYN
of 5 and 8 nPa. Solid lines, particle trapped in the Northern Hemisphere (also labeled by
N), dashed lines, in the Southern Hemisphere (also labeled by S). IMF B
Y
= 5 nT for all cases.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
12 of 17
istic effect and replace m with M = M
0
+ "M
1
. However, the
evaluation of the latter requires the detailed information of a
particles initial position and velocity, while initial guiding
center position, velocity magnitude, and pitch angle are suf-
ficient for solving the BrizardChan equations.
7.2. Validity of Guiding Center Approximation
[28] The other aspect of numerically integrating guiding
center equations under the conditions of nonzero B
Y
that
should be pointed out is the allowed values of initial velocities
(or total energy). The results presented in this paper do not
depend on the total energy; thus one might be tempted to
choose higher values of initial velocity to speed up integration
of (6)(7), which, with the appropriate accuracy, are com-
putationally slow. It could be argued that as long as the
gyroradius of the particle is small with respect to the field
variation length scale, the guiding center approximation is
applicable. More specifically, the length scale of magnetic
field change experienced by a charged particle during the
gyration period should be much larger than its gyration
radius, which also implies the conservation of the first adia-
batic invariant. The often used criterion [Sergeev et al., 1983;
Figure 14. Comparison of results obtained with geometry calculations (lines) and four cases of guiding
center drift trajectory solutions (four crosses). The curves are the same as in Figure 11 (top), while the
crosses are from the four cases in Table 1 and Figure 4.
Figure 15. Results of tracing an electron with initial position (0, 8, 0) R
E
in GSM coordinates, velocity
0.3c, pitch angle 60 under Tsyganenko 02 model using the standard Northrop equations (12)(13). (left)
The second invariant as a function of time along the electrons trajectory. (right) The magnetic field
strength the electron experiences along its trajectory, with the top envelope corresponding to the mirror
field magnitude.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
13 of 17
Bchner and Zelenyi, 1989; Delcourt et al., 1996; Wolf et al.,
2009] is

R
c;min
a
c; max
) 1
where R
c,min
is the minimum curvature radius of the magnetic
field line and a
c,max
is the maximumgyration radius along the
field line. The critical values of for the conservation of the
first adiabatic invariant suggested by Sergeev et al. [1983]
and Delcourt et al. [1996] are

8
p
and 3, respectively. There
are also other ways to measure the nonadiabatic scattering of
the first adiabatic invariant, such as discussed for ions by
Anderson et al. [1997].
[29] In our study of the drift shell bifurcation effect, the
existence of nonzero interplanetary magnetic fields decreases
the curvature radius of magnetic field lines. We could not use
too energetic particles, whose gyration radii become com-
parable to the curvature radii, especially near the plasma
sheet on the night side, to carry out guiding center tracing.
This was somewhat unexpected as this type of scattering of
the first invariant is usually associated with highly stretched
magnetotailplasma sheet field lines and not the inner mag-
netosphere (L < 8). Yet too energetic particles should not be
traced through guiding center equations, since the assumption
for guiding center equations is violated.
[30] We tested fullparticle tracing for different particle
velocities by integrating the relativistic NewtonLorentz
equation (1) under the same solar wind and IMF condi-
tions as in previous guiding center numerical simulations.
Figure 17 (top) shows the magnetic field magnitude along the
0.99c electrons trajectory and the corresponding values.
Instead of showing the value for each magnetic field line,
here we calculate the instant value (square root of the ratio
of the curvature radius to the gyration radius) for each traced
point along the electrons trajectory. At the bottom of
Figure 17, we show solutions of (1) but now for an electron
with 0.9c velocity magnitude, all other conditions being the
same. Comparison of the two cases suggests a critical
value of 9 to ensure good conservation of the first adiabatic
invariant, as implied by the conservation of the mirror field
magnitude (refer to the definition of the first adiabatic invari-
ant). We use this value as our guideline for selecting electrons
and verifying the validity of guiding center simulation.
[31] We are especially careful on this topic in our study
because of the existence of nonzero interplanetary magnetic
fields. For magnetic fields with eastwest and northsouth
symmetries, much more energetic particles could be used in
guiding center tracing.
7.3. Probability of Being Trapped in a Given
Hemisphere
[32] It would be of interest to estimate the average chan-
ges of the second invariants during drift shell bifurcation
under nonzero IMF B
Y
, which requires knowledge of the
probability of the particle becoming trapped in a given
hemisphere for averaging purposes. The change of the
second invariant (for given initial I and B
m
) depends on the
hemisphere the particle is trapped in, which is determined by
the bounce phase of the particle on the bifurcation field line.
Shabansky [1971] hypothesized that the probability of the
particle entering the northern or Southern Hemisphere should
be proportional to the corresponding particle bounce period
between mirror points in the vicinity of the bifurcation field
line, when averaged over uniform distributions of the bounce
phases. The bounce period can be expressed as

2
v
_
sm2
sm1
1
B s
B
m
_ _
1=2
ds:
[33] The integrand approaches infinity at s
m1
and s
m2
. By
Taylor expansion, the integral near the mirror point becomes
integrable,
2
v
_
sm1"
sm1
1 1
B
0
s
m1

B
m
s s
m1

_ _ _ _1=2
ds
2
v
_
"
0

B
m
B
0
s
m1

ds

s
p
2
v

B
m
B
0
s
m
1

2

"
p
;
where B(s
m1
) < 0. We would get a similar result at mirror
point s
m2
. However, when this mirror point is the local
maximum near the equatorial plane, the first derivative will
Figure 16. Results of tracing the same electron as in 15 using the Northrop equations (12)(13) but with
the two extra terms, which are the same as the simulation results (Figure 4c) obtained by solving the
BrizardChan equations.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
14 of 17
become zero. Adding the second derivative term in the
Taylor expansion yields
2
v
_
sm2
sm2"
1 1
B
0 0
s
m2

B
m
s s
m2

2
2
_ _ _ _
1=2
ds
2
v
_
"
0

2B
m
B
0 0
s
m2

ds
s
;
which is an improper integral.
[34] We designed an ensemble of particles similar to the
way it was done by ztrk and Wolf [2007]. Over 100 par-
ticles located between mirror points along a magnetic field
line were launched from local midnight with both upward
and downward velocities and traced through guiding center
equations. The pitch angles of these particles should satisfy
B(x
i
, y
i
, z
i
) = B
m
sin
2
a
i
to ensure they stay on the same drift
shell. Under the several different cases we investigated, we
did not find an acceptable criterion for the ratio of particles
going into the Northern and Southern hemispheres; we have
to leave this question open.
7.4. DiffusionType Coefficient Associated With Drift
Shell Bifurcation
[35] The changes in the second adiabatic invariant eval-
uated in this work clearly have implications for modeling of
radiation belts dynamics. Although it is not straightforward
to translate these calculations into more familiar and more
commonly used forms such as diffusion coefficients [e.g.,
Albert, 2009; Summers, 2005], we provide a simpleminded
estimate of average changes, as follows. We would like to
consider an ensemble of electrons with different pitch angles
Figure 17. (top) Plots of (left) magnetic field magnitude and (right) the corresponding values of k along
the full particle trajectory. The electron is launched with initial position (8, 0, 0) R
E
in GSM coordinates,
initial velocity 0.99c, initial pitch angle 45. (bottom) Same as top, but electrons initial velocity is
reduced to 0.9c. The solar wind and IMF conditions are the same as in previous numerical simulations.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
15 of 17
and different initial positions in space, and calculate the
average changes hDIi, h(DI)
2
i, etc. (their expressions and
values described below), over this ensemble, for the nominal
case of IMF B
Y
= 5 nT.
[36] We place an ensemble of 418 electrons on a surface
located on the dawn side and obtained by interpolating from
magnetic field lines with equatorial crossing distance rang-
ing from 4 to 10 R
E
. Then we adopt the following approach
to ensure that it is equivalent to placing a uniform density of
electrons per unit area on the surface: we prescribe a mesh
of uniformly distributed points in the X
GSM
= 0 plane on the
dawn side and assign each projected point on the surface with
an electron and a weight factor proportional to the surface
area enclosed by its four nearest neighbors; the weight factor
w
k
(k is the electron number) is later included in the expres-
sions of the ensemble averages.
[37] To set up the ensemble, we place a 341 keV electron
with a randomvalue of cosabetween 1 and 1 at each point on
the surface. Here a is the pitch angle; cosa is uniformly dis-
tributed over the range [1, 1]. The initial gradient/curvature
drift velocity perpendicular to the surface V
?gc,k
is also
calculated.
[38] The ensemble of electrons with initial positions and
pitch angles prescribed above are started from the surface on
the dawn side and traced using the Brizard and Chan [1999]
guiding center equations. Out of them, 66 electrons expe-
rience drift shell bifurcation, being trapped either into the
Northern Hemisphere or into the Southern Hemisphere on the
dayside. Their second invariant changes rI
k
and drift periods
t
k
are recorded. Electrons that drift through the magnetopause
or fail to undergo drift shell bifurcation are eliminated from
the calculation. Our results are
DI h i

k
w
k
DI
k
V
?gc;k

k
w
k
V
?gc;k
1:0R
E
;
DI
2
_ _

k
w
k
DI
k

2
V
?gc;k

k
w
k
V
?gc;k
3:2R
2
E
;
DI

_ _

k
w
k
DI
k
=
k
V
?gc;k

k
w
k
V
?gc;k
0:001R
E
=s;
DI
2
2
_ _

k
w
k
DI
k

2
=2
k
_ _
V
?gc;k

k
w
k
V
?gc;k
0:002R
2
E
=s;
where the last quantity can be considered a diffusiontype
coefficient. The energy dependence in the last two expres-
sions should be that of the gradient/curvature drift velocity,
which is proportional to gv
2
(g and v are the relativistic
factor and the total velocity, respectively). So a factor of
(W
K
/ 273 keV)(1 + W
K
/ 1022 keV) / (1 + W
K
/ 511 keV)
should be multiplied to the righthand side for different
energies. Here W
K
is the electron kinetic energy. For elec-
trons of the energy considered in this calculation, the ratio of
the diffusiontype coefficient to the one (0.411 r
m
2
(G
dawn
2
+
G
dust
2
) / t
d
; equation(49)) given by ztrk and Wolf [2007]
and evaluated for typical parameters suggested in that paper
is 2 10
4
, showing that the impact due to the field asymmetry
is likely to dominate.
8. Conclusions
[39] We presented analysis of the second adiabatic invari-
ant changes of trapped energetic particles in the magneto-
sphere due to drift shell bifurcation near the magnetopause for
the case of magnetic fields lacking eastwest and northsouth
symmetry. For the magnetic field configurations we studied,
electrons subject to the DSBeffect were in the energy range of
keV up to MeV, while protons were supposed to have lower
energies due to their larger gyroradii and possible violation
of the first adiabatic invariant conservation. We found that
typical changes in the second invariant after a particle crosses
the dayside bifurcation region are comparable to the initial
value of the invariant, and we demonstrated the result in two
ways, by solving guiding center equations to trace particle
trajectories and also by analyzing the magnetic field only
(both results agree with each other). We studied dependence
of the effect on the parameters controlling the Tsyganenko
T02 magnetic field model and found that, for a given value of
the mirror field, the absolute value of the second invariant
change DI tends to depend on IMF B
Y
and somewhat on B
Z
and is larger with increasing values of the solar wind dynamic
pressure; the dependence on Dst is rather weak. When par-
ticles come back to the same value of local time that they
started from, if they get a net increase in the second invariant,
they are at a larger radial distance and have a higher possi-
bility of getting lost to the magnetopause once they drift to the
dayside. These results have implications for modeling of
radiation belts in the inner magnetosphere, although we did
not parameterize the effect in terms of loss terms for typical
radiationtype calculations. In addition, we clarified why the
Brizard and Chan [1999] guiding center equations had an
advantage over the standard Northrop [1963] equations when
numerically tracing charged particles in realistic magneto-
spheric magnetic fields.
[40] Acknowledgments. This work was supported by the NASA
Geospace Theory grants NNG05GH93G and NNX08AI55G.
[41] Masaki Fujimoto thanks Xin Tao and another reviewer for their
assistance in evaluating this paper.
References
Albert, J. M. (2009), The coupling of quasilinear pitch angle and energy
diffusion, J. Atmos. Sol.Terr. Phys., 71(16), 16641668.
Anderson, B. J., R. B. Decker, N. P. Paschalidis, and T. Sarris (1997),
Onset of nonadiabatic particle motion in the nearEarth magnetotail,
J. Geophys. Res., 102(A8), 17,55317,569, doi:10.1029/97JA00798.
Brizard, A. J., and A. A. Chan (1999), Nonlinear relativistic gyrokinetic
VlasovMaxwell equations, Phys. Plasmas, 6(12), 45484558.
Bchner, J., and L. M. Zelenyi (1989), Regular and chaotic charged particle
motion in magnetotaillike field reversals: 1. Basic theory of trapped
motion, J. Geophys. Res., 94(A9), 11,82111,842, doi:10.1029/
JA094iA09p11821.
Cary, J. R., and A. J. Brizard (2009), Hamiltonian theory of guiding center
motion, Rev. Mod. Phys., 81(2), 693738, doi:10.1103/RevMod-
Phys.81.693.
Delcourt, D. C., and J.A. Sauvaud (1999), Populating of the cusp and
boundary layers by energetic (hundreds of keV) equatorial particles,
J. Geophys. Res., 104(A10), 22,63522,648, doi:10.1029/1999JA900251.
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
16 of 17
Delcourt, D. C., J.A. Sauvaud, R. F. Martin Jr., and T. E. Moore (1996),
On the nonadiabatic precipitation of ions from the nearEarth plasma
sheet, J. Geophys. Res., 101(A8), 17,40917,418, doi:10.1029/
96JA01006.
Kim, K. C., D. Y. Lee, H. J. Kim, L. R. Lyons, E. S. Lee, M. K. ztrk,
and C. R. Choi (2008), Numerical calculations of relativistic electron
drift loss effect, J. Geophys. Res., 113, A09212, doi:10.1029/
2007JA013011.
Littlejohn, R. G. (1983), Variational principles of guiding center motion,
J. Plasma Phys., 29, 111125.
Mead, G. D. (1964), Deformation of the geomagnetic field by the solar wind,
J. Geophys. Res., 69(7), 11811195, doi:10.1029/JZ069i007p01181.
Northrop, T. G. (1963), The Adiabatic Motion of Charged Particles, John
Wiley, New York.
Northrop, T. G., and J. A. Rome (1978), Extensions of guiding center
motion to higher order, Phys. Fluids, 21(3), 384389.
Northrop, T. G., and E. Teller (1960), Stability of the adiabatic motion of
charged particles in the Earths field, Phys. Rev., 117, 215225.
ztrk, M. K., and R. A. Wolf (2007), Bifurcation of drift shells near the
dayside magnetopause, J. Geophys. Res., 112, A07207, doi:10.1029/
2006JA012102.
Press, W. H., S. A. Teukolsky, W. T. Wetterling, and B. P. Flannery (1992),
Numerical Recipes in Fortran, Cambridge Univ. Press, New York.
Roederer, J. G. (1970), Dynamics of Geomagnetically Trapped Radiation,
Springer, Berlin.
Sergeev, V. A., E. M. Sazhina, N. A. Tsyganenko, J. A. Lundblad,
and F. Soraas (1983), Pitchangle scattering of energetic protons in
the magnetotail current sheet as the dominant source of their isotropic
precipitation into the nightside magnetosphere, Planet. Space Sci.,
31, 1237.
Shabansky, V. P. (1971), Some processes in magnetosphere, Space Sci.
Rev., 12(3), 299418.
Shabansky, V. P., and A. E. Antonova (1968), Topology of the drift shells
of particles in the magnetosphere, Geomag. Aeron., 8, 799802.
Summers, D. (2005), Quasilinear diffusion coefficients for fieldaligned
electromagnetic waves with applications to the magnetosphere, J. Geophys.
Res., 110, A08213, doi:10.1029/2005JA011159.
Tao, X., A. A. Chan, and A. J. Brizard (2007), Hamiltonian theory of
adiabatic motion of relativistic charged particles, Phys. Plasmas, 14(9),
092107, doi:10.1063/1.2773702.
Tsyganenko, N. A. (2002a), A model of the magnetosphere with a dawn
dusk asymmetry 1. Mathematical structure, J. Geophys. Res., 107(A8),
1179, doi:10.1029/2001JA000219.
Tsyganenko, N. A. (2002b), A model of the near magnetosphere with a
dawndusk asymmetry: 2. Parameterization and fitting to observations,
J. Geophys. Res., 107(A8), 1176, doi:10.1029/2001JA000220.
Walt, M. (1994), Introduction to Geomagnetically Trapped Radiation,
Cambridge Univ. Press, Cambridge, UK.
Wolf, R. A., Y. F. Wan, X. Xing, J. C. Zhang, and S. Sazykin (2009),
Entropy and plasma sheet transport, J. Geophys. Res., 114, A00D05,
doi:10.1029/2009JA014044.
M. K. ztrk, Department of Information Systems and Technologies,
Yeditepe University, 34755 Istanbul, Turkey.
S. Sazykin, Y. Wan, and R. A. Wolf, Physics and Astronomy
Department, Rice University, Houston, TX, 77005 USA. (yifei@rice.edu)
WAN ET AL.: DRIFT SHELL BIFURCATION IMF B
Y
A10205 A10205
17 of 17

You might also like