The Dynamics of Planetary Magnetospheres

Planetary and Space Science 49 (2001) 10051030
www.elsevier.com/locate/planspasci
The dynamics of planetary magnetospheres

C.T. Russell
Department of Earth and Space Sciences and Institute of Geophysics and Space Physics, University of California, Los Angeles, 3845 Slichter Hall,
MS 156704, Los Angeles, CA 90095, USA
Received 8 December 1999; received in revised form 9 October 2000; accepted 10 October 2000
Abstract
Mercury, Earth, Jupiter, Saturn, Uranus, Neptune, and the moon, Ganymede, have presently-active internal dynamos while Venus, Mars,
at least two of the Galilean moons, the Earths moon, comets and asteroids do not. These active dynamos produce magnetic 5elds that have
su6cient strength to stand o7 the pressure of the exterior plasma environment. Because of changes in these exterior plasma environments
these magnetospheres are very dynamic. The jovian magnetosphere includes a strong time-varying energy source that adds to the dynamics
of its magnetosphere and produces a quite di7erent circulation pattern than that found at Earth and, presumably, Mercury. Not only intrinsic
planetary magnetic 5elds produce magnetospheres but also unmagnetized planets. Venus, Mars and comets have induced magnetospheres
associated with the solar wind interaction with their atmospheres. Cometary magnetospheres, parts of which can be remotely sensed,
exhibit spectacular disruptions called tail disconnections. Even the atmosphereless bodies with weak magnetic 5elds can interact with the
solar wind. Small magnetic anomalies on the moon and possibly asteroids cause weak de>ections of the solar wind. The dynamics of these
c 2001 Elsevier Science Ltd. All rights reserved.
various magnetospheres provide a rich spectrum of behavior which we review herein.
1. Introduction
At 5rst sight the bodies of the solar system can be classi5ed into three types according to their plasma environments:
the atmosphereless bodies, with no intrinsic magnetic 5eld
such as the Earths moon; the bodies with atmospheres like
Venus and Mars but with a weak or absent intrinsic 5eld
where the interaction is governed by the interaction of the
plasma with the neutral atmosphere; and the bodies such as
Mercury, Earth and Jupiter, in which the intrinsic magnetic
5eld is so strong that it de>ects the external plasma at altitudes far above the surface. The magnetic 5elds of the bodies whose 5elds are derived from internal dynamos or, in
the case of the moon, from magnetized rocks are steady on
the time scales of space exploration but their plasma environments are not. Thus each of the solar system bodies exists in a very dynamical plasma and magnetic environment
because of changes external to the body.
We have su6ciently explored these bodies to be able to
recognize a continuum of behavior over the spectrum of
planetary interactions. Of course, we understand best the
behavior of just a few bodies: the terrestrial and jovian
Corresponding author. Tel. +1-310-825-3188; fax: +1-310-206-8042.

E-mail address: ctrussel@igpp.ucla.edu (C.T. Russell).
magnetospheres, the Venus and cometary magnetospheres,

and the lunar interaction. We stress herein the information
gleaned from our studies of these bodies. However, we attempt to be complete by including the brief glimpses we
have gained into the behavior of other magnetospheres as
well. We emphasize that our purpose is to compare the dynamical behavior of the solar system magnetospheres and
thus our discussion of the basic structure of these magnetospheres will be brief. By dynamical behavior we mean the
circulation and motion of the magnetized plasma on 5eld
lines that thread the body and especially the sudden changes
in that circulation and plasma content.
We begin our review with the Earth because we understand best the behavior of the Earth. We then examine the
solar wind interaction with the Moon and Mercury, two basically atmosphereless bodies, and the magnetospheric interaction with Ganymede, an atmosphereless moon in Jupiters
magnetosphere. Next we examine induced magnetospheres,
including those of Venus and Mars and comets. Finally, we
close with a discussion of the jovian magnetosphere and
with a few words about the magnetospheric dynamics of
the remaining outer planets. We stress that our goal is to
provide a tutorial overview of the dynamics of planetary
magnetospheres with an emphasis on the jovian magnetosphere to provide context for the other papers associated with
the 1999 symposium on the Magnetospheres of the Outer
c 2001 Elsevier Science Ltd. All rights reserved.

0032-0633/01/$ - see front matter
PII: S 0 0 3 2 - 0 6 3 3 ( 0 1 ) 0 0 0 1 7 - 4
1006
C.T. Russell / Planetary and Space Science 49 (2001) 10051030
Planets. This paper is not intended as a complete review of

the jovian, terrestrial or other literature.
2. The Earth
The Earth has a dipole magnetic moment of 8 1015 Tm3
that produces a magnetic 5eld strength at the equator on the
Earths surface of about 30; 000 nT, and at 10 Earth radii
(RE ) of about 30 nT. The solar wind interaction slightly more
than doubles this value on the dayside so that the pressure
in the magnetic 5eld is about 2 nPa. The sun emits a magnetized plasma consisting of mainly protons and electrons
with a density of about 7 cm3 at the orbit of the Earth (1
astronomical unit or AU) at a velocity of about 440 km s1 .
The pressure exerted by this >owing plasma is also about
2 nPa, thus balancing the pressure exerted by the magnetospheric 5eld.
Disturbances of three types propagate in this magnetized
solar wind plasma. The fast mode wave compresses the magnetic 5eld and plasma; the intermediate mode wave bends
the >ow and magnetic 5eld but does not compress it; and
the slow mode wave rare5es the 5eld while it compresses
the plasma and vice versa. The solar wind travels faster than
the propagation speed of all three of these waves so when
it reaches the Earths magnetosphere the pressure waves
needed to de>ect the solar wind plasma cannot propagate
upstream into the solar wind without creating a shock front.
The geometry of this shock, the de>ected >ow and the
magnetopause is shown in Fig. 1. The fastest wave is the
aptly named fast mode wave. It does the yeomans work in
slowing, de>ecting and heating the solar wind downstream
of the bow shock so that the plasma can >ow around the magnetosphere. Nevertheless it cannot cause all of the changes
in the plasma needed to move both the plasma and the magnetic 5eld around the bullet-shaped magnetosphere and the
intermediate and slow modes also play a role. The net result
of these standing waves is a >ow that bends to >ow parallel to the magnetopause, the boundary between the magnetosheath and the magnetosphere. The pressure normal to the
surface is transmitted by the thermal motions of the plasma
and by the magnetic 5eld.
2.1. The size of the magnetosphere
In order to determine the scale size of the magnetosphere
we need to understand the pressure applied to the magnetosphere by the solar wind. Fortunately, we do not have to
solve the complex non-linear solar wind interaction problem
to do so. We can obtain a quantitative formula for the distance from the center of the Earth to the magnetopause in a
straightforward manner. Conservation of the momentum in
a stream tube of varying cross-section, S, gives us
(u2 + nkT + B2 =20 )S = constant;
(1)
Fig. 1. The cross-section of the magnetosphere and the bow shock in

the plane of the incoming >ow to the stagnation point and the magnetic
5eld. The dashed lines show the >ow and how it is de>ected at the shock
and is bent around the magnetopause. The solid lines are magnetic 5eld
lines that also are straight until they encounter the shock at which point
they are bent around the magnetopause. The shaded region is the ion
foreshock where ions move back from the bow shock into the solar wind.
This backward re>ection can occur because of the inability of some ions
to penetrate the shock barrier and to re>ect back along the 5eld lines
and due to the thermal heating in the shock=magnetosheath that results
in some particles moving backward at greater than the incoming bulk
velocity. The geometry of this 5gure is patterned after early gas dynamic
simulations by Spreiter et al. (1966).
where ; u; n; T and B are the mass density, speed, number density, temperature of the solar wind and magnetic 5eld
strength, respectively. This formula allows us to use the incoming solar wind dynamic pressure, u2 , which dominates
over the thermal and magnetic pressures, in front of the bow
shock instead of having to calculate these pressures in the
magnetosheath downstream from the shock front, given that
we know the expansion of the cross-section of the stream
tube, S.
It is instructive to compare the size of the terms in (1).
The ratio of the 5rst two terms is
u2 =nkT = u2 =(kT=mi ) = u2 =cs2 = Ms2 ;
(2)
where is the polytropic index, cs is the sound velocity and

Ms is the sonic Mach number. The ratio of the second 5rst
and third terms is
2
= 2MA2 ;
u2 =(B2 =20 ) = 2u2 =(B2 =0 ) = 2u2 =vA
where MA is the Alfven or intermediate Mach number.
(3)
1007
Finally, the ratio of the second and third terms is

nkT=(B2 =20 ) = = 2(MA =Ms )2 = :
(4)
Since, as stated above, the solar wind >ows faster than any of
the three waves in the plasma and usually much faster, Mach
numbers are much greater than unity. Thus Eqs. (2) and (3)
tell us that the dynamic pressure dominates over the thermal
and magnetic pressures in front of the bow shock. Eq. (4) indicates that, when the magnetic pressure dominates (low ),
the speed of Alfven waves, vA , exceeds that of sound waves,
cs , and magnetic forces dominate in the plasma frame. The
intermediate wave propagates at the Alfven speed along the
magnetic 5eld. The fast mode wave propagates at a speed
2
2 2
2
0:707{cs2 + vA
+ [(cs2 + vA
) 4cs2 vA
cos2 ]1=2 }1=2 where
is the angle between the magnetic 5eld and the direction of
propagation of the phase fronts of the wave. Perpendicular to
2
the magnetic 5eld this speed is equal to (vA
+cs2 )1=2 . The slow
2
2
2
2 2
2
cos2 ]1=2 }1=2 .
speed is 0:707{cs + vA [(cs + vA ) 4cs2 vA
The fast mode is the only mode that can transmit energy
across a magnetic 5eld.
Returning to the question of the stando7 distance of the
nose of the magnetopause, we now know that we can approximate the solar wind pressure contribution by the momentum
>ux u2 diminished by a factor accounting for the expansion of the stream tube. This e7ect is small, roughly 10%.
The magnetospheric pressure is dominated by the pressure
in the magnetic 5eld. The pressure equals (aB0 =L3mp )2 where
a is a shape-dependent factor, equalling 2:4 for the shape
of the Earths magnetosphere, and Lmp being the distance
to the magnetopause from the center of the Earth. Equating
the pressure to the solar wind dynamic pressure we obtain
a stando7 distance.
2 1=6
Lmp = 107:4(nsw usw
)
;
(5)
where nsw is the solar wind proton number density in

cm3 ; usw is the proton bulk velocity in km=s, and Lmp is
the stando7 distance in RE .
2.2. Tangential stress
The dynamic pressure determines the overall size of the
magnetosphere and to zeroth order its shape but tangential
stresses also a7ect the shape and cause momentum transfer across the boundary. Several di7erent mechanisms have
been proposed for the source of tangential stress and momentum transfer to the magnetosphere. Fig. 2 illustrates a number of the popular mechanisms. The upper left-hand panel
illustrates di7usive entry. An ion enters the magnetospheric
magnetic 5eld and instead of returning to the magnetosheath
with the same velocity with which it started, it becomes scattered and drifts within the magnetosphere carrying with it
whatever momentum parallel to the boundary it had initially.
This process relies on scattering centers that seem not to
be present in the boundary layers inside the magnetosphere.
The lower left panel shows a variant of di7usive entry in
Fig. 2. Sources of viscosity at the magnetopause. Schematic illustrations

of di7usive entry, impulsive penetration, surface wave induced momentum
transfer and the KelvinHelmholtz instability.
which an entire tube of magnetosheath plasma crosses the

magnetopause into the magnetosphere. This mechanism is
not expected to be e7ective because as long as there is a 5nite
angle between the magnetosheath and magnetospheric 5elds
the tubes cannot penetrate one another. If they do become
aligned in some region, the three-dimensional geometry of
the interaction causes them to be at some signi5cantly large
angle not far away from the point of alignment. Thus the
two magnetized plasmas are kept separate.
The top right-hand panel illustrates momentum transfer
by wave processes. In a dissipative medium, such as the
ionosphere to which the magnetospheric 5eld lines are connected, the eddies formed by the passage of the surface
wave are increasingly smaller with distance from the boundary. The net result is a >ow in the magnetosphere parallel to the magnetosheath >ow. Finally, the lower left panel
shows the result of boundary-wave amplitude growth via
the KelvinHelmholtz instability when the magnetosheath
velocity passes an instability threshold. The boundary shape
becomes non-sinusoidal and the momentum transfer by the
process discussed above proceeds at even a greater rate.
This process takes place independently of whether the solar wind magnetic 5eld is parallel or antiparallel to that of
1008
Fig. 3. Reconnecting magnetospheres for southward interplanetary magnetic 5elds (top) and northward interplanetary magnetic 5elds (bottom)
(Dungey, 1961, 1963). The diagrams are not to scale nor are the details
of the solar wind interaction taken into account.
the magnetosphere, but the tangential stress on the magnetosphere clearly depends on the magnetic 5eld orientation.
There seems to be little momentum transfer that is independent of control by the north-south component of the interplanetary magnetic 5eld.
The mechanism by which the magnetized solar wind powers the magnetosphere was 5rst proposed by Dungey (1961,
1963) as sketched in Fig. 3. In the top panel the interplanetary magnetic 5eld is southward and becomes connected to
the terrestrial magnetic 5eld at the subsolar point in a process known as reconnection. The resulting V-shaped magnetic 5eld accelerates plasma (whose origin is both in the
solar wind and the magnetosphere) as the 5eld lines straightens. Then the magnetic 5eld slows the plasma as the 5eld
lines are stretched behind the terminator. Over the dayside
energy >ows from the magnetic 5eld into the plasma, but in
the tail there is a Poynting >ux of energy into the magnetic
5eld from the solar wind plasma. This process results in the
storage of energy in the magnetotail. This energy is in turn
tapped at a reconnection point in the tail that causes the >ow
of plasma into the magnetosphere proper and back down
the tail. The plasma and 5eld continue to move toward the
dayside reconnection point where the cycle repeats itself.
In this way the magnetospheric plasma can be made to circulate even in a dissipative system as energy is continually
supplied by the solar wind. Estimates of the rate of energy
input into the magnetosphere during active times range up
to about 2 TW (2 1012 W).
When the interplanetary magnetic 5eld is northward reconnection can still occur but it has a quite di7erent e7ect on
the magnetosphere. The panel on the bottom of Fig. 3 shows
this situation. The interplanetary magnetic 5eld now reconnects with the terrestrial magnetic 5eld above and behind
Fig. 4. A cut away diagram illustrating the three-dimensional magnetosphere, its plasma regions and current systems.
the poles. The reconnected 5eld line is added to the dayside

and a corresponding >ux tube is removed from the nightside and transported down the tail. E7ectively this transports
magnetic >ux from the nightside to the dayside magnetosphere. The reconnected dayside tube moves along the magnetopause boundary (in three dimensions out of the plane
of the page) and replaces the >ux tube that was lost down
the tail. This mechanism makes a boundary layer of plasma
within the magnetopause and maintains circulation of the
plasma for northward interplanetary magnetic 5eld. When
the magnetic 5eld is northward, but not due northward, this
process can still proceed, but the same >ux tube is unlikely
to reconnect at both ends. This results in plasma circulation
but no change in the magnetic >ux on open and closed magnetic 5eld lines (those with 1 and 2 feet in the ionosphere
respectively).
The tangential stress in the outer magnetosphere must
ultimately apply stress to the ionospheric plasma as the ionosphere is the ultimate site of dissipation in the magnetospheric system. Although there are several ways for energy
to be deposited in the ionosphere from the magnetosphere,
such as particle precipitation whereby energetic charged particles in the magnetosphere enter the atmosphere and are
lost, or as wave transport in which waves generated in the
magnetosphere propagate into the ionosphere and heat it,
the most signi5cant energy dissipation mechanism is joule
dissipation in electric currents. Fig. 4 illustrates where the
current systems in the magnetosphere >ow. When the forces
of the solar wind are directed along the magnetopause normal and there are no tangential stresses, the magnetopause
current and the tail current are dissipationless currents >owing on the surface of the magnetopause with j E = 0 where
j is the current density and E is the electric 5eld. When
there is a magnetic 5eld normal to the magnetopause these
currents can accelerate or decelerate the >ow in the dayside and tail regions respectively. The current crossing the
1009
plasma sheet, labelled neutral sheet current, is an extension

of the tail current system. It too can lead to plasma acceleration when there is a normal component of the magnetic 5eld across the tail current sheet. The magnitude of
the magnetopause current is such as to rotate the 5eld direction and increase its strength from that of the magnetosheath to that of the magnetosphere. The neutral sheet
current strength is of magnitude to reverse the direction of
the magnetic 5eld from the northern lobe of the tail to the
southern lobe.
2.3. The ring current
The ring current consists of the current due to the
eastward (electron) and westward (proton) drift in the radiation belts. Since this current does not pass through the
ionosphere, it too is basically dissipationless. The energy
content of the radiation belt is generally fairly constant except during periods known as geomagnetic storms. The ring
current causes a net decrease in the magnetic 5eld on the
surface of the Earth (Dessler and Parker, 1959; Sckopke,
1966) as opposed to the magnetopause current that causes
an increase. The energy of these circulating particles can be
easily calculated from their e7ect on the ground-level magnetic 5eld. In major magnetic storms this energy can reach
10 or more petaJoules (1015 J) and the energization rate can
exceed several teraWatts (1012 W), equaling and possibly
exceeding the energy dissipation in the auroral ionosphere.
The e7ect of the tail current system is to oppose the Earths
5eld and has a stronger e7ect on the nightside thus, causing
a daynight gradient in the 5eld due to external sources.
Thus a sudden compression of the size of the magnetosphere by an increase in the solar wind pressure will cause
a greater e7ect on the dayside of the magnetosphere than
on the nightside.
2.4. Magnetosphereionosphere coupling
The current system ultimately responsible for the majority
of the dissipation is labeled 5eld-aligned current in Fig.
4. This current system closes in the magnetosphere on the
pressure gradients in the plasma and thus is controlled by
the magnetospheric stresses. In the ionosphere these currents
cross the magnetic 5eld in the collisional ionosphere. These
collisions heat the atmosphere and the cross-5eld currents
accelerate the ionosphere plasma against the drag of the atmosphere. These currents attempt to maintain the >ow in the
ionosphere to follow the >ow in the magnetosphere. When
the stress in the magnetosphere increases, the bend in the
magnetic 5eld increases and the current >ow along the 5eld
lines increase. Fig. 4 describes a steady-state magnetosphere.
However, transient events also can cause stress-induced current systems that close in the ionosphere. Substorms that are
described below are particularly important in causing such
currents.
Fig. 5. The coupling of the magnetosphere to the ionosphere via

5eld-aligned currents. The current >ows across magnetic 5eld lines in
both the magnetosphere and the ionosphere and along the magnetic 5eld
in between. This causes the J B force applied to the magnetospheric
plasma at high altitudes to be applied to the ionosphere. The electric 5eld
associated with the motion of the plasma times the magnetic 5eld perturbation provides a Poynting >ux, S, into the ionosphere (Strangeway
et al., 2000).
Fig. 5 illustrates the physics involved in the coupling of

the magnetosphere to the ionosphere via 5eld-aligned currents. At the top of the 5gure is the magnetopause where
the stress applied to the magnetosphere has pulled the top
of a bundle of magnetic 5eld lines into the page. These 5eld
lines are now slanted with respect to their neighbors. This
shear in the magnetic 5eld is the equivalent of a parallel current. This current that began in the generator region of the
magnetosphere closing across 5eld lines in a pressure gradient and acting in the direction to slow the solar wind, now
experiences a load in the ionosphere where it applies a
stress to the ionospheric plasma in the same direction as the
stress in the magnetosphere so that the ionosphere begins to
>ow in the same direction as the solar wind. If the plasma
>ows in response to this stress, then the product of the magnetic 5eld distortion, B, and the electric 5eld, V B, is a
Poynting >ux, S, into the ionosphere. It is this energy that
is dissipated in the ionosphere as Joule dissipation.
In the Earths magnetosphere there are many types of auroral processes. Any process that causes the precipitation of
energetic charged particles from the magnetosphere into the
atmosphere in signi5cant numbers is likely to cause di7use
aurora. However, the most intense discrete aurora is caused
by electrons accelerated down into the ionosphere by parallel electric potential drops along magnetic 5eld lines in upward parallel electric currents. Thus most intense auroras are
in regions of velocity shear in the magnetospheric plasma.
2.5. Geomagnetic storms
A geomagnetic storm occurs when the energy content of
the radiation belts increases to unusually large values. The
conditions in the solar wind that lead to the generation of
geomagnetic storms are rare. The interplanetary magnetic
1010
Fig. 6. Development of a geomagnetic storm. The top panel shows the

dynamic pressure of the solar wind as measured by the WIND spacecraft.
The second panel shows the solar magnetospheric Z component of the
interplanetary magnetic 5eld as measured by the WIND spacecraft. The
third panel is the interplanetary electric 5eld computed from the product
of the radial solar wind velocity and the north-south component of the
interplanetary magnetic 5eld. The 5fth panel contains the AU index and
AL index the di7erence between which is the AE index. These indices are
computed from the maximum (upper) and minimum (lower) horizontal
component of the magnetic 5eld in the auroral zone and measure the
strength of auroral processes. The bottom panel shows the Dst index
computed from the average near equatorial surface 5eld.
5eld must be strong and steadily southward for several hours

(Russell et al., 1974). Fig. 6 illustrates the solar wind condition and magnetospheric response during the build up of a
geomagnetic storm that has been chosen because it clearly
illustrates the di7erent types of responses of the magnetosphere to geomagnetic activity. The top panel shows the
solar wind dynamic pressure. This pressure compresses the
magnetosphere and increases the 5eld strength on the surface of the Earth but has little other e7ect. In particular the
Dst index, or the average worldwide surface magnetic 5eld
in the near-equatorial regions with the quiet-day 5eld removed, shows an increase in association with the pressure
increase but this increase signals not a change in the ring
current but a change in the magnetopause current. The second panel shows the north-south or solar magnetospheric Z
component of the interplanetary magnetic 5eld during the
interval. The magnetic 5eld is slightly southward at the beginning of the interval plotted and becomes strongly southward in the second half of the plot. This leads to currents
in the auroral zone as seen in the fourth panel from the top.
This activity remains roughly constant over the entire pe-
riod plotted even though the character of the interplanetary

magnetic 5eld changes drastically.
After the pressure pulse arrives the interplanetary
magnetic 5eld strength increases and begins to have strong
>uctuations in the northsouth direction. This hardly a7ects
either the auroral currents as measured by the AE index
in the fourth panel or the ring current as registered by the
Dst index. Eventually the interplanetary magnetic 5eld becomes steadily southward and strong and lasts this way for
several hours. Now the ring current builds up (Russell et
al., 1999a, b) and the Dst plunges to values more negative
than 200 nT.
We can understand the injection of energetic plasma into
the ring current with a fairly simple model of the solar wind
magnetosphere interaction (Burton et al., 1975). The basic
premise of this model is that the energy >ows into the ring
current from the solar wind at a rate proportional to the
interplanetary electric 5eld in the dawndusk direction but
not when the electric 5eld is in the opposite direction. Once
in the ring current, the energy exponentially decays with a
5xed time constant so that a 5xed percentage of the ring current energy is lost per unit time. The Dst index, which is the
average horizontal component of the magnetic 5eld around
the near equatorial region is a good measure of the ring current, but there are other contributions to Dst . In particular
the magnetopause currents contribute to Dst as well. Burton
et al. (1975) proposed the following recipe for calculating
Dst from the solar wind parameters:
dDst =dt = F(Ey ) aDst0 ;
Dst0 = Dst b(P)1=2 + c;
F(Ey ) = d(Ey 0:5); Ey 0:50 mV=m;
F(Ey ) = 0; Ey 0:50 mV=m;
where a = 3:6 105 s1 ; b = 15:8 nT=nPa1=2 ; c = 20 nT;
and d = 1:5 103 nT=(mv m1 s).
This recipe states that the rate of change of the energy in
the ring current increases due to an energy coupling function F(Ey ) and decreases by a 5xed percentage each minute
because of loss processes. The ring current itself is Dst0 and
di7ers from the surface 5eld disturbance by contributions
proportional to the square root of the solar wind dynamic
pressure, P, and the quiet day ring current, c. The coupling
function F(Ey ) extracts energy from the solar wind when the
solar wind electric 5eld exceeds 0:50 mv m1 in the dawn
dusk direction and is zero otherwise.
The energy contained in the ring current according to
the DesslerParkerSckopke formula (Dessler and Parker,
1959; Sckopke, 1966) is
ERC (J ) = 2:8 1013 B (nT)
so that the 200 nT storm shown in Fig. 6 contains about 6
1015 J (6 PJ), which it was losing to the Earths atmosphere
at a rate of about 2 1011 W or 0:2 TW. At the peak of
the injection of power into the ring current the injection
rate was about 1012 W or 1 TW. We note that this buildup
1011
of energy was not predictable from the currents in the auroral zone as can be seen by the lack of a clear relationship
between the quantities plotted in the fourth and 5fth panels.
We note also that during disturbed solar wind conditions the
convected magnetic energy in the solar wind, that is the solar wind Poynting >ux, is about a teraWatt integrated over
the entire dayside magnetopause. Thus reconnection would
have to be 100% e7ective for the Poynting >ux to power a
magnetic storm. Instead the magnetosphere taps a small fraction of the 60 TW of mechanical energy >ux that the solar
wind convects toward the dayside magnetopause under disturbed conditions. We also note that studies of the e6ciency
of the interplanetary electric 5eld for causing geomagnetic
activity suggests that the rate of reconnection diminishes for
very high solar wind Mach numbers when the beta of the
magnetosheath becomes large and the magnetic 5eld weak
(Scurry and Russell, 1991). This could be important at Earth
during times of extremely intense solar wind disturbances
but should be more important in the outer heliosphere where
high Mach number conditions are more prevalent.
2.6. Substorms
Strictly speaking the Dungey model for the solar wind interaction sketched in Fig. 3 is a steady state model, but it can
readily be converted (e.g. Russell and McPherron, 1973) to
be a time-varying model as shown in Fig. 7. The top panel
shows the Dungey model for southward interplanetary magnetic 5eld. The magnetic 5eld has just turned southward and
>ux is being eroded from the dayside magnetosphere and
moved into the polar cap or tail lobe region. Thus, at the beginning of the substorm period, the merging rate, M , goes
up, decreasing the >ux !day in the bottom panel. Until reconnection begins between the two tail lobes, the magnetic
>ux in the tail, !lobe , will increase. When the reconnection
rate, R, in the center panel climbs rapidly, then the substorm onset has occurred and that >ux is returned 5rst to
the plasma sheet and then to the dayside. Almost by definition the substorm involves the release of energy that is
much more rapid than its accumulation time. Finally, we
note that even though its name suggests that a substorm is
a small storm or a process that leads to a storm, there appears to be little connection between the processes that lead
to substorm storage and release of energy into the auroral
zone illustrated in Fig. 6 and the storm process. The substorm appears to intimately involve the tail for storage and
release of energy. The storm appears to be associated with
the penetration of the solar wind induced plasma circulation
deep into the magnetosphere.
3. The moon and asteroids
The Earths moon and the asteroids are too small to
have currently active dynamos because any internal liquid
core has long since cooled and ceased its dynamo action.
Fig. 7. Schematic illustration of a substorm. The top panel shows the

Dungey magnetosphere when the IMF has just turned southward. In the
center panel the time variation of the reconnection and convection in
each of the three locations: dayside reconnection, tail reconnection and
convection to the dayside. The bottom panel shows the variation of total
magnetic >ux in each of the three regions.
Nevertheless, such bodies could have remanent magnetism,

created when the surface material cooled while an early dynamo was operating. Even those bodies that are too small
to ever have had a core may be part of an earlier larger
body that possessed a core. Thus a priori we cannot rule
out the presence of some magnetized regions on any of
these bodies. The Moon proves this point. Measurements
from the Explorer 35 and Apollo subsatellite magnetometers show that, when magnetized regions interact with the
solar wind >ow past the Moon, disturbances are created in
the solar wind >ow called lunar limb compression (Colburn et al., 1971; Russell and Lichtenstein, 1975). Recently,
these results were con5rmed by the Lunar Prospector mission (Lin et al., 1998). Thus the solar wind does interact strongly with magnetic regions at least down to the
size of the proton gyro radius. Reports of analogous disturbances from the asteroid, Gaspra, suggest that these disturbances can be created for even smaller scale-size features
(Kivelson et al., 1993).
It is worth reviewing the behavior of a large unmagnetized
body, such as most of the Moon appears to e7ectively be,
that is, too weakly magnetized to prevent the solar wind
hitting the surface. This interaction is shown in Fig. 8 for an
1012
Fig. 8. The solar wind interaction with the Moon when the interplanetary
magnetic 5eld is perpendicular to the solar wind >ow. The solar wind
is completely absorbed on streamlines that intersect the Moon, leaving a
cavity on the downstream side that 5lls by ion motion along the magnetic
5eld at the ion thermal velocity. Because of the charge neutrality condition
in the plasma the electrons move with the ions. In MHD terms the region
in which the plasma is moving toward the wake is called an expansion
fan (Spreiter et al., 1970).
interplanetary magnetic 5eld perpendicular to the solar wind

>ow. Not shown is the >ow-aligned case that occurs much
more rarely. In both cases the >owing plasma is absorbed
by the moon leaving an empty wake behind the Moon. In
the aligned->ow case the plasma cannot >ow into the cavity
behind the moon but the wake does narrow to a diameter
less than that of the moon. In the case with the interplanetary
magnetic 5eld perpendicular to the >ow, the plasma closes
behind the Moon at the ion thermal velocity. Since the ions
are much more massive than the electrons and since charge
neutrality requires electrons and ions to stay together in the
solar wind, ion motion governs the electrons as well.
An important aspect of this interaction is the electric 5eld.
The solar wind is a >owing, magnetized plasma and hence
has an electric 5eld in the frame of reference of the Moon.
Thus ions produced on one side of the moon by photoionization of its tenuous atmosphere will be accelerated down
on to the surface, while on the other side ions will be removed from the moon (Freeman and Ibrahim, 1975). In this
way the solar wind electric 5eld both implants ions into the
lunar surface and removes them from the lunar atmosphere.
However, the currents through the body of the Moon, driven
by this electric 5eld, are very, very small because of the extremely low electrical conductivity of the lunar surface. The
solar wind does cause currents in the interior of the moon
by carrying a spatially varying magnetic 5eld past the moon
that the moon sees as a time varying magnetic 5eld and that
induces a voltage across the moon. These currents >ow entirely within the moon and do not penetrate the crust. Finally, we note that Mars tiny moons Phobos and Diemos
have been reported to cause disturbances in the solar wind
(Riedler et al., 1989; Dubinin et al., 1990; Sauer et al., 1998)
but since these moons orbit close to the bow shock when
they are in the solar wind it is di6cult to separate lunar from
planetary e7ects.
Fig. 9. The average con5guration of the magnetic 5eld in the Mercury

magnetosphere as drawn in the noon-midnight meridian based on the
Mariner 10 >ybys. (Russell et al., 1988).
4. Mercury
To the non-specialist Mercury looks much like the Moon.
It has a cratered surface and no signi5cant atmosphere
but unlike the Moon it has a magnetic 5eld that de>ects
the solar wind well above the surface. The magnetic 5eld
con5guration in the noon-midnight meridian is shown in
Fig. 9 as inferred from two >ybys by Mariner 10 in 1974
and 1975. Some recon5guration of the magnetosphere was
detected on the 5rst >yby and interpreted in terms of a
magnetospheric substorm as on Earth (Siscoe et al., 1975),
but, since Mercury has no signi5cant ionosphere, stresses
might be communicated much more rapidly in the Mercury magnetosphere than in the terrestrial magnetosphere.
Under the assumption that Mercurys magnetosphere was
responsive to the interplanetary magnetic 5eld orientation in a manner similar to that on the Earth, Luhmann
et al. (1998) modi5ed Tsyganenkos (1996) terrestrial magnetic 5eld model to apply to Mercury. Fig. 10 shows the
equivalent magnetic 5eld models for three IMF conditions
obtained by Luhmann et al. (1998). They then assumed that
these model 5elds were immediately attained when the IMF
changed and calculated what IMF conditions would create
the magnetospheric conditions observed. Their conclusion
was that the dynamics of the Mercury magnetosphere could
be directly driven with little or no storage of energy in the
magnetic tail, unlike the terrestrial magnetosphere.
1013
Fig. 11. The con5guration of the Ganymede magnetosphere in Jupiters

magnetic meridian when Ganymede is at high jovian magnetic latitudes
(after Kivelson et al., 1996a).
magnetosphere is more a7ected by the external (jovian)

magnetic 5eld than the corotating magnetosphere plasma.
This 5eld rocks back and forth mainly in the meridian plane
as Ganymede moves back and forth across the magnetic
equator. This alters the high altitude magnetic 5eld direction
shown in Fig. 11, but the low altitude magnetic 5eld remains
5xed. There are no reports yet of substorms or storms in this
tiny magnetosphere.
6. Venus, Mars and comets
Fig. 10. The Mercury magnetic 5eld in the noon-midnight meridian if the
5eld is responsive to the IMF in a manner similar to that of the terrestrial
5eld (Luhmann et al., 1998).
5. Ganymede
The last intrinsic magnetosphere in a terrestrial-sized body
that we discuss is that of Jupiters moon, Ganymede, that
sits well inside the jovian magnetosphere at 15 jovian radii
(RJ ) in a plasma of density about 4 cm3 >owing about
130 km s1 relative to Ganymede. Its magnetic moment is
1:4 1013 Tm3 (Kivelson et al., 1997a) and it sits in a
background magnetic 5eld of strength 100 nT. The static
pressure of the Jovian magnetic 5eld is about 4 nPa and
that of the >owing plasma is 2 nPa so that the Ganymede
Venus and comets are completely devoid of intrinsic magnetic 5elds. Mars has some signi5cant regions of remanent
magnetic 5eld (Acuna et al., 1998) but not so strong that
it dominates the solar wind interaction with the planet. The
solar wind interaction for Venus and Mars is basically as
sketched in Fig. 12 and described in detail in the volume
Venus Aeronomy (Russell, 1991) and Venus and Mars: Atmospheres, Ionospheres and Solar Wind Interaction (Luhmann et al., 1992). Solar EUV shining on the planetary
atmosphere creates an ionosphere. While the ions are produced from the atmosphere over a wide altitude range, they
can recombine only at low altitudes where there are collisions. This sets up a circulation pattern in the ionosphere
that is downward in the neighborhood of the subsolar point
and toward the antisolar point at other solar zenith angles.
The thermal pressure of the ionosphere is generally greater
than the dynamic pressure of the solar wind so that it can de>ect the solar wind prior to the solar wind hitting the atmosphere. When the solar wind pressure is low this de>ection
occurs well above the collisional regime in the ionosphere
and there is a thin layer of current separating the solar wind
and the ionosphere so that the magnetic 5eld penetrates very
1014
Fig. 12. The con5guration of the magnetic 5eld in the Venus and Mars
ionospheres for low and high dynamic pressures (after Saunders and
Russell, 1986).
little into the ionosphere. In the case of a high solar wind

dynamic pressure the interaction takes place at a lower collisional altitude and a thick current layer is formed with
magnetic 5eld deep into the ionosphere. In both cases the
supersonic solar wind is de>ected with a bow shock much
like the Earth but, when the EUV is weak at solar minimum,
the solar wind apparently is able to interact directly with
the atmosphere and is absorbed. This absorption allows the
shock to move closer to the planet.
Some ions are created in the >owing solar wind itself
from a hot oxygen exosphere whose high thermal velocity
is su6cient to carry neutral atoms to high altitudes. As at
the Moon the solar wind electric 5eld is responsible for
accelerating these ions. Some of them get accelerated back
to the planet. Others are accelerated away from the Venus
and have been detected as far away as Earth (Grunwaldt
et al., 1997). The geometry of the magnetic 5eld lines in
the tail is in the sense to accelerate ions via the J B
force away from the sun (McComas et al., 1986). Some
evidence for substorm-like activity in the tail or at least the
production of energetic ions in the tail of Venus has been
presented by Vaisberg et al. (1994). The loss rate of ions
from Venus is poorly known but it may be close to 1024
oxygen ions per second. To accelerate these ions to the solar
wind velocity requires about 3 GW, a number that is a small

percentage of the over 100 GW incident on Venus in solar
wind mechanical energy >ux.
The solar wind interaction with Mars much resembles
that with Venus. However, the solar wind magnetic 5eld is
weaker and the radius of Mars smaller than at Venus. Hence
while MHD provides a suitable model for the solar wind
interaction at Venus, kinetic e7ects become important at
Mars (Brecht, 1997). Also, while there is no global planetary
5eld there are strong localized 5elds that also cause some
di7erences with the Venus interaction (Acuna et al., 1998).
A more detailed discussion of the solar wind interaction
can be found in the article by Vaisberg et al. (1990) and
the volume edited by Luhmann et al. (1992). Evidence for
substorm-like phenomena have not yet been reported for
the Martian tail but the Mars plasma environment continues
to be studied. Data are still being returned from the Mars
Global Surveyor and the Japanese Nozomi spacecraft is due
to enter Mars orbit in late 1993 (Yamamoto and Tsuruda,
1998).
The European Giotto and Russian Vega 1 and 2 spacecraft directly probed the coma of comet Halley and the
ISEE-3 mission, renamed ICE, has probed the coma of
comet Giacobini-Zinner. In addition Giotto encountered a
second comet, Grigg-Skjellerup. The cometary plasma environment has been described in detail in the volume Cometary
Plasma Processes (Johnstone, 1991). The cometary interaction covers a larger volume of space than the interaction with
Venus and Mars because of the lack of signi5cant gravitational pull on the neutral gas that escapes from the nucleus
at over 1 km=s and is ionized in about a day. These measurements represent only snapshots of the region around the
comet and to study the dynamics of the cometary plasma
environment one must rely on remote sensing data.
These remote sensing data provide a strong case for a
substorm-like phenomenon. Fig. 13 shows an example of a
cometary tail disconnection event where it appears that the
tail has been pulled out of the cometary coma. The bright
head of the cometary coma is thought to be much like the
Venus ionosphere and magnetosheath but on a larger scale
because of the absence of a signi5cant gravitational 5eld.
As at Venus and Mars a magnetic-5eld-free region can form
close to the nucleus if there is su6cient outgassing and
EUV to form an ionosphere that has su6cient pressure to
stando7 the solar wind dynamic pressure. The ions created
in the solar wind >ow form a ring around the magnetic 5eld
in velocity space that is unstable to the formation of both
ion cyclotron waves and mirror mode waves. The former
oscillate at the gyro frequency of the heavy ions created.
The latter form depressions in the magnetic 5eld strength
with a diameter of several gyroradii.
There are two outstanding theories of cometary tail disconnection events. The 5rst model assumes that the phenomenon involves reconnection in the cometary coma when
the interplanetary magnetic 5eld reverses (Niedner and
Brandt, 1978). The second assumes that reconnection in the
1015
Fig. 13. A cometary tail disconnection event, seen in comet Morehouse

1908a at 1943 UT, October 1, 1908 (Niedner and Brandt, 1978).
tail causes the disconnection as sketched in Fig. 14 (Russell

et al., 1986). Just as with terrestrial substorms, such a tail
disconnection may be triggered by a solar wind event.
7. Jupiter
Jupiter is the planet of solar system records. It has
the fastest rotation, the most mass, the largest radius, the
strongest surface magnetic 5eld, largest magnetic moment,
most intense radiation belts, strongest radio emissions,
and the largest moon, Ganymede. Fittingly it also has the
largest magnetosphere, one so large that, if it could be seen
from Earth, it would appear larger than the Earths moon.
The size of the magnetosphere, like that of the Earth, is
determined by a balance between the solar wind dynamic
pressure and the pressure exerted by the magnetosphere.
Part of the reason for the enormity of the jovian magnetosphere is that the mass density of the solar wind and hence
its dynamic pressure decreases as the inverse square of the
heliocentric radius. By the time the solar wind has traveled
from 1 AU to Jupiter the dynamic pressure has decreased a
factor of 27. Thus, if Jupiter were like the Earth, this e7ect
alone would increase the radius of the magnetosphere by
a factor of 4. Under most conditions the Earths magnetic
5eld can be approximated by a dipole. A dipole magnetic
Fig. 14. A tail reconnection model for the disconnection of cometary tails.
5eld is a force-free magnetic con5guration in which the outward magnetic pressure is balanced by the inward curvature
force. Its 5eld strength decreases as the cube of the radius.
The jovian magnetic 5eld is decidedly not force free and
its interior pressure falls o7 more slowly than an inverse
cube in its outer portions. This e7ect increases the size of
the magnetosphere relative to that of the Earth over and
above its factor of 18,000 greater magnetic moment. The
weaker outward pressure gradient also makes the size of the
magnetosphere more sensitive to changes in the solar wind
dynamic pressure than the Earths magnetosphere (Slavin
et al., 1985; Huddleston et al., 1998a). Thus the nose of the
jovian magnetosphere has been found at distances from 40
to over 100 R jovian radii (RJ ).
As we discuss in greater detail below, the additional
force in the jovian magnetosphere that is not present in the
terrestrial magnetosphere is centrifugal force due to a very
strong source of plasma at the moon Io. This additional
stress stretches out the magnetic 5eld forming a magnetodisk beyond about 24RJ (Smith et al., 1975). The resulting
magnetic con5guration resembles the sketch of the jovian
1016
Fig. 15. Magnetic 5eld lines in the noon-midnight meridian of the jovian
magnetosphere showing the current sheet in the magnetodisk region (after
Russell et al., 1998a, b).
magnetic 5eld in the noon-midnight meridian shown in

Fig. 15. As can be seen in this 5gure the nose of the magnetosphere is sharper than that of the Earth. Just as the
aerodynamic shape of a supersonic airplane allows the bow
shock to form very close to the nose of that airplane, the
more streamlined shape of the jovian magnetopause allows
the bow shock to be formed closer to the magnetosphere
than at Earth (Stahara et al., 1989).
The existence of a variable source of mass in the inner
jovian magnetosphere provides an extra dimension to the
dynamics of the jovian magnetosphere. There is possible
control by the rate of mass addition as well as by the solar
wind and the interplanetary magnetic 5eld. This mass addition could a7ect the size and the shape of the magnetosphere.
We do not yet know how variable is this mass-loading rate,
so we cannot yet estimate how important this e7ect is on the
size of the magnetosphere. If mass loading were to totally
cease we estimate that the magnetopause stando7 distance
would be only about 40RJ which is similar to the smallest
stando7 distances seen, but these conditions also most probably correspond to periods of higher than usual solar wind
dynamic pressure.
As we discussed above, the Earths magnetosphere is very
much a7ected by the strength and orientation of the interplanetary magnetic 5eld, or more correctly, the product of
the solar wind velocity and the component of the magnetic
5eld perpendicular to the solar wind >ow. While the magnetic 5eld strength is almost a factor of 10 smaller at Jupiter
than at the Earth, the enormous size of the magnetosphere
might compensate for this decrease. We can estimate the importance of the solar wind electric 5eld on a magnetosphere
by comparing the solar wind electric 5eld, the product of
the magnetic 5eld perpendicular to the solar wind >ow and

the solar wind speed, with the corotational electric 5eld of
the planetary magnetosphere that is equal to the corotational
speed !R times the north-south component of the magnetic
5eld. Since the corotational speed increases as R and the
magnetic 5eld decreases as R3 (in a dipole) the electric 5eld
of a rotating dipolar magnetosphere decreases as L2 . Thus
the terrestrial corotational electric 5eld is 14L2 mV m1
and that of Jupiter 4900L2 mV m1 where L is the distance in planetary radii. The solar wind electric 5eld at 1
and 5:2 AU respectively is typically 3 and 0:4 mV=m. If all
of this 5eld were able to penetrate the terrestrial and jovian
magnetospheres, the interplanetary and corotational 5elds
would be equal at 2:1RE and 100RJ respectively. Since at
Earth only about 10% of the solar wind electric 5eld penetrates the magnetosphere, the typical distance at which the
electric 5elds balance is 6RE . If the same rule applied to
Jupiter the balance point would be about 300RJ . In fact, we
have reason to believe that reconnection is even less e7ective
at Jupiter than at Earth. While >ux transfer events, one manifestation of magnetopause reconnection, were observed at
the jovian magnetopause they were typically smaller and less
frequent than on Earth (Walker and Russell, 1985). Moreover, the reconnection is apparently less e6cient for high
beta conditions that occur behind high Mach number shocks
(Scurry et al., 1994), and the jovian shock has a signi5cantly higher Mach number than the terrestrial shock. Finally
and most importantly, jovian auroral phenomena behave differently than terrestrial aurora (Clarke et al., 1996; Prange
et al., 1998). Jovian aurora rotate with Jupiter and are associated with the inner magnetodisk portion of the magnetosphere. Unlike terrestrial auroras they do not cluster about
the boundary between open and closed 5eld lines. It is clear
that the jovian magnetosphere works much di7erently than
the terrestrial magnetosphere.
The electric 5eld associated with corotation arises because the ionosphere rotates with the atmosphere and the atmosphere rotates with the planet. Since electrons can move
quite freely along the magnetic 5eld, the magnetic 5eld lines
are equipotentials and transmit this electric 5eld to the equator regions. It is, of course, possible that this electric 5eld
is altered in some way. If some process held the >ux tube
5xed in the equatorial plane, it would either have to bend
because it was also 5xed to the ionosphere, or it would
have to slip with respect to the ionosphere. If it slipped with
respect to the ionosphere, a potential drop would have to
appear across the point where the >ux tube slipped. As discussed for the Earth this velocity shear leads to intense aurora. Thus, to zeroth order, auroral pictures of Jupiter may
simply show us where this slippage is taking place.
7.1. Mass addition at Io
Io is the engine that drives the jovian magnetosphere and
mass addition is the fuel that powers the magnetosphere.
1017
Fig. 16. Schematic diagram of the addition of ions to the Io torus (after Huddleston et al., 1998b).
In the following sections we treat 5rst the immediate vicinity of Io and then move out into the Io torus and then into
the middle and distant magnetosphere. The interaction of Io
with the jovian magnetosphere is shown in Fig. 16. Io orbits Jupiter at 17 km s1 whereas corotating plasma would
be travelling at 74 km s1 . Ions that are produced from neutral atoms at rest with respect to Io must be accelerated by
57 km s1 . If the mass loading rate is large and the generally accepted 1 ton per second is large, the plasma must
slow down. Where there is good coupling of the ionosphere
to the magnetospheric >ux tube, the slow down at the equator can be taken up by a bending of the 5eld. In steady state
the bend in the magnetic 5eld can remain but the velocity
of the plasma must reach the corotation level if there is no
slippage. If the bend is to be erased the >ow must exceed
the corotational velocity for a while to catch up with the
ionosphere. The existence of an auroral spot, not only at the
magnetic footpoint of Io, but extended in the direction of rotation suggests that the >ux tubes that pass near Io slip with
respect to the ionosphere for a long distance downstream of
Io (Prange et al., 1996).
Fig. 16 also illustrates some important aspects of the distribution of the picked up charged particles about the magnetic 5eld line. The initial motion of the charged particle is
in a cycloid about the 5eld line. The thermal or gyro velocity of this motion is equal to its bulk velocity. Because
Io is close to the magnetic equator, but not usually at the
equator, there is a small velocity of the ion along the magnetic 5eld. Initially the distribution of the ions can be represented as two delta functions perpendicular to the 5eld
but with time the particles scatter and eventually become
Maxwellian.
If the mass loading rate is large enough, then the addition

of plasma measurably slows the >ow. This is seen at Io
(Frank et al., 1996) and is illustrated in Fig. 17. The plasma is
heated at the edge of the wake where newly created ions are
added to the >ow with a velocity equal to the >ow velocity.
The decrease in the magnetic 5eld strength in the wake is
di6cult to explain simply by plasma diamagnetism. The
missing magnetic energy in the wake is up to 4:5 MeV=cm3
which would require a density of 22; 500 ions cm3 at a
temperature of 200 eV. While the density in the wake is at
its peak over 104 ions=cm3 , the temperature is very cold,
about 10 eV and does not explain the depression. Moreover,
any slow down of the >ow and pile up of magnetic 5eld
should increase the 5eld in the near wake, not decrease it.
The middle panel shows estimates of the source density
of new ions (Bagenal, 1997) based on the plasma measure+
ments, and of speci5cally, NSO
, based on the amplitude of
2
the ion cyclotron waves. The ions produced on >ux tubes
that move across Ios poles and into the wake region is a
small fraction of the total production, only a few tens of
kg s1 (Russell et al., 1997). The ions produced in the >ow
that is de>ected by Io, as crossed by Galileo, is between
180 and 580 kg s1 (Bagenal, 1997) signi5cantly less than
the 1000 kg s1 usually quoted for the mass loading rate
(Hill et al., 1983). Although re5ned estimates of the plasma
parameters in the ion torus now exist (Frank and Paterson,
1999a, b), they have not changed these values substantially.
The error bars simply represent the uncertainty in the shape
of the mass loading region around Io. Unless the Galileo
measurements occurred during a quiescent period of Io activity, and the density data suggest if anything the opposite
is true, then we come to either of two conclusions: the mass
1018
Fig. 17. Averaged pro5les of plasma parameters from the Galileo-Io >yby.
From top to bottom the panels show magnetic 5eld magnitude (Kivelson
et al., 1996), ion velocity and temperature from the plasma analyzer
instrument (Frank et al., 1996) total iogenic source density (Bagenal,
1997) and SO+
2 component pickup estimated by Huddleston et al. (1998b)
and a comparison of the rms amplitudes of compressional and transverse
components of B (Huddleston et al., 1999).
loading rate is smaller than generally believed, or the mass

loading region extends well beyond the near vicinity of Io.
As we discuss in the next section, the latter may well be
true. If a 1000 kg s1 were accelerated to corotational energies from orbital energies by the Io-torus interaction, then
2:5 TW (2:51012 W) must be provided. This is equivalent
to the power introduced for short periods into the Earths
magnetosphere during the largest storms but at Jupiter this
energy source is active continually. If the mass loading rate
were as low as 200 kg=s, then the energy supplied to the
torus is 0.5 TW. Because of its enormous cross-section
the mechanical energy >ux in the solar wind intercepted
by the jovian magnetosphere is much larger than that intercepted by the Earths magnetosphere and it would dwarf the
energy >ux into the jovian magnetosphere at Io, if it could
be tapped. Reconnection, that draws energy from the solar wind mechanical energy at Earth, appears to be weak at
Jupiter and in any event would not a7ect the rapidly rotating plasma of the magnetosphere whose corotational electric 5eld much exceeds that of the solar wind. Thus only
over the polar cap and near its boundaries do we expect to
see activity powered by solar wind coupling. Fig. 17 also
Fig. 18. Trajectory of Galileo on its 5rst >yby of Io illustrating the

phenomena observed and the inferred >ow pattern. Vectors along the
trajectory indicate >ow direction as seen by the plasma analyzer (after
Russell et al., 1997).
emphasizes that the main mass loading region is about 2 Io

diameters across. The amount of magnetic >ux convected
into the 4RIo wide region is about 80; 000 Wb s1 . Thus,
about 80; 000 Wb s1 accompany the newly added plasma
until they reach a region in which they can be separated.
Such a separation is needed in steady state because the magnetic >ux in the magnetosphere is set by the jovian dynamo
and changes at most slowly. In contrast ions need to escape
from Jupiter to avoid reaching an in5nite density.
A 5nal aspect, illustrated by Fig. 17, is the growth of
waves in the neighborhood of Io. On either side of Io these
waves are principally ion cyclotron waves, while at the
edges of the wake there are mirror mode waves (Kivelson
et al., 1996a, b; Russell et al., 1998b; Huddleston et al.,
1999; Russell et al., 1999b). The appearance and amplitude
of the waves is consistent with the observed properties of
the plasma and adds credence to interpretation of the nature
and the strength of the interaction.
Fig. 18 shows a schematic of the interaction of the torus
plasma with Io during the 5rst Galileo >yby illustrating
the probable >ow streamlines and where various phenomena were observed along the trajectory. One phenomenon
not relevant to the mass content of the torus but possibly
caused by the mass loading is the presence of electron beams
in the wake (Frank and Paterson, 1999a; Williams et al.,
1999). These beams possibly mark >ux tubes with electric
potential drops in the ionosphere where the mass loaded
Fig. 19. Three dimensional plot of the reduced distribution function versus
energy per charge for spectral measurements made in Voyager 1s C cup
between 0730 UT (7RJ ) and 1145 UT (4:9RJ ) on March 5, 1979. The
back panel shows the total positive-ion elementary-charge concentration
as a function of time from 5ts to the corresponding spectra (Belcher,
1983).
>ux tube is being accelerated back to corotational velocity. These beams are important for the electron heat balance
because electron-electron heat transfer is more rapid than
ion-electron heat transfer.
7.2. The Io torus
While Voyager did not pass as close to Io as did Galileo,
it provided measurements of the Io torus that have proven
invaluable to our understanding of the energetics of the
Jovian engine. A good summary of our pre-Galileo understanding can be found in the review by Spencer and Schneider (1996). Fig. 19 shows energy per charge readings on
March 5, 1979 as Voyager 1 cut inbound through the Io
torus, prior to its outbound pass underneath Io. Before 0800
UT the torus plasma had a low density, 900 cm3 , and
was warm. As the radial distance of the Io orbit is reached,
the plasma density rises and the plasma remains warm. After 1000 UT, when the spacecraft is well inside the Io orbit
a very cold population of plasma is seen with a composition that indicates its iogenic origin. (The varying density
and charge state has been explained in terms of centrifugal
and mirror forces by Bagenal (1985).) The reason for the
low temperature in this region is poorly understood. Either
the plasma is produced cold or it cools in the inner torus.
The plasma in the wake directly behind Io (Frank and Pa-
1019
terson, 1999b) that was picked up at very low velocities is

cold but it is not clear how to transport this plasma into the
inner torus. Neutral gas from Io will get picked up in the
inner torus with a much smaller gyro velocity than in the
outer torus. This also leads to a lower temperature. Finally,
the residence time in the inner torus may be long enough
for Coulomb collisions to cool the ions.
Two mechanisms may be responsible for maintaining Ios
neutral atmosphere than in turn supplies the Io plasma torus.
The 5rst is solar heating of the SO2 frost on Io. This would
produce an atmosphere centered on Io noon. The second
process is sputtering by the Io torus plasma. This would produce an atmosphere centered on the upstream site of Io. To
explain the appearance of the neutral sodium cloud, Wilson and Schneider (1999) involve both these sources. However, sodium is a minor constituent of the atmosphere and
the photochemistry of sulfurdioxide is di7erent than that of
sodium and we cannot assume that the neutral SO2 cloud is
the same as the Na cloud. One way to determine the extent
of the SO2 neutral cloud is to determine where the ion cyclotron waves arise when the neutral cloud becomes ionized.
There is very little evidence for an extensive spherical exosphere at Io as there are essentially no waves upstream of
Io (Russell et al., 2001a). Rather the neutral cloud appears
to be a disk extending perhaps 0:5RJ inside and outside of
Ios orbit and far downstream but not upstream.
The radial extent of the neutral torus around Io is at 5rst
surprising because the atmosphere of Io is relatively cold.
However, photoionization followed by acceleration of ions
in the outward electric 5eld associated with the corotation of
the torus plasma followed by charge exchange can produce
fast neutrals close to Io where the neutral density is large
(Wilson and Schneider, 1999; Wang et al., 2000). The new
ions travel in cycloidal paths in roughly the direction of
corotation. If the local density of neutrals is su6ciently large,
these ions can charge exchange to become a neutral cloud
mainly outward from Jupiter but with an inward extension as
well. This allows the total mass-loading associated with Io to
be large even though close to Io the mass-loading is smaller
than expected. The Galileo data also allow a test of whether
the sulphur dioxide atmosphere responsible for the torus
comes from a dayside Io atmosphere or a wake-sputtered
atmosphere. If the former is the source the neutral cloud
would vary with the orbital phase of Io. In the latter case it
would not. Galileo data show a very strong variation with
the orbital phase of Io (Russell et al., 2001a).
Turning now to the origin of the hot outer torus, we show
in Fig. 20 isodensity contours of the torus derived from the
Voyager 1 data by Bagenal (1994). We have integrated the
number density along the 5eld and in azimuth to give a torus
density per jovian radius (RJ ) in the radial direction. If we
adopt a simple model of mass conservation with a one ton=s
source at 5:9RJ and radial expansion we obtain the velocities
shown above the plot that is required to maintain this pro5le. A source strength of 200 kg s1 would give numbers
one-5fth of these values. At the higher mass loading rate,
1020
Fig. 20. Io torus density as observed by Voyager 1 (Bagenal, 1994)

Numbers at top of 5gure indicate the vertically and azimuthally integrated
mass of the torus in radial bins of extent 1RJ and the estimated radial
>ow velocity for iogenic mass loading of a ton=s.
the plasma would take about 3 months to reach a distance

of 7RJ where the density drops more rapidly and the radial
velocity presumably increases. This is very similar to the
scale time for the rise in S+ and its decay after a Na outburst
studied by Brown and Bouchez (1997). Thus it is possible
that the 1 ton=s is achieved occasionally but is not a permanent rate. Long term torus studies indicate much variability
of its emissions (Thomas, 1993).
An alternative loss mechanism to radial transport is loss
down the magnetic 5eld lines into the ionosphere. This is
desirable because it removes the particle from the magnetic
5eld lines. If signi5cant particle precipitation is occurring
then our estimate of radial velocities based on mass conservation are over estimated. Radial transport in a collisionless
plasma transports the magnetic 5eld too. While Jupiter needs
to shed these ions to maintain a steady state, it does not need
to shed magnetic >ux. At radial distances close to that of
Io, the torus is quiet but the >uctuation level increases with
increasing radial distance. Because the background 5eld is
strong, the fractional amplitude of these waves is small. The
transverse and compressional wave amplitudes are less than
0.1% of the background 5eld at frequencies near the ion
cyclotron frequency throughout most of the torus (Russell
et al., 2001b). Whether these waves can reduce the mass
content of the >ux tubes depends both on the size of the loss
cone and on the radial transport times. Estimates of the loss
rate as the >ux tubes convect outward through the dipolar
region of the magnetosphere indicate that this is only a minor loss mechanism (Russell et al., 2001b). This is in accord
with the observed low auroral activity in this region. In the
magnetodisk region the loss cone is small and the plasma is
trapped in the current sheet region.
Fig. 21. Linearly detrended time series near outer edge of torus at 7:7RJ
obtained on December 7, 1995 from 1521 to 1531 UT showing step-like
changes in the magnetic 5eld that might be produced by the interchange
in stability (Russell et al., 1999a, b).
Radial transport is usually ascribed to the interchange instability in which a heavily laden >ux tube interchanges radially with a lightly laden tube so that the heavily laden tube
moves outward. This preserves magnetic >ux and transports
ions outward. This process has been treated by many authors
(Southwood and Kivelson, 1987, 1989; Siscoe and Summers, 1981; Pontius and Hill, 1989; Vasyliunas, 1989; Hill,
1994). Observational evidence for >ux tube interchange can
be found in the Io torus. Kivelson et al. (1997b) have reported an empty inward moving >ux tube and the >ux tubes
near the outer edge of the Io torus, as shown in Fig. 21, appear to be divided into tubes of di7erent plasma content as
one might expect from the interchange instability (Russell et
al., 2000b). If this process is not accompanied by scattering,
tubes will retain their total content and be full until they
get emptied at some very large radius, e.g. by loss of the
ions down the tail. Then the interchanging tubes would consist of two types, full and empty. The variation seen in the
magnetic 5eld in Fig. 21 suggests that there is a continuum
of >ux tube contents at 7:7RJ . Plasma observations show
that this region at the outer edge of the torus has unusually
intense 5eld-aligned electron beams with directional energy
>uxes of up to 90 erg=cm2 s sr parallel and anti-parallel to
the 5eld, Frank and Paterson (1999a). If these >uxes extend
all around the edge of the Io torus and extend over a 1RJ
band, then taking a >ux of 15 erg=cm2 we obtain a 0:3 GW
energy source if the beams each have a width of 0:5 sr. Thus
these beams represent a signi5cant heating source for the
plasma. We note that the appearance of these beams coincides with a steep gradient in the plasma number density,
i.e. the outer edge of the torus, reinforcing the interpretation of these >uxes as being associated with the interchange
instability. A second but possibly related phenomenon, is
the occurrence of transient energetic charged particle injections deep in the middle magnetosphere (Mauk et al., 1999).
1021
Fig. 23. Magnetic 5eld observed by the Galileo spacecraft just outside
of the orbit of Europa showing wake-like disturbances that appear to be
convecting outward from Europa. The two disturbances seen at 0900 and
1100 UT can be used to estimate the radial >ow velocity (Russell et al.,
1999a, b).
Fig. 22. Evidence for a plasma plume from Europa (Intriligator and Miller,
1982). Top panel shows the >ux of oxygen and presumably sulfur ions
observed by Pioneer as it moved outward past the orbit of Europa. The
bottom panel shows a sketch of the trajectory 5xed in a magnetic dipole
ordered coordinate system.
factor. Nevertheless, this outward motion, no matter what

its absolute value, suggests to us that radial transport is operative. Nevertheless, we recall that for this transport to be
e7ective we need to 5nd a mechanism that empties the >ux
tube of ions and returns empty >ux tubes to the inner magnetosphere.
7.4. The inner edge of the magnetodisk
While they envision an external trigger and possibly radial

transport inward, an internal source, possibly the interchange
instability is also a candidate for these injections.
7.3. Transport in the middle magnetosphere
Our examination of the transport in the Io torus left us with
two possible loss mechanisms for the ions, precipitation and
transport. We can help decide between the two mechanisms
with the aid of the Europa plume (Intriligator and Miller,
1982). Pioneer plasma observations shown in Fig. 22 (top)
have been interpreted as a plume emitted by Europa and
wrapping around Jupiter dragged by the corotational >ow
but slowly moving out at about 400 m s1 as sketched in the
bottom panel of Fig. 22. Disturbances seen in the Galileo
magnetometer data (Russell et al., 1999a) also have been
interpreted as due to a Europa wake. Fig. 23 shows a pair
of these disturbances. If these were emitted by Europa on
successive rotations of Jupiter, then the radial separation of
these two observations is equivalent to an outward speed of
the plume of 500 m s1 , quite consistent with the inference
of Intriligator and Miller (1982). Unfortunately, if the two
events detected were not produced on succession orbits of
Europa then the speed would be reduced by some integral
The journey outward is not steady. Evidence of the

time-varying behavior of the magnetosphere comes 5rst
about 24RJ where the magnetic 5eld switches from nearly
dipolar inside of this distance to decidedly non-dipolar outside of this radial distance. Fig. 24 shows the change in the
radial component of the magnetic 5eld between the regions
above and below the current sheet and the component normal to the current sheet (Russell et al., 1999c). The normal
component varies from pass to pass and thus the J B
force in the plane varies. We can conjecture that the force
balance in this region is principally between the magnetic
forces, the centrifugal force of the cold nearly co-rotating
plasma and the pressure gradient in the hot plasma that
causes the depression around the current sheet. Two of
these components can be estimated from the magnetometer
and used to solve for the centrifugal force and if corotating,
for the mass density. This is done in Fig. 25 for the 5rst
four inbound passes. These show some consistency in the
radial pro5le of density from pass to pass but with occasional large decreases in the density. As before with the
torus plasma we can invoke conservation of mass to obtain
a velocity pro5le. These points together with our earlier
derived speeds are shown in Fig. 26. The left-most line is
derived from the assumption that Io provides 1 ton=s. A
1022
Fig. 25. The mass density of a ring of the magnetodisk current sheet of
radial extent, 1RJ , estimated from the radial force balance (Russell et al.,
1999a, b).
Fig. 24. Magnetic 5eld components (radial and normal to the current sheet)
measured across and at the current sheet respectively. These components
allow the radial J B force to be calculated. We note that while there
is little variation in the radial magnetic 5eld and hence the azimuthal
current, the normal component is variable (Russell et al., 1999a, b).
very consistent pattern of acceleration with distance is seen

with almost stagnant plasma in the inner torus, increasing
in radial velocity to 500 m s1 at 10RJ to about 8000 m s1
at 20RJ to about 25 km s1 at 30RJ . If we instead use the
200 kg s1 value of Bagenal (1997) we displace this line
by a factor of 5 downward. This value is consistent with
that derived from the LECP data by Kane et al. (1995), but
it becomes less consistent with the Europa plume estimate
based on the once-around assumption. The estimate based
on the Europa wake detection at radial distances beyond the
Europa orbit are higher than these two curves. This suggests
that the Europa wake sightings occur during periods of
higher than usual outward transport. This is consistent with
their rare occurrence. We emphasize that these two curves
bracket our inferred radial pro5le of the average outward
velocity under the assumption of no particle precipitation.
If particles are lost along the 5eld line so that part of the
Fig. 26. The radial variation of the out>ow velocity estimated from
conservation of mass and Voyager observations, the two independent
observations of the Europa plume by Pioneer 10, in December 1973 and
Galileo in September 1996, by the stress balance in the jovian current sheet
as deduced from Galileo magnetic observations and from the Voyager
LECP anisotropy measurements (Russell et al., 1999a, b).
1023
Fig. 27. Four crossings of the magnetodisk current sheet in current sheet ordered coordinates illustrating the multiple crossings of the current sheet
associated with surface waves of about 10 min period as well as the evolving structure of the normal component to the current sheet (Russell et al.,
1999a, b).
density decrease is due to particle loss along the 5eld, the

required radial out>ow speed would be further reduced.
Another means to estimate the radial velocity is to use
magnetic >ux conservation. We do this despite the fact that
as much magnetic >ux must be convected inward as is being
convected outward. The third curve in Fig. 26 shows the
velocity required to give the radial fall-o7 in the magnetic
5eld crossing the equatorial plane if the plasma is convecting
the magnetic >ux outward. For this calculation we assume
that there is slow out>ow over nearly 360 with a narrow

region of rapid in>ow. If the >ux were to return at the same
speed over half the circumference, then the speed would be
twice that shown. As we discuss later we believe that the
magnetic >ux does return rapidly over a narrow region, at
least in the middle magnetosphere and torus, so that our
estimate is approximately correct.
There is a di7erent slope in the inner magnetosphere and
torus than in the magnetodisk. This could be due to the
cessation of particle losses along the 5eld once the

magnetodisk is encountered. We further note that the
magnetic-5eld based calculation is about a factor of two
less than the lower of the two density-based calculations.
While both calculations were made assuming 360 symmetry and so both su7er from lack of knowledge of the
true >ow pattern, the magnetic estimates especially at high
radial distances push the accuracy of the magnetic measurements because the normal 5elds are very small, approaching
0:1 nT. Thus the lower two curves should be viewed as
mutually consistent. Based on the 200 kg s1 curve it takes
about 2.5 years for the torus plasma to reach Europa from
Io and about 0.5 years from Europa to Ganymede. From
Ganymede to Callisto takes about one month and then from
there to 50RJ takes about a week.
As our conjectured >ow moves outward through the magnetodisk, the current sheet becomes less stable. As shown
in Fig. 27 the magnetic 5eld crossing the current sheet
1024
(component Bb ) is moderately strong near 25RJ averaging

about 5 nT in the center of the sheet with some high frequency noise. At about 40RJ the normal component has diminished to about 1 nT with >uctuations that may in part
be due to surface waves that cause changes in the current
sheet orientation. These waves carry the current back and
forth over the spacecraft. At 50RJ the motion of the current
sheet is even greater and the possible e7ect on the apparent
normal component across the average current sheet is even
greater but the reversals seen in the normal component may
not be real. Their correlation with the >uctuation in current
sheet location suggests that the current sheet orientation is
rocking as the surface waves move by. Finally, at 55RJ we
see pulses of reversed 5eld components at the current sheet
that cannot be explained by current sheet motion. These narrow pulses of reversed 5eld seem to be small tearing islands
embedded within the current sheet. We suggest that these
small tearing islands may act as the seeds for reconnection
when this region rotates to the nightside but that while it is
on the dayside the tearing islands remain quiescent within
the current sheet.
The multiple crossings of the current sheet, evident on
three of the four current sheet encounters displayed here,
are quite fascinating as they indicate a magnetodisk that is
constantly in motion, agitated by some source, possibly deep
inside the magnetosphere and possibly due to the solar wind
interaction.
Fig. 28. The Galileo orbit used to study the substorm-like phenomena
presented in Figs. 29 and 30 (Russell et al., 1998a, b).
7.5. Separating the ions from the magnetic 7eld

When the Galileo spacecraft reached the distant midnight
magnetosphere, it was able to detect the process separating the ions from the magnetic 5eld. Fig. 28 shows the
trajectory of the Galileo spacecraft in May and June 1997
when it had a long sequence of continuous data through the
post midnight sector. Fig. 29 shows the magnetic measurements from June 1 to June 21 when Galileo moved from
apogee to about 50RJ at a local time of about 4 a.m. This
section of data reveals many features of the dynamics of
the jovian magnetosphere. The total 5eld shows the evidence of crossing the current sheet in its periodic decreases
but it has longer modulations as well. These variations
on the period of days could well re>ect changes in the
solar wind dynamic pressure as the solar wind is seen to
vary on such time scales. The current sheet also appears
to move mainly above and below the spacecraft for days
at a time. This also appears to be due to the solar wind,
in this case being due to variations in the direction of the
solar wind >ow. Finally, there are variations in the component of the 5eld across the current sheet. These variations
while brief are the most dramatic of all. They rise rapidly
to 5eld strengths even greater than the originally existing
5eld outside the current sheet. We interpret these as
large-scale reconnection events (Russell et al., 1998a,
2000c).
Fig. 29. The magnetic 5eld observed by Galileo on orbit G8 from apogee
in to about 50RJ (Russell et al., 1999a, b).
Fig. 30 shows a very strong event in which the 5eld

strength increases nearly a factor of four when the 5eld
turns southward. We interpret such a strong resultant 5eld as
due to explosive reconnection when the reconnection point
reaches a region of very low density out of the current sheet.
Here the Alfven velocity is high and the plasma can be accelerated to high speeds and reconnection can proceed very
rapidly. This process creates an X-point in the magnetic 5eld
in a swept-back meridian phase as shown in Fig. 31. These
largest events contain about 1010 Wb or about 10 times the
magnetic >ux in one of the tail lobes of the Earth but only
1025
Jupiter does not negate the model. The observed circulation is in fact quite consistent with the qualitative picture
of Vasyliunas. The Galileo, Voyager and Pioneer observations used in this review have merely made this model more
quantitative. We have not yet completely solved our earlier
posed dilemma, because we have not returned the emptied
>ux tubes into the inner magnetosphere. We examine that
process next.
7.6. Returning the 8ux to the inner magnetosphere
Fig. 30. An example of explosive reconnection as detected by Galileo.

Magnetic 5eld components are directed outward, southward and in the
corotation direction.
Fig. 31. Interpretation of the measurements shown in Fig. 29. At point

A the magnetic 5eld becomes strongly southward and strong as if the
magnetic 5eld had suddenly reconnected outside of the radial distance
of Galileo. Events with northward turnings are also seen (Russell et al.,
1998a, b).
about 0.1% of the >ux in the jovian magnetotail. At the rate

of events seen in Fig. 28, once every 4 hours, an average of
about 70; 000 Wb per second can be emptied of ions and be
prepared for return to the inner magnetosphere. It is reassuring to note that the amount of magnetic >ux that is mass
loaded at Io every second, assuming that Ios mass loading
can be approximated by a step function four Io radii across
(in the radial direction perpendicular to the >ow), is about
80; 000 Wb per second. Thus the substorm process and the
mass loading process are in approximate balance.
Putting these observations in a more global context we
show in Fig. 32 the Vasyliunas (1989) prediction for the circulation in the jovian magnetosphere. Like Dungeys models of the reconnection process at Earth, this model powers
the magnetosphere in a steady-state manner. The fact that
we 5nd a temporally varying magnetosphere at Earth or at
The above observations strongly support a model of the

continued, inexorable, outward >ow of plasma, moving
slowly at 5rst but accelerating rapidly until a radial velocity of close to 50 km=s is reached about 40RJ . The radial
plasma >ow carries magnetic >ux with it because we cannot
separate the plasma from the magnetic 5eld until reconnection takes place beyond about 50RJ in the midnight to
3 a.m. sector. Reconnection that reaches the lobes above
and below the current sheet, as these powerful reconnection
events do, create nearly empty magnetic >ux tubes that are
lighter and have a much lower beta than the mass-loaded
>ux tubes. In the rapidly rotating jovian magnetosphere
light >ux tubes are buoyant and should be able to >oat in
toward Io. If so then where are they? In fact we do see
such depleted >ux tubes. One such tube was discussed by
Kivelson et al. (1997b). Others, but smaller, are seen in the
Io torus (Russell et al., 2000e) as shown in Fig. 33. They
distinguish themselves by their increased 5eld strength. The
increase in magnetic energy density matches that expected
for a tube drained of plasma at this location in the magnetosphere. Such high-resolution data are rare in the Galileo
records and thus we cannot provide an extensive survey of
the occurrence of such tubes but we have covered the Io
torus quite thoroughly (Russell et al., 2000e). One might
wonder at the paucity of such depleted >ux tubes in the
limited records on hand as they occur only about 0.4% of
the time. Naively one might expect that the depleted returning >ux tubes would occupy half of the magnetosphere.
However, since they are not transporting plasma, the >ux
tubes can move rapidly. Thus if the depleted >ux tubes in
the Io torus move at about 200 times the outward velocity
of the outward laden >ux tubes then there is >ux balance.
If the outward velocity here is about 10 m s1 then the
inward velocity might be 2 km s1 . In fact Kivelson et
al. estimated an inward velocity much greater than this,
over 100 km s1 .
Depleted >ux tubes have also been observed between Europa and Ganymede (Russell et al., 1999a). These tubes last
longer than the tubes in the Io torus. Perhaps they are moving more slowly but because the outward >ow is more rapid
here we would expect the empty tubes to occupy a larger
fraction of the records unless they also moved inwards more
rapidly in this region. It would be desirable but is very dif5cult to get an inventory of the amount of inward moving
1026
Fig. 32. The circulation pattern and magnetic topology in Vasyliunas (1989) reconnecting jovian magnetosphere.
Fig. 33. Empty >ux tubes in the Io torus. An average magnetic 5eld
magnitude has been removed in each panel leaving only the change in
5eld magnitude. The 5eld strength during these periods is about 1500 nT.
>ux. In fact there may be much >ux moving inward with

scale sizes too small for Galileos magnetometer to detect
because of its generally poor ( 20 s) resolution.
8. Saturn, Uranus and Neptune

Our entire body of knowledge of the magnetospheres of
Saturn, Uranus, and Neptune rests on the information gained
from the >ybys of Pioneer 11, Voyager 1, and Voyager 2
for Saturn, and Voyager 2 only for Uranus and Neptune.
Expressed in terms of planetary radii the magnetospheres of
these three planets are similar in size although in absolute
terms the magnetospheres are over a factor of two smaller.
Because as we stated earlier, the Mach number of the shocks

standing in front of these magnetospheres are very strong,
the plasma behind the shocks is very hot. This produces a
high beta plasma in which magnetic pressure is low and
the dynamical processes in the magnetosheath are dominated by pressure gradients in the plasma and not magnetic
forces. Thus we do not expect reconnection to be important at these magnetospheres. We do, however, see evidence
for some reconnection in terms of magnetic 5elds along
the normal to the magnetopause (Huddleston et al., 1998a).
We also note that as the magnetic 5eld of the solar wind
weakens with distance from the sun with a constant solar
wind velocity, the electric 5eld of the solar wind becomes
even weaker with respect to the corotational electric 5eld
of each of these planets that have moderately rapid rotation
and moderately strong intrinsic magnetic 5elds. Nevertheless, it is possible that the polar regions and their extensions
the magnetotails are linked to the solar wind though reconnection, and exhibit some solar wind associated phenomena.
In fact such a dynamic event has been reported for Uranus
(Kane et al., 1991). It is unlikely that we will learn more
about the magnetospheres of Uranus and Neptune for quite
some time but we should soon have data from the Cassini
mission in orbit about Saturn beginning in July 2004 and
extending for four years. This mission includes a full complement of instruments to study the magnetosphere and its
interaction with its moons. At this writing the spacecraft and
its payload are well on their way to Saturn and performing
>awlessly.
9. Conclusions
Dynamic pressure and the strength of planetary magnetic
moments principally control the size of a planetary magnetosphere. A secondary factor is the presence of an internal
mass or energy source such as provided by Io deep in the
jovian magnetosphere. No other planet has a strong source

of plasma so deep in the magnetosphere. Reconnection plays
a signi5cant role in the dynamics of most magnetospheres
with intrinsic magnetic 5elds and possibly for some induced
magnetospheres, most notably for comets. Field-aligned currents are the pathways by which stresses are communicated
from one part of a planetary magnetosphere to another, such
as from the outer magnetosphere to the ionosphere on Earth,
or from Io to the jovian ionosphere. These currents can lead
to electric potential drops along 5eld lines. These potential
drops can accelerate charged particles into the atmosphere
to excite aurora and to decouple the magnetosphere from
the ionosphere. This process seems to be quite active at the
Earth or possibly Jupiter.
Reconnection with the interplanetary magnetic 5eld is
important in the magnetospheres of Earth and Mercury.
Theory and modeling predict that the conductivity of the
ionosphere can mitigate reconnection at the magnetopause.
However, there is evidence that reconnection at the magnetopause proceeds rapidly even when the polar cap potential
drop appears to saturate. It is important to study the Mercury magnetosphere, whose ionosphere is weak, to determine how important the ionosphere is in the energy transfer
process. Studies of Mercury suggest it is rapidly responsive
to solar wind changes. Thus Mercury may not have substorms of the same type as Earth with energy stored for later
release. Part of the delay in the terrestrial substorm may be
the time required for the reconnection region to eat its way
through the dense plasma surrounding the current layer to
reach the lower density, high Alfven velocity region on either side of the tail current sheet. The onset of terrestrial
substorms is very rapid compared to the other phases. The
jovian equivalent of substorms also has an extremely rapid
and powerful onset.
Mass loading is an important process in the solar system. It produces the obstacle that obstructs the solar wind
at comets. It contributes importantly to the solar wind interaction at Venus and Mars and it powers the jovian magnetosphere. The variability of the mass loading rate at comets
produces a spectrum of cometary behavior as comets approach and recede from the sun. Solar EUV >ux variations
a7ect the mass loading at Venus and Mars and at Jupiter
the volcanic activity of Io modulates the strength of the
interaction.
The terrestrial magnetosphere is characterized by a series
of storage and energy releases over a period of hours that
has been incorrectly termed a substorm, incorrectly because
substorms are not small magnetospheric storms. Occasionally, a special set of conditions occur in the solar wind that
produces a strong steady coupling of the magnetosphere to
the solar wind and a magnetic storm ensues in which a ring
current circling the magnetosphere about 35 Re develops,
storing and dissipating a tremendous amount of energy in
the Earths upper atmosphere. Likewise Jupiter is encircled
with an enormous ring of current but the variations in this
ring are relatively small, at least as seen in the inner mag-
1027
netosphere. Thus the inner magnetosphere of Jupiter seems

relatively immune from storms despite the variability of Ios
volcanoes and the tremendous >ux of energy intercepted by
the jovian magnetosphere. It is possible that Jupiter has dif5culty in tapping this energy reservoir because reconnection with solar wind magnetic 5eld is ine6cient. Jupiter is
unlike the Earth in that its corotation electric 5eld, caused
by the rotation of the ionosphere, which is in turn coupled
to the jovian atmosphere, dominates any possible electric
5eld provided by the solar wind. Thus while there may be
phenomena in the region of the polar cap a7ected by the
interplanetary electric 5eld, it is unlikely that phenomena
deep in the magnetosphere are so a7ected. Of course, since
the strength of the solar wind dynamic pressure a7ects the
size of the magnetosphere and the direction of the solar
wind a7ects its orientation and the possible beaming and
shadowing of radio signals, it is possible that there are solar wind correlations seen from Earth in the various radio
phenomena.
Plasma circulates in all planetary magnetospheres. At
Mercury and the Earth as well as the unmagnetized bodies
with atmospheres the direction of the interplanetary magnetic 5eld is critical in the control of the circulation. However, at Jupiter this circulation is driven from within. The
mass loading at Io sets in motion a gradual out>ow of plasma
that >ows radially even faster as it corotates with the ionosphere. Eventually beyond about 50RJ in the post midnight
sector reconnection empties the magnetic 5eld lines creating
magnetized islands that >ow down the jovian tail and empty
>ux tubes that return to the inner magnetosphere. These >ux
tubes appear to be both small and fast moving in order to return the requisite magnetic >ux to the inner magnetosphere,
yet be rarely seen. Evidence for this process and the rapid
motion of these >ux tubes is found. The rate at which >ux
tubes are loaded at Io and emptied in the distant magnetosphere appears to match quite well.
The magnetospheres of Saturn, Uranus and Neptune seem
quite benign compared to that of Jupiter. Saturn has mass
sources in the rings, in the moon Enceladus and at Titan.
However, Titan lies close to the outer edge of the magnetosphere, simplifying the process of shedding the mass
added to the magnetosphere by Titan. The mass added by
the rings and Enceladus are certainly measurable and signi5cant but is unlikely to cause any dynamical phenomena
akin to those in the jovian magnetosphere. Uranus has exhibited some evidence for a dynamic magnetosphere but it is
di6cult to tell if this just had some e7ect on the tail plasma
or whether this event had an e7ect on the inner portions of
the magnetosphere. At Neptune there is little evidence for
activity within the magnetosphere, albeit there is some reconnection at the magnetopause just as there is at the other
planets.
The comparison of the behaviors of the magnetospheres
of Mercury, Jupiter and the Earth has thus far been very
bene5cial and has given us new insights into the workings of each. With the advent of new data from Mercury
1028
with the Messenger Discovery mission and the continued

analysis of Galileo data we expect further such insights.
Most importantly we will soon receive data from Cassini
in orbit about Saturn. This mission should raise the Saturn
magnetosphere to the level of understanding that we now
have for the terrestrial and jovian magnetospheres and likewise add to our understanding of magnetospheres as plasma
systems, engines generating aurora, heating, acceleration
and radio emissions.
Acknowledgements
I am extremely fortunate to have worked with Janet G.
Luhmann, Margaret G. Kivelson and R.L. McPherron and
many others of the worlds experts in the study of planetary
magnetospheres, as well as some very bright graduate students. This work was supported by the National Aeronautics
and Space Administration through the Jet Propulsion Laboratory.
References
Acuna, M.H., Connerney, J.E.P., Wasilewski, P., Lin, R.P., Anderson,
K.A., Carlson, C.W., McFadden, J., et al., 1998. Magnetic 5eld and
plasma observations at Mars: initial results of the Mars Global Surveyor
mission. Science 279, 16761680.
Bagenal, F., 1985. Plasma conditions inside Ios orbit: Voyager
measurements. J. Geophys. Res. 90, 311324.
Bagenal, F., 1994. Empirical model of the Io plasma torus: Voyager
measurements. J. Geophys. Res. 99, 11,04311,062.
Bagenal, F., 1997. The ionization source near Io from Galileo wake data.
Geophys. Res. Lett. 24, 21112114.
Belcher, J.W., 1983. The low energy plasma in the jovian magnetosphere.
In: Dessler, A.J. (Ed.), Physics of the Jovian Magnetosphere.
Cambridge University Press, New York, pp. 69105.
Brecht, S.H., 1997. Solar wind proton deposition into the Martian
atmosphere. J. Geophys. Res 102, 11,28711,294.
Brown, M.E., Bouchez, A.H., 1997. The response of Jupiters
magnetosphere to an outburst on Io. Science 278(5336), 268271.
Burton, R.K., McPherron, R.L., Russell, C.T., 1975. An empirical
relationship between interplanetary conditions and Dst. J. Geophys.
Res. 80, 42044214.
Clarke, J.T., Ballester, G.E., Trauger, J., Evans, R., Connerney, J.E.P.,
et al., 1996. Far-ultraviolet imaging of Jupiters aurora and the Io
footprint. Science 274, 404409.
Colburn, D.S., Mihalov, J.D., Sonett, C.P., 1971. Magnetic observations
of the lunar cavity. J. Geophys. Res. 76, 2940.
Dessler, A.J., Parker, E.N., 1959. Hydromagnetic theory of magnetic
storms. J. Geophys. Res. 64, 22392259.
Dubinin, E.M., Lundin, R., Pissarenko, N., Barabash, S., Zakharov, A.,
Koskinen, H., Schwingenschuh, K., Yeroshenko, Y., 1990. Indirect
evidence for a gas=dust torus along the Phobos orbit. Geophys. Res.
Lett. 17, 861864.
Dungey, J.W., 1961. Interplanetary magnetic 5eld and the auroral zones.
Phys. Rev. Lett. 6, 4748.
Dungey, J.W., 1963. The structure of the exosphere or adventures
in velocity space. In: Dewitt, C., Hieblolt, J., Lebeau, A. (Eds.),
Geophysics: The Earths Environment. Gordon and Breach, New York,
pp. 505550.
Frank, L.A., Paterson, W.R., 1999a. Intense electron beams observed

at Io with the Galileo spacecraft. J. Geophys. Res. 104, 28,657
28,669.
Frank, L.A., Paterson, W.R., 1999b. Observations of plasmas in the
Io torus with the Galileo spacecraft. J. Geophys. Res. 105, 16,017
16,034.
Frank, L.A., Paterson, W.R., Ackerson, K.L., Vasyliunas, V.M., Coroniti,
F.V., Bolton, S.J., 1996. Plasma observations at Io with the Galileo
spacecraft. Science 274, 394395.
Freeman, J.W., Ibrahim, M., 1975. Lunar electric 5elds, surface potential
and associated plasma sheaths. Moon 14, 103114.
Grunwaldt, H., et al., 1997. Venus tail ray observation near Earth.
Hill, T.W., 1994. The size and shape of convection cells in the jovian
magnetosphere. In: Solar System Plasmas in Space and Time. American
Geophysical Union, Washington, DC, pp. 199 205.
Hill, T.W., Dessler, A.J., Goertz, C.K., 1983. Magnetospheric models. In:
Dessler, A.J. (Ed.), Physics of the Jovian Magnetosphere. Cambridge
University Press, New York, pp. 353394.
Huddleston, D.E., Russell, C.T., Kivelson, M.G., Khurana, K.K., Bennett,
L., 1998a. Location and shape of the Jovian magnetopause and bow
shock. J. Geophys. Res. 103, 20,07520,082.
Huddleston, D.E., Strangeway, R.J., Blanco-Cano, X., Russell, C.T.,
Kivelson, M.G., Khurana, K., 1999. Mirror-mode structures at the
Galileo-Io >yby: instability criterion and dispersion analysis. J.
Geophys. Res. 104, 17,47917,489.
Huddleston, D.E., Strangeway, R.J., Warnecke, J., Russell, C.T., Kivelson,
M.G., 1998b. Ion cyclotron waves in the Io torus: wave dispersion,
free energy analysis, and SO+
2 source rate estimates. J. Geophys. Res.
103, 19,88719,889.
Intriligator, D.S., Miller, W.D., 1982. First evidence of a Europa plasma
torus. J. Geophys. Res. 87, 80818090.
Johnstone, A.D. (Ed.), 1991. Cometary Plasma Processes. American
Geophysical Union, Washington, DC, 364pp.
Kane, M., Mauk, B.H., Keath, E.P., Krimigis, S.M., 1991. Structure and
dynamics of the Uranian magnetotail: results from hot plasma and
magnetic 5eld observations. J. Geophys. Res. 96, 11,48511,499.
Kane, M., Mauk, B.H., Keath, E.P., Krimigis, S.M., 1995. Hot ions in
Jupiters magnetodisc: a model for Voyager 2 low-energy charged
particle measurements. J. Geophys. Res. 100, 19,47319,486.
Kivelson, M.G., Bargatze, L.F., Khurana, K.K., Southwood, D.J., Walker,
R.J., Coleman Jr., P.J., 1993. Magnetic 5eld signatures near Galileos
closest approach to Gaspra. Science 261, 331334.
Kivelson, M.G., Khurana, K.K., Russell, C.T., Walker, R.J., Warnecke, J.,
Coroniti, F.V., Polanskey, C., Southwood, D.J., Schubert, G., 1996a.
Discovery of Ganymedes magnetic 5eld by the Galileo spacecraft.
Nature 384, 537541.
Kivelson, M.G., Khurana, K.K., Walker, R.J., Warnecke, J., Russell, C.T.,
Linker, J.A., Southwood, D.J., Polanskey, C., 1996b. Ios interaction
with the plasma Torus: galileo magnetometer report. Science 274,
396398.
Kivelson, M.G., Khurana, K.K., Coroniti, F.V., Joy, S., Russell, C.T.,
Walker, R.J., Warnecke, J., Bennett, L., Polanskey, C., 1997a. The
magnetic 5eld and magnetosphere of Ganymede. Geophys. Res. Lett.
24, 21552158.
Kivelson, M.G., Khurana, K.K., Russell, C.T., Walker, R.J., 1997b.
Intermittent short-duration magnetic 5eld anomalies in the Io
torus: evidence for plasma interchange. Geophys. Res. Lett. 24,
21272130.
Lin, R.P., Mitchell, O.L., Curtis, D.W., Anderson, K.A., Carlson, C.W.,
McFadden, J., Acuna, M.H., Hood, L.L., Binder, A., 1998. Lunar
surface magnetic 5elds and their interaction with the solar wind: results
from lunar prospector. Science 281, 14801484.
Luhmann, J.G., Russell, C.T., Tsyganenko, N.A., 1998. Disturbances in
Mercurys magnetosphere: are the Mariner 10 substorms simply
driven? J. Geophys. Res. 103, 91139119.

Luhmann, J.G., Tatrallyay, M., Pepin, R.O. (Eds.), 1992. Venus and
Mars: Atmospheres, Ionospheres and Solar Wind Interaction. American
Geophysical Union, Washington, DC, 430pp.
Mauk, B.H., Williams, D.J., McEntire, R.W., Khurana, K.K., Rostoker,
J.G., 1999. Storm-like dynamics of Jupiters inner and middle
magnetosphere. J. Geophys. Res. 104, 22,75922,778.
McComas, D., Spence, H.E., Russell, C.T., Saunders, M.A., 1986. The
average magnetic 5eld draping and consistent plasma properties of the
Venus magnetotail. J. Geophys. Res. 91, 79397953.
Niedner Jr., M.B., Brandt, J., 1978. Interplanetary gas XXIII. Plasma
tail disconnection events in comets: evidence for magnetic 5eld
reconnection at interplanetary sector boundaries? Astrophys. J. 223
(2), 655670.
Pontius Jr., D.H., Hill, T.W., 1989. Rotation driven plasma transport: the
coupling of macroscopic motion and microdi7usion. J. Geophys. Res.
94, 15,04115,053.
Prange, R., Rego, D., Pallier, L., Connerney, J.E.P., Zarka, P., Queinnec,
J., 1998. Detailed study of FUV Jovian auroral features with the
post-COSTAR HST faint object camera. J. Geophys. Res. 103, 20,
19520,215.
Prange, R., Rego, D., Southwood, D., Zarka, P., Miller, S., Ip, W.-H.,
1996. Rapid energy dissipation and variability of the Io-Jupiter
electrodynamic circuit. Nature 379, 323325.
Riedler, W., et al., 1989. Magnetic 5elds near Mars: First results. Nature
341, 604607.
Russell, C.T., 1991. Venus Aeronomy. Kluwer Academic Publishers,
Dordrecht, 489pp.
Russell, C.T., Lichtenstein, B.R., 1975. On the source of lunar limb
compressions. J. Geophys. Res. 80, 47004711.
Russell, C.T., McPherron, R.L., 1973. The magnetotail and substorms.
Space Sci. Rev. 15, 205266.
Russell, C.T., McPherron, R.L., Burton, R.K., 1974. On the cause of
geomagnetic storms. J. Geophys. Res. 79, 11051109.
Russell, C.T., Bagenal, F., Cheng, A.F., Ip, W.-H., Roux, A., Smyth,
W.H., Bolton, S.J., Polanskey, C.A., 1997. Ios interaction with the
Jovian magnetosphere. EOS Trans. AGU 78, 93.
Russell, C.T., Baker, D.N., Slavin, J.A., 1988. The magnetosphere of
Mercury. In: Vilas, F., Chapman, C., Matthews, M. (Eds.), Mercury.
Univ. Arizona Press, Tucson, pp. 514561.
Russell, C.T., Blanco-Cano, X., Strangeway, R.J., 2001b. Ultra
low frequency waves in the jovian magnetosphere: causes and
consequences. Planet Space Sci. 49, 291301.
Russell, C.T., Huddleston, D.E., Khurana, K.K., Kivelson, M.G., 1999a.
The >uctuating magnetic 5eld in the middle Jovian magnetosphere:
initial Galileo observations. Planet. Space Sci. 47, 133142.
Russell, C.T., Huddleston, D.E., Khurana, K.K., Kivelson, M.G., 1999c.
Structure of the magnetodisk current sheet: Initial Galileo observations.
Planet Space Sci. 47, 11011109.
Russell, C.T., Huddleston, D.E., Khurana, K.K., Kivelson, M.G., 2000b.
Waves and >uctuations in the jovian magnetosphere. Adv. Space Res.
26, 14891498.
Russell, C.T., Huddleston, D.E., Strangeway, R.J., Blanco-Cano, X.,
Kivelson, M.G., Khurana, K.K., Frank, L.A., Paterson, W., Gurnett,
D.A., Kurth, W.S., 1999b. Mirror-mode structures at the Galileo-Io
>yby: observations. J. Geophys. Res. 104, 17,47117,477.
Russell, C.T., Khurana, K.K., Huddleston, D.E., Kivelson, M.G., 1998a.
Localized reconnection in the near Jovian magnetotail. Science 280,
10611064.
Russell, C.T., Khurana, K.K., Kivelson, M.G., Huddleston, D.E., 2000c.
Substorms at Jupiter: galileo observations of transient reconnection in
the near tail. Adv. Space Res. 26, 14991504.
Russell, C.T., Kivelson, M.G., Khurana, K.K., Huddleston, D.E., 1998b.
Magnetic >uctuations close to Io: ion cyclotron and mirror mode wave
properties. Planet. Space Sci. 47, 143150.
Russell, C.T., Kurth, W.S., Kivelson, M.G., Gurnett, D.A., 2000e.
Implications of depleted >ux tubes in the jovian magnetosphere.
1029
Russell, C.T., Saunders, M.A., Phillips, J.L., Fedder, J.A., 1986. Near-tail
reconnection as the cause of cometary tail disconnections. J. Geophys.
Res. 91, 14171423.
Russell, C.T., Wang, Y.L., Blanco-Cano, X., Strangeway, R.J., 2001a.
The Io mass loading disk: constraints provided by ion cyclotron wave
observations. J. Geophys. Res., in press.
Sauer, K., Dubinin, E., Baumgartel, K., Tarasov, V., 1998. Low-frequency
electromagnetic waves and instabilities within the Martian bi-ion
plasma. Earth, Planets, Space 50, 269278.
Saunders, M.A., Russell, C.T., 1986. Average dimension and magnetic
structure of the distant Venus magnetotail. J. Geophys. Res. 91, 5589
5604.
Sckopke, N., 1966. A general relation between the energy of trapped
particles and the disturbance 5eld near the Earth. J. Geophys. Res. 71,
3125.
Scurry, L., Russell, C.T., 1991. Proxy studies of energy transfer in the
magnetosphere. J. Geophys. Res. 96, 95419548.
Scurry, L., Russell, C.T., Gosling, J.T., 1994. Geomagnetic activity and
the beta dependence of the dayside reconnection rate. J. Geophys. Res.
99, 14,81114,814.
Siscoe, G.L., Ness, N.F., Yeates, C.M., 1975. Substorms on Mercury? J.
Geophys. Res. 80, 43594363.
Siscoe, G.L., Summers, D., 1981. Centrifugally driven di7usion of iogenic
plasma. J. Geophys. Res. 86, 8471.
Slavin, J.A., Smith, E.J., Spreiter, J.R., Stahara, S.S., 1985. Solar wind
>ow about the outer planets: gas dynamic modeling of the Jupiter and
Saturn bow shocks. J. Geophys. Res. 90, 62756286.
Smith, E.J., Connor, B.V., Foster Jr., G.J., 1975. Measuring the magnetic
5elds of Jupiter. IEEE Trans. Magnetics 11, 962980.
Southwood, D.J., Kivelson, M.G., 1987. Magnetospheric interchange
instability. J. Geophys. Res. 92, 109.
Southwood, D.J., Kivelson, M.G., 1989. Magnetospheric interchange
motion. J. Geophys. Res. 94, 299308.
Spencer, J.R., Schneider, N.M., 1996. Io on the eve of the Galileo mission.
Ann. Rev. Earth Planet, Sci. 24, 125190.
Spreiter, J.R., Summers, A.L., Alksne, A.Y., 1966. Hydromagnetic >ow
around the magnetosphere. Planet. Space Sci. 14, 223253.
Spreiter, J.R., Summers, A.L., Rizzi, A.W., 1970. Solar wind >ow past
nonmagnetic planets: Venus and Mars. Planet. Space Sci. 18, 1281
1299.
Stahara, S.S., Rachiele, R.R., Spreiter, J.R., Slavin, J.A., 1989. A
three dimensional gas dynamic model for the solar wind >ow past
non-axisymmetric magnetospheres: application to Jupiter and Saturn.
J. Geophys. Res. 94, 13,35313,365.
Strangeway, R.J., Elphic, R.C., Peria, W.J., Carlson, C.W., 2000.
FAST observations of electromagnetic stresses applied to the polar
ionosphere. In: Ohtani, S.-I., Fujii, R.-I., Lysak, R., Hesse, M. (Eds.),
Magnetospheric Current System, AGU Monograph, Washington,
2129.
Thomas, N., 1993. The variability of the Io torus. J. Geophys. Res.
18,73718,750.
Tsyganenko, N.A., 1996. E7ects of the solar wind conditions on the global
magnetospheric con5guration as deduced from data-based models. In:
Proceedings of Third International Conference on Substorms (ICS-3)
Eur. Space Agency Spec. Publ., ESA SP-389, pp. 181-195.
Vaisberg, O., Luhmann, J.G., Russell, C.T., 1990. Plasma observations
of the solar wind interaction with Mars. J. Geophys. Res. 95,
14,84114,852.
Vaisberg, O., Smirnov, V., Fedorov, A.O., Avanov, L., Dunjushkin, F.,
Luhmann, J.G., Russell, C.T., 1994. Structure of the Venus Tail. In:
Burch, J.L., Waite, J.H. Jr. (Eds.), Solar System Plasmas in Space and
Time. American Geophysical Union, Washington, DC, pp. 207220.
Vasyliunas, V.M., 1989. Maximum scales for preserving >ux tube content
in radial di7usion driven by interchange motions. Geophys. Res. Lett.
16, 14651468.
Walker, R.J., Russell, C.T., 1985. Flux transfer events at the Jovian
magnetopause. J. Geophys. Res. 90, 73977404.
1030
Wang, Y.L., Russell, C.T., Raeder, J., 2000. The Io mass loading disk:
model calculations. J. Geophys. Res., submitted for publication.
Williams, D.J., Thorne, R.M., Mauk, B., 1999. Energetic electron beams
and trapped electrons at Io. J. Geophys. Res. 104, 14,73914,753.
Wilson, J.K., Schneider, N.M., 1999. Ios sodium directional feature:

evidence for ionospheric escape. J. Geophys. Res. 104, 16,567.
Yamamoto, T., Tsuruda, K., 1998. The Planet B mission. Earth, Planets
and Space 50, 175181.

The Dynamics of Planetary Magnetospheres

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Dynamics of Planetary Magnetospheres

Uploaded by

Copyright:

Available Formats

Planetary and Space Science 49 (2001) 10051030

The dynamics of planetary magnetospheres

Corresponding author. Tel. +1-310-825-3188; fax: +1-310-206-8042.

magnetospheres, the Venus and cometary magnetospheres,

c 2001 Elsevier Science Ltd. All rights reserved.

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

Planets. This paper is not intended as a complete review of

Fig. 1. The cross-section of the magnetosphere and the bow shock in

where is the polytropic index, cs is the sound velocity and

where MA is the Alfven or intermediate Mach number.

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

Finally, the ratio of the second and third terms is

where nsw is the solar wind proton number density in

Fig. 2. Sources of viscosity at the magnetopause. Schematic illustrations

which an entire tube of magnetosheath plasma crosses the

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

the poles. The reconnected 5eld line is added to the dayside

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

plasma sheet, labelled neutral sheet current, is an extension

Fig. 5. The coupling of the magnetosphere to the ionosphere via

Fig. 5 illustrates the physics involved in the coupling of

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

Fig. 6. Development of a geomagnetic storm. The top panel shows the

5eld must be strong and steadily southward for several hours

riod plotted even though the character of the interplanetary

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

Fig. 7. Schematic illustration of a substorm. The top panel shows the

Nevertheless, such bodies could have remanent magnetism,

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

interplanetary magnetic 5eld perpendicular to the solar wind

Fig. 9. The average con5guration of the magnetic 5eld in the Mercury

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

Fig. 11. The con5guration of the Ganymede magnetosphere in Jupiters

magnetosphere is more a7ected by the external (jovian)

6. Venus, Mars and comets

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

little into the ionosphere. In the case of a high solar wind

wind velocity requires about 3 GW, a number that is a small

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

Fig. 13. A cometary tail disconnection event, seen in comet Morehouse

tail causes the disconnection as sketched in Fig. 14 (Russell

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

magnetic 5eld in the noon-midnight meridian shown in

the magnetic 5eld perpendicular to the solar wind >ow and

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

If the mass loading rate is large enough, then the addition

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

loading rate is smaller than generally believed, or the mass

Fig. 18. Trajectory of Galileo on its 5rst >yby of Io illustrating the

emphasizes that the main mass loading region is about 2 Io

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

terson, 1999b) that was picked up at very low velocities is

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

Fig. 20. Io torus density as observed by Voyager 1 (Bagenal, 1994)

the plasma would take about 3 months to reach a distance

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

factor. Nevertheless, this outward motion, no matter what

While they envision an external trigger and possibly radial

The journey outward is not steady. Evidence of the

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

very consistent pattern of acceleration with distance is seen

C.T. Russell / Planetary and Space Science 49 (2001) 10051030

density decrease is due to particle loss along the 5eld, the

that there is slow out>ow over nearly 360 with a narrow