Moon–Magnetosphere Interactions: a Tutorial

Advances in Space Research 33 (2004) 2061–2077 www.elsevier.com/locate/asr

Moon–magnetosphere interactions: a tutorial

M.G. Kivelson a,b,*

a University of California, Institute of Geophysics and Planetary Physics, UCLA, Los Angeles, CA 90095-1567, USA b Department of Earth and Space Sciences, UCLA, Los Angeles, CA 90095-1567, USA

Received 8 May 2003; received in revised form 13 August 2003; accepted 18 August 2003

Abstract

The interactions between the Galilean satellites and the plasma of the Jovian magnetosphere, acting on spatial scales from that of ion gyroradii to that of magnetohydrodynamics (MHD), change the plasma momentum, temperature, and phase space distribution functions and generate strong electrical currents. In the immediate vicinity of the moons, these currents are often highly structured, possibly because of varying ionospheric conductivity and possibly because of non-uniform pickup rates. Ion pickup changes the velocity space distribution of energetic particles, f (v), where v is velocity. Distributions become more anisotropic and can become unstable to wave generation. That there would be interesting plasma responses near Io was fully anticipated, but one of the surprises of Galileo’s mission was the range of effects observed at all of the Galilean satellites. Electron beams and an assortment of MHD and plasma waves develop in the regions around the moons, although each interaction region is different. Coupling of the plasma near the moons to the Jovian ionosphere creates auroral footprints and, in the case of Io, produces a leading trail in the ionosphere that extends almost half way around Jupiter. The energy source driving the auroral signatures is not fully understood but must require field aligned electric fields that accelerate elections at the feet of the flux tubes of Io and Europa, bodies that interact directly with the incident plasma, and at the foot of the flux tube of Ganymede, a body that is shielded from direct interaction with the background plasma by its magnetospheric cavity. Ó 2004 COSPAR. Published by Elsevier Ltd. All rights reserved.

Keywords: Moon–magnetosphere interactions; Jupiter; Jupiter’s moons

1. Introduction 2. Properties of importance to the interaction

The major moons of Jupiter and Saturn, solid bodies Various intrinsic properties of the moons critically comparable in scale with Earth’s moon (whose radius is affect their interactions with the plasma that flows onto 1734 km), are embedded in the flowing plasma of a them. Of particular importance are the internal mag- planetary magnetosphere. Some relevant properties are netic fields of the moons, which may be permanent and/ presented in Table 1. Interactions with the surroundings or induced. Neutrals liberated from some of the moons depend on details of the bodies and of the plasma that can upon ionization become the source of a plasma that flows onto them. This tutorial presentation first intro- is not only locally denser than the ambient plasma of the duces the physical processes that must be considered in magnetosphere but may contain ion species not present understanding how the moons interact with the system in it. The electrical conductivity of the surfaces and the and then presents and interprets selected measurements interiors of the moons, their ionospheres, and the plas- in the vicinity of the moons. ma clouds that surround them are critical elements of the interaction. There are interesting similarities and differences between the interaction regions (magnetospheres or equivalent) that form around the planets and the * Tel.: +1-310-825-3435; fax: +1-310-206-8042. interaction regions that surround the Galilean moons. E-mail address: [email protected] (M.G. Kivelson). Critical is the fact that planets are embedded in a

Table 1 Selected properties of the Galilean moons and of Titan Io Europa Ganymede Callisto Titan Radius (km) 1818 1560 2634 2400 2575 Density (kg m3) 3530 2990 1940 1851 1881 Orbital period 42.46 (h) 85.22 (h) 171.71 (h) 400.54 (h) 15.9 (days) super-magnetosonic plasma, the solar wind, whereas the Galilean moons are embedded in a sub- or trans-magnetosonic plasma, in this case Jupiter’s magnetospheric plasma, that overtakes them from their (orbital) trailing side. Properties of the magnetospheric plasma at the orbit of a moon that affect the interaction include: the MHD Mach numbers (fast, intermediate, and slow), the plasma beta, the Alfven conductance (which char- acterizes the effectiveness of the plasma in carrying current across the magnetic field), and the relatively predictable orientation and temporal variation of the external magnetic field. We shall return to a discussion of these features but let us first consider how the large- scale perturbations relate to MHD wave modes produced in the interaction region.

3. Pertinent MHD wave properties

The group velocity of a wave constrains the regions of space within which perturbations can be imposed by its action. Fig. 1 shows the properties of the group velocity of MHD waves for two different assumed plasma conditions in a uniform magnetic field (Kivelson, 1995). In the plasma rest frame, the fast mode carries information in all directions relative to the background field, B,and the speed depends on the angle relative to B. The Alfven wave or intermediate mode carries information strictly along the background field. The slow mode carries information in directions close to the background field but, like the fast mode, it does not carry electrical current along the field direction. Only the Alfven wave carries field-aligned current! Different types of perturbations are imposed by different wave modes. The fast mode is compressional. Fig. 1. Group velocities of the MHD fast (F), intermediate (I), and Thermal and magnetic pressure increase and decrease in slow (S) mode waves plotted vs. direction of wave vector relative to the phase producing pressure gradients that exert forces on background magnetic field. Labels above the plots indicate that in the upper (lower) panel the Alfven speed, V , exceeds (is less than) the the plasma. The slow mode is also compressional but A sound speed, cs. thermal and magnetic pressure vary in antiphase and therefore the total pressure (thermal plus magnetic) changes little, although density changes develop. The 4. Flow in the interaction region: analogy to planetary intermediate mode or Alfven wave does not change the magnetospheres field magnitude but plays an important role because it carries field-aligned current that couples regions sepa- There are similarities and differences between the in- rated along the background field direction. Where the teractions that occur at the moons and the interactions field changes direction without changing its magnitude, between the solar wind and the planets. In both cases, this mode is present. Its action reaccelerates downstream the flow slows when the plasma incident from upstream flow to the speed of the incident plasma. first senses the presence of an obstacle to the flow. In the M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2063 super-magnetosonic solar wind, the slowing occurs only the upstream boundary is the locus defined by abrupt downstream of a standing fast magnetosonic bow shock, bends of the magnetic field. Downstream of this surface whether the interaction is with a magnetized planet lies the region coupled to the moon by field-aligned (Mercury, Earth, Jupiter, for example) or an unmagne- currents. Because field-aligned current is carried only by tized body with an atmosphere (Venus, for example). the Alfven wave, whose group velocity in the plasma rest Downstream of the bow shock and within planetary frame is along B, the perturbations carried by this magnetospheres, flows are typically sub-magnetosonic. mode appear downstream of a front represented sche- Within Jupiter’s magnetosphere magnetospheric plasma matically in Fig. 3. The angle of bendback, measured 1 approximately corotates with Jupiter and overtakes the relative to )B is given by a ¼ tan ðu=VAÞ in terms of moons, whose Keplerian orbital speeds are smaller than the flow velocity u and the Alfven speed of the back- the speed of plasma flow along the orbits. The relative ground plasma VA. In spacecraft measurements, the flow is sub-magnetosonic, so perturbations that slow signature is a field rotation with significant jdBxj where and divert the flow as the plasma approaches a moon x^ ¼ u=u. The surface defined by the field rotation is re- develop gradually. The need to decelerate the flow ferred to as an Alfven wing (Drell et al., 1965). Con- across a standing shock wave is absent. Indeed, no up- finement to a region downstream of a tilted front is stream shocks have been observed in the vicinity of the sometimes discussed in terms of wave characteristics moons on Galileo’s multiple passes, nor was a shock (Neubauer, 1980). Fig. 4 shows schematically the inter- identified in the, vicinity of Titan on the Voyager 1 pass. action region viewed from downstream in the flow. Mach numbers >1 may occur near Callisto at times Current flows into moon on the side facing toward when it is crossing the magnetic equator, but none of the Galileo encounters occurred in this plasma regime. Fig. 2 represents the qualitative behavior of the plasma flow near a moon. Both, at the moons and the planets, the flow slows when the plasma incident from upstream first senses the presence of an obstacle to the flow which, in the super- magnetosonic solar wind, occurs only downstream of a standing fast magnetosonic bow shock. Many features of the diversion of the flow behind the bow shock can be understood as analogues of arguments that we will in- troduce in the context of the interactions at the moons. An element of the interaction that is unique to the moons and of particular interest is the perturbation that couples the interaction region to Jupiter’s ionosphere. Where the flow slows, the field curvature changes, and correspondingly currents transverse to the background Fig. 3. Schematic representation of an interaction region. The pat- field develop. Currents must be divergence-free, so the terned obstacle is embedded in a background plasma threaded by transverse currents of limited spatial extent must link to vertical field lines (dashed). Alfven waves travel along the field at speed VA while the plasma flows across the field at speed u. Some perturbed field-aligned currents, flowing toward the interaction field lines (heavy lines) are illustrated (but bends out of the plane are region on the side closer to Jupiter and away on the side not represented). Dotted lines bound the region (gray) containing more remote from Jupiter. In the context of interactions, perturbations linked to Alfven waves.

Fig. 4. Schematic view looking from downstream in the flow at a Fig. 2. Schematic of slowing and accelerating flow around a moon section of the interaction region defined by B and the direction radial (black circle) effected by fast and slow mode perturbations. Flow is ralative to Jupiter. Currents flow into the moon on the Jupiter side, fastest where the surroundings are darkest. closing in the vicinity of the moon. 2064 M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077

Jupiter and out on the other side, closing through dense ionized material at the moon. The configuration of the sub-Alfvenic interaction region appears superficially very different from that relevant to the super-Alfvenic solar wind interaction with planets, but some of the differences are predominantly geometric. On field lines connected to the solar wind, the asymptotic planetary magnetopause is the locus of a converting kink of the reconnected magnetic field (an Alfven wing), a surface within which are found the field- aligned currents that link the external plasma to the Fig. 6. (J. Clarke, Space Telescope Image) Jupiter’s auroral ionosphere ionosphere of the planet. The kink propagates into the observed in UV emissions. The bright ring is the main auroral oval and localized emissions at the feet of the flux tubes through Io, Europa, external plasma at a speed VA while being swept anti- sunward at the (faster) magnetosheath flow speed: and Ganymede are labeled. Vm0sheath. As illustrated in Fig. 5, the angle subtended can be parametrized by an Alfvenic Mach number ionosphere are the sources of localized ultraviolet MA ¼ Vm0sheath=VA > 1, thereby producing a boundary emissions that are observed equatorward of the main 1 bent asymptotically at an angle a tan MA > 45°. auroral oval. An extended region of emission extends The schematic Figs. 3 and 4 show field aligned cur- ahead of Io’s footprint, most probably arising from rents linked to the moon but do not follow them far out currents that link the slowed plasma of Io’s wake with into the planetary magnetosphere. In the MHD limit, Jupiter’s ionosphere (Hill and Vasyliunas, 2002; Dela- the continuity equation rj þ oq=ot ¼ 0 reduces to mere et al., 2003). rj ¼ 0, i.e., currents are divergenceless. Thus we must consider where they close away from the moons. Some of the current is likely to be diverted from the flux tubes 5. Beyond MHD: parallel electric fields to close across B through plasma within the torus, particularly near its outer boundary where there are Although it is evident that field-aligned current con- reasons to believe that some of the wave power is re- nects the moons with Jupiter’s ionosphere, the question flected. However, much of the current flows along the of how the current is carried from Io into the Jovian background field all the way to Jupiter and consequently ionosphere is somewhat less apparent. Near Io, current the conductivity of the Jovian ionosphere is an addi- carriers can be found aplenty in the torus plasma, con- tional parameter that contributes to the full description fined by centrifugal acceleration to regions near the of the interaction region (Hill et al., 1983). This con- centrifugal equator. In Jupiter’s ionosphere, there are clusion rests on unambiguous evidence from the images ions and electrons confined by gravity. But, as illus- of the Jovian aurora (Clarke et al., 2002) acquired by the trated in Fig. 7, there is not much plasma in between. Hubble Space Telescope near Earth (see Fig. 6). By Thus, one needs to understand how current is carried modeling the magnetic field lines through the Galilean between these regions. moons, one can confirm that the footprints in the Jovian Fig. 8 shows that field-aligned (or parallel) electric fields develop and these serve to accelerate electrons through the low-density regions. The need for parallel

Fig. 7. Jupiter and the Io flux tube (not to scale) showing concentration Fig. 5. The asymptotic structure of a planetary magnetopause. Beyond of plasma near the magnetic equator as a result of the field-aligned the cusp, the magnetosheath flow (Vmsh) becomes super-Alfvenic and component of the centrifugal acceleration (gray arrows) and near the the magnetopause flare is partially governed by an angle (shown by surface of Jupiter as a result of the field-aligned component of the double lines) related to the Alfvenic Mach number in a super-Alfvenic gravitational force (black arrows). Little plasma is present in the re- flow. gions between. M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2065

Fig. 9. Schematic neutral (gray circle) newly ionized by photo-ionization (photon energy shown as hm) in a magnetized plasma flowing at Fig. 8. As for Fig. 7, but with arrows showing the parallel electric field velocity u. (Here u is the ambient flow velocity which may differ from needed to accelerate current-carrying electrons that couple the moon corotation speed.) Both ion and electron join the flow and their gy- and the ionospheric footprint. The field is toward (away) from the rocenters move at u, but the presence of a convection electric field leads ionosphere at the higher (lower) latitude side of the flux tube. to a displacement of the gyrocenter of the ion relative to the gyrocenter of the electron. (The ion gyrocenter follows the dotted arrow in the illustration.) electric fields (Ek) to close currents is familiar from Earth’s auroral regions and they develop for the same reasons in both cases. Field-aligned electric fields are momentum must be conserved. The momentum that required when currents must flow through a region in goes into new ions is extracted from the background which there is a deficit of current-carrying electrons. plasma, reducing u. The new ions acquire gyrospeeds Assuming that ion velocities are much smaller than equal to the flow speed (see Fig. 9) but their Vk does not electron velocities, one can approximate the current change (e.g., Goertz, 1980). This means that the distri- density jk as jk ¼neeVek where ne is the electron den- butions become increasingly anisotropic and possibly sity, Vke is the parallel electron velocity, and e is the unstable to wave generation. The plasma temperature magnitude of the electron charge. Large parallel velocity (or second moment of the distribution) is also affected, can compensate for small ne and thus satisfy the demand with the perpendicular temperature increasing if in the for current. Accelerated electrons moving downward background plasma vthermal < uplasma and decreasing if in into the ionosphere produce the ionospheric signatures background plasma vthermal > uplasma (Linker et al., of satellite footprints. 1988). Here vthermal is the thermal energy of the back- Parallel electric fields appear not only on the Io flux ground plasma and uplasma is the flow speed. tube, but also in the wake of Io. Because of the large ion Charge exchange differs from other forms of pickup. source localized near Io, the plasma in the immediate It refers to a process in which an incident ion moving wake of this moon is greatly slowed (Frank et al., 1996; with the flow exchanges its charge with a neutral at rest Hinson, 1998). Currents develop along flux tubes that in the moon’s frame. The products of the interaction are link Io’s wake to Jupiter’s ionosphere, couple to the a neutral with the momentum of the impacting ion and ionosphere with currents that require Ek, and act to an ion at rest (because collisional momentum is con- reaccelerate the slowed plasma downstream of Io. Su served). Charge exchange does not change mass density et al. (2003) have worked out many details to account because for each plasma ion lost, a new ion is added. for the ‘‘footprint’’ and the leading trail. Associated However, the ion initially at rest must join the flowing density cavities can account for the radio frequency plasma, requiring that momentum be extracted from the (decametric, hectometric) waves that also link to satellite flow as in the case of pickup. If many charge exchange footprints. interactions occur, the plasma in the final state will have slowed and large numbers of fast moving neutrals will have been created. The neutral atoms speeding off on 6. Beyond MHD: effects on the scale of ion gyroradii straight paths at the local plasma rotation velocity spread to great distances. For example, near Io in re- Another important class of satellite interactions oc- gions where the flow has not yet slowed because of the curs below the MHD scale. Many of the non-MHD interaction, the neutral velocity is 57 km/s (relative to an phenomena arise because the neutral atoms and/or ion moving with Io) tangential to the azimuthal direc- molecules that surround the moons can be ionized tion at the point of origin, implying that in one Jovian through photoionization, electron impact, charge ex- rotation period, the neutral travels 29Rj (Rj ¼ radius of change and related processes (Smyth and Marconi, Jupiter). Wang et al. (2001) discuss and illustrate the 1998). As a result of photoionization and electron im- process. Instruments that can detect rapidly moving pact ionization, ions and electrons are added to the neutral particles can provide images of their source ambient plasma. As soon as they are added to the flow, distribution. An example of such an image produced forces act that accelerate them to the speed of the sur- from data acquired by the Cassini orbiter during its 2002 rounding plasma. The flow then slows because total flyby of Jupiter is shown in Fig. 10. The neutral cloud 2066 M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077

Table 2 Gyroradii of pickup ions

Moon Gyroradius/Rmoon Io 0.0014–0.0016 Europa 0.010–0.012 Ganymede 0.01–0.08 Callisto 0.16–1.73

As remarked above, pickup distributions develop enhanced anisotropy. Anisotropic distributions may be unstable to generation of plasma waves if resonant interactions dominate the damping effect of the background plasma. Resonance with a wave of angular frequency x and wave vector k occurs for particles Fig. 10. Neutral atom image of Jupiter’s surroundings from the Cassini whose velocity satisfies the condition x k V X energetic particle investigation (Krimigis et al., 2002). Jupiter is the þ k k ¼ ci black dot in the center and the grey scale variation relates to intensity. with x < Xci. Thus in a bi-Maxwellian plasma, waves can grow at frequency fwaves with fwaves < fci. In the extends well beyond the 10 s of Rj represented in this following section, examples of such waves in the vicinity image. It extends to hundreds of Rj and has been imaged of Jupiter’s moons will be shown. from Earth (Mendillo et al., 1990) by using the strong Additional non-MHD processes relevant to the in- reflection of the widely dispersed sodium atoms (the teractions near the moons arise because the gyroradii of sodium cloud). energetic ions can become comparable with the scale size One important consequence of ion pickup is to con- of the interaction regions, characterized by the radii of tribute a cross-field current source as illustrated in the moons. Table 2 shows that even for pickup ions, the Fig. 9. Fig. 9 illustrates the displacement of the ion gy- gyroradii at Callisto can become significant relative to rocenter that occurs immediately after ionization of a the radius of the moon. For more energetic particles, neutral. This displacement produces a short-lived cross- finite gyroradius effects can produce spatial asymmetries field current. Because rJ ¼ 0, the cross-B current of particle fluxes near and downstream of satellites. must link to field-aligned currents that extract from Strong selective losses that vary with pitch angle pro- Jupiter’s ionosphere the momentum needed to accelerate duce flux dropouts of limited spatial extent (referred to the new ions. This means that a cloud of pickup ions as microsignatures) in the regions downstream of the acts like a conducting obstacle, producing Alfven wings moons. The effect has been explored principally in re- within which the field is bent (e.g., Neubauer, 1998). lation to the energetic particle signatures downstream of 28 Near Io, a few 10 ions/s are added, and addition at Saturn’s moons (e.g., Carbary et al., 1983). lower rates occurs at the other moons. The current circuit that results closes across the field 7. Interactions at the Galilean moons in the Jovian ionosphere, and in the equatorial regions across the moon’s ionosphere and pickup cloud (or The previous section described significant aspects of possibly through its interior if conducting paths exist). the interaction between a flowing magnetized plasma In the ionosphere, one talks of the Pedersen conductivity and a conducting body surrounded by a neutral cloud. (r , parallel to E ) that arises from collisions with 1 ? In this section, we shall examine the relevance of the neutrals, but conductivity does not require collisions. In concepts introduced previously to the interpretation of the presence of pickup, the relevant conductivity the interactions near specific moons. The best-studied en x v r e ce in ; interactions are those of the Galilean moons of Jupiter. 1 ¼ 2 2 B xce þ vin All of the moons slow and divert the flow and generate can be expressed in terms of an effective collision fre- Alfven wing currents whose properties have been studied. As established in the introductory overview, spe- quency vin (Neubauer, 1998) that includes the contribu- cifics of the interaction depend directly on the properties tions of pickup currents as well as collisions. Finite r1, arises through pickup even in the absence of collisions. of the individual moon such as ionospheric conductivity, Note the region of brightness that leads the Io foot- sputtered neutrals in its surroundings, and the properties print in Fig. 6 and extends over many degrees of of its internal magnetic field. longitude. This region maps magnetically to the near- equatorial regions in which the slowed cloud of plasma 8. Io and its environment introduced near Io is reaccelerated to the rotation speed of the local plasma by the action of a current system that Many think that Alfven wing analysis was introduced links to Jupiter’s ionosphere (Hill and Vasyliunas, 2002). in the study of Io but, in earlier studies, the disturbance M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2067 arising when a conducting body moves relative to a consistent with characteristics of a large pickup ion magnetized plasma was described by Drell et al. (1965) component (Gurnett et al., 1996a). in relation to the motion of satellites through the Earth’s Following the initial wake pass, Galileo acquired data ionosphere. This paper appeared shortly after Bigg in other regions near Io. One pass skimmed flux tubes (1964) reported that Io’s orbital position controls Jupi- linked to Io and identified only minor perturbations of ter’s decametric emissions. Subsequently Piddington and fields and particles (Kivelson et al., 2001a; Gurnett et al., Drake (1968) and Goldreich and Lynden-Bell (1969) 2001). However, in passes above the northern polar re- described the control mechanism in terms of field- gions (Frank and Paterson, 2002) and below the aligned currents linking Io to Jupiter’s ionosphere, and southern polar regions, Galileo crossed flux tubes linked introduced the Alfvenic interaction in this context. to Io and again found order of magnitude increases of The first confirmation of the presence of a tilted wing plasma density and greatly slowed flows. Alfven wing carrying field-aligned currents as described by the perturbations (i.e., bendback of the magnetic field as in Alfven wing model came from Voyager 1 data. The the schematic Fig. 3) appeared clearly in the magnetic spacecraft passed just upstream of the southern Alfven field measured on polar passes (Kivelson et al., 2001b). wing, about 10RIo south of Io, and observed large Fig. 12 compares data from the energetic particle magnetic and flow perturbations, consistent with ex- detector for the wake pass (J0) previously described and pectations (Acuna et al., 1981; Neubauer, 1980). Galileo for two upstream near equatorial passes at different fields and particles detectors found dramatic signatures distances from Io (I24 and I25). The I24 pass shown in of the Io interaction on the first pass by Io in December the middle panel of Fig. 3 recorded only insignificant 1995 (see Fig. 11). The flows accelerated relative to co- changes in the fields and particle instruments and, in rotation as Galileo approached the flanks of the Io particular, there were no changes of the anisotropy of wake, and decreased to near stagnation in the center of energetic electrons. The I27 pass came closer to the the wake where the density increased by almost an order surface of Io and encountered field lines linked to Io. of magnitude (Frank et al., 1996). The magnetic field Onesided loss cones appeared in the energetic electrons decreased by up to 40% in the flanks, but recovered coincident with irregular transverse field fluctuations partially in the wake center (Kivelson et al., 1996a,b). (Mauk et al., 2001). The energetic particle fluxes decreased markedly in the Strong plasma waves were observed at the ion gyro- þ þ wake region and developed strong anisotropy with field frequency of SO2 (fso2 ) at distances up to 20RIo ra- aligned beams in the energetic electrons (see top panel of dially away from Jupiter on the J0 pass. On this same Fig. 12) (Williams et al., 1996). Plasma waves appeared pass, plasma waves consistent with proton pickup were encountered in the wake. On subsequent passes, strong wave power at and near the cyclotron frequencies of þ þ molecular ions SO2 and SO were present (Russell et al., 2001; Russell and Kivelson, 2001) as illustrated in Fig. 13. Waves near the S+ gyrofrequency have also been observed (Blanco-Cano et al., 2001; Wang et al., 2001). The plasma near Io is dominated by Oþ and Sþ ions, so one may ask why only the two molecular ion species produce resonant wave growth in the vicinity of Io. The explanation requires an analysis of the phase space distribution of ions of different mass per unit charge. The background distribution is Maxwellian at low en- ergies, and such distributions damp ion cyclotron waves. Pickup ion distributions are non-Maxwellian. As previously noted, the newly added ions acquire substantial gyro-energy perpendicular to the background magnetic field. In the plasma rest frame this implies a phase space distribution forming a ring about the origin that is shifted by twice the flow velocity in the moon’s rest frame as shown in Fig. 14. If no ions of the same mass/ charge (and therefore the same ion cyclotron frequency) Fig. 11. From Williams et al. (1996) showing plasma flow vectors from are present in the background distribution, and if Frank et al. (1996) placed on the Galileo trajectory and proposed flow contours. Data traces are the magnetic field magnitude in nT and the v? vk, in the plasma rest frame, the anisotropy leads intensity of 15–29 keV electrons vs. time showing marked changes in to wave growth with x ¼Xci. If the ring distribution Io’s wake. is superimposed on a Maxwellian distribution of 2068 M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077

Fig. 12. From Mauk et al. (2001) examples of electron flux measurements in the near vicinity of Io from three different passes (identified in the text). The data are plotted vs. time marked on a projection of the trajectory onto a plane defined by Jupiter’s spin axis direction. Inserts show distributions vs. particle pitch angle measured at different locations along the trajectory. M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2069

Fig. 13. Dynamic spectra of the magnetic ﬁeld measured near Io on the J0 pass of December 1005, inbound top left, outbound top right, on the I24 þ þ þ pass of October 1999, bottom left, and on the I27 pass of February 2000, bottom right. The ion cyclotron frequencies of SO (SO2 ) and H2S are plotted as white traces.

background ions as shown in the lower panel of Fig. 14, wave growth is inhibited and becomes unlikely. It follows that pickup of ions of species present in the background plasma does not normally generate wave power but some molecular ions that are absent in the background plasma may be present in the moons. For Io, þ þ SO2 and SO appear near Io but these molecular ions dissociate sufficiently rapidly that they are virtually absent elsewhere in the torus. Near Io, they cause the emissions shown in Fig. 13. Galileo’s first (J0) pass by Io occurred near the north– south center of the plasma torus. Strong ion cyclotron, þ waves at the SO2 gyrofrequency filled a large volume radially outward of Io and terminated as Galileo passed inside, Io’s orbit. SOþ waves were absent or weak. Waves at both f þ and f þ were seen on I24 and ic; SO ic; SO2 I25, but they were more closely confined to the immediate vicinity of Io. On Voyager 1, whose trajectory Fig. 14. Ring distribution of pickup ions in the absence (above), and in crossed Io’s flux tube 10RIo below the moon, no waves the presence of a background distribution (below). were present at either f þ or f þ . ic; SO ic; SO2 2070 M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077

Fig. 16. Cross-section of Ganymede normal to the plasma flow direction showing the trajectory of Galileo in the G2 polar pass. On the left, vectors with length proportional to magnetic field magnitude and Fig. 15. Schematic of the distribution of molecular ions in the vicinity direction as measured are drawn from positions separated by 1 min of Io. along the trajectory. On the right, the vectors are provided for an assumed internal dipole as discussed in the text. Russell and Kivelson (2000) interpret the distribution of emissions of molecular ions as evidence for a disk-like Ganymede, like Io, is embedded in the flowing plas- distribution of pickup ions around Io illustrated in ma of Jupiter’s magnetosphere. Because the internal Fig. 15, but there may be temporal variations as well. field is sufficiently strong to stand off the flow above the The differences in the wave emissions observed on the surface, a very small magnetosphere forms, a unique different passes may also relate to Io’s north–south po- example to date of a magnetosphere within a magneto- sition within the plasma torus. Only the J0 pass, on sphere. Because the plasma conditions differ so greatly which emissions were observed over an extremely large from those of the solar wind at Earth, not only the size range of distances from Io, occurred near the centrifugal of Ganymede’s magnetosphere differs greatly from equator where the ambient plasma density is higher than Earth’s but also its shape as illustrated in Fig. 17. at other locations on the Io flux tube. Despite the considerable differences in the shapes of the two magnetospheres, the qualitative features of the structure are similar, with three types of field lines evi- 9. Ganymede dent. In both cases there are closed field lines that connect to the primary body at both ends, open field From the perspective of plasma interactions, the lines in the polar regions that connect only once, and critical property of Ganymede is that it has a substantial external solar wind or Jovian field lines that do not internally generated magnetic field. Prior to the Galileo connect at either end. Low latitude reconnection occurs flyby it was thought that Ganymede would have cooled on the upstream surface, the magnetopause that sepa- over geologic time sufficiently that its interior would rates field lines with at least one end on Ganymede from have fully solidified. Because planetary magnetic fields those that do not intersect Ganymede. The reconnection of significant magnitude are thought to require dynamo may be quite different from that familiar at Earth be- action in a conducting fluid core, the presence of an cause the upstream conditions are relatively stable and internal magnetic field was thought to be improbable. the external field remains southward and therefore fa- Thus it is particularly intriguing that this moon has an vorably oriented. In both systems, the newly opened internal magnetic moment with a surface equatorial field lines can direct energetic charged particles from the magnitude of about 700 nT. This surface magnitude is external plasma to the surface near poles. Fig. 18 shows 2% of Earth’s but about twice Mercury’s surface field a map of the surface of Ganymede that suggests to and 7 times the field of Jupiter at Ganymede’s orbit. The Khurana et al. (2004) that the ice properties are modi- magnetic moment is small compared with Earth but the fied by the impacts of the energetic particles entering on ratio (3/2000) scales approximately with the volume. the polar cap (open) field lines, giving strong support for Galileo passed over Ganymede’s north pole and the magnetic field model obtained from the Galileo flyby measured changes of the magnetic field, shown on the measurements. left side of Fig. 16 as lines of varying length and orien- As at Earth, Ganymede’s aurora appears at boundary tation placed along the trajectory. On the right side of between open and closed field lines. Because of the flow the figure is the analogous plot with field values taken asymmetry, the boundary lies at lower latitude on the from a model field, that of a dipole oriented north south downstream side than the upstream side as illustrated in at the center of Ganymede. The dipole orientation is Fig. 19. Observations made by Space Telescope instru- same as at Earth, i.e., the moment is aligned with hBJUPi ments in Dec 2000 show asymmetry between leading at Ganymede. and trailing faces (reported by McGrath, 2002). As at M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2071

Fig. 17. Comparison of sizes and shapes of the magnetospheres of Earth (left) and Ganymede (right). Shown are cross-section in planes through the center of the body containing the ﬂow direction of the external plasma.

Fig. 18. A composite map of the surface of Ganymede analyzed by Khurana et al. (2003). Irregular curves bound the edge of the bright polar caps, whereas the smoothly varying line mark the boundaries between closed and open ﬁeld lines that link to the moon’s surface at various times during each orbital cycle. Black represents missing data.

Earth, the auroral boundaries at 10° lower latitude on to roughly half the electron gyrofrequency appear in the the downstream side (the night side for Earth) than on spectrum (Gurnett et al., 1996b). Fig. 20 shows. that the the upstream side (the day side for Earth). The flow magnetic field increased from O(100 nT) outside pattern shown in Fig. 19 is consistent with data available the magnetosphere to almost 1200 nT at the closest but more analysis is called for and further computer approach altitude of <200 km and was well approxi- simulations would be helpful. mated by an internal dipole field model. The charac- All the fields and particles detectors identified prop- teristics of energetic particles were noted above. erties of the mini-magnetosphere with analogues in Of particular importance for planetary studies was planetary magnetospheres. The magnetopause crossing the evidence of that Ganymede responds inductively to was identified as a sharp change of the flow-aligned x- the changing magnetic field imposed by Jupiter’s mag- component of the field (Kivelson et al., 1996a,b, 1998) netosphere. Data were obtained on multiple flybys with and noise bursts appeared in the plasma wave detector. closest approaches above different regions of Gany- The flow speed decreased abruptly (Williams et al., mede’s surface and at different system III (SIII) longi- 1998). Within the magnetosphere, the plasma charac- tudes. Fig. 21-shows that the radial component of teristics changed markedly (Frank et al., 1997). Plasma Jupiter’s magnetic field at Ganymede’s orbit varies with waves encountered in planetary magnetospheres were SIII longitude, thereby imposing a time-varying signal identified. For example, whistler waves at frequencies up on the moon’s interior. This made it possible to look for 2072 M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077

Fig. 19. Convective flows driven by magnetic reconnection in Gany- mede’s magnetosphere (white arrows) and the boundary between open and closed field lines in the north polar regions. The cross-section contains the background field and the flow direction. changes in the internal field governed by the phase of the temporal variation, which proved useful for establishing that Ganymede responds inductively and probably has a global layer of melted material below the surface ice (Kivelson et al., 2002). For planetary structure this is a critical observation. But the ratio of magnitudes of the induced and the permanent magnetic field is small (1/15) (Schilling et al., 2004), so from the perspective of magnetospheres, the inductive response contributes only Fig. 20. G2 north polar pass 09/06/96. Top to bottom: plasma wave power at different frequencies (Gurnett et al., 1992) vs. UT (magnetic small perturbations. (electric) field above (below)), fce plotted in white; magnetic field data The presence of field-aligned currents linking Gany- and dipole field model of components and magnitude (Kivelson et al., mede to Jupiter’s ionosphere has been confirmed in 1992); energetic particle fluxes (Williams et al., 1992) with locations of various ways. Ultraviolet emissions at the foot of Gan- magnetopause crossing; density, temperature, and flow velocity com- ymede’s flux tube are evident in auroral images (see ponents and magnitude from the plasma detector (Frank et al., 1992) for heavy ions (protons) outside (within) the magnetopause. Fig. 6), again no doubt as the result of field-aligned currents connecting to Jupiter’s ionosphere. In addition, Ganymede controls radio emission in the band 3.2–5.6 which changes at the synodic period of Jupiter because MHz (Menietti et al., 1998) a response analogous to the of the 10° tilt of Jupiter’s dipole moment relative to its control of decametric (slightly higher frequency) emis- spin axis. The dominant component of the time-varying sions by Io, with occurrence rate related to orbital perturbation is radially oriented. The inductive response phase. The occurrence rate of emission controlled by is well approximated as a time-varying dipole field with Ganymede is considerably lower than for Io but the its pole in the moon’s equatorial plane pointing at its power levels can be nearly the same. maxima towards and away from Jupiter. The currents driving the inductive fields flow in the icy crustal layers relatively close to the surface. The magnitude of the 10. Europa and Callisto induced field everywhere outside the surface is less than the driving field; thus the field is too weak to carve a Unlike Ganymede, the icy moons Europa and Cal- magnetospheric cavity out of the surrounding plasma. listo do not harbor stable magnetic moments. Their in- Even though there are no internally generated fields ternally generated magnetic moments vary periodically, large enough to serve as cocoons and shield portions of generated as an inductive response to the time-varying the surfaces of Europa and Callisto from direct inter- external magnetic field of the Jovian magnetosphere action with the ambient plasma, the surroundings of (Kivelson et al., 1999; Zimmer et al., 2000). The field these moons do reveal characteristics that could allow us sensed by these moons varies principally in orientation, to think of their surroundings as induced magneto- M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2073

Fig. 21. Left: SIII radial, latitudinal, and azimuthal components of Jupiter’s magnetic field and its magnitude at Ganymede as a function of SIII longitude. The latitudinal component remains nearly constant but the radial component changes periodically at the synodic period of Jupiter. Right: Galileo’s multiple passes near Ganymede projected into the plane perpendicular to Jupiter’s spin axis (above) and into the plane perpendicular to the flow direction (below). Coverage of both polar and near equatorial regions is critical to fitting internal field models. spheres. The plasma–moon interaction shares many north–south asymmetries of structure (Neubauer, 1999) characteristics with the Ganymede interaction, such as as illustrated in Fig. 22(a). The Galileo magnetometer the development of Alfven wing currents and slowing data shown in Figs. 22(b) and 23(b) have been rotated and diversion of flow in the vicinity. into a coordinate system with x-aligned with the back- Europa’s and Callisto’s induced internal fields pro- ground flow and the magnetic field in the xz plane. The duce interesting and unique interaction signatures with bendback that arises from Alfven wing currents appears periodic variability. The presence of an induced internal as an abrupt change of Bx. The schematic illustration magnetic field modifies an Alfven wing, leading to 23(a) indicates Galileo’s path for the pass plotted in

Fig. 22. (a) Schematic of the Alfven wing for an induced magnetic field at Callisto after (Neubauer, 1999). The heavy arrow represents the Galileo trajectory on C10 through both Alfven wings. (b) Magnetic field in the vicinity of Callisto on Galileo’s C10 pass that crossed from the northern to the southern Alfven wing as illustrated in a. Gray shading delimits the interval during which Galileo crossed the region in which Callisto blocks the unperturbed flow. Between 00:01 and 00:11 UT, and between 0 0:21 and 00:31 UT the field rotated as Galileo crossed the northern (Nin and Nout) and southern (Sin and Sout) Alfven wings. 2074 M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077

Fig. 23. (a) As for Fig. 22(a) but showing the path of Galileo through one Alfven wing to represent the C3 pass. (b) As for Fig. 22(b), but for the C3 pass which crossed only one Alfven wing. Multiple rotations within the Alfven wing are interpreted as evidence of structured conducting regions or multiple narrow current structures linking to patchy regions of ion pickup.

Fig. 23(b). Galileo’s crossings of the Alfven wings at Callisto confirmed the predicted displacements as shown in Figs. 22 and 23 in which the regions with abrupt rotations are tinted and clearly displaced from the gray shaded projected cross-section of Callisto. The Bx perturbations are quite abrupt, localized in a region shifted towards Jupiter with a negative Bx north of Callisto and shifted outward with a positive Bx to the south, consistent with the model. The pass plotted in Fig. 23(b) crossed only the northern Alfven wing, and this, too, is shifted outward with negative Bx perturbations. For a homogeneous conductor, one would expect the current to flow towards the moon on one side of the Alfven wing, and out on the other, thereby producing Fig. 24. Representation of structured field-aligned currents linking to two sharp rotations. The Alfven wings at Callisto are patchy conducting regions consistent with the evidence of transverse not of this form; but instead include multiple rotations. perturbations in Fig. 23(b). Such a form can be understood if the conducting regions are discontinuous as would be the case for a structured 11. Titan conductor or for patchy pickup ion distribution. Fig. 24 shows how field aligned currents might flow in and out The focus on the Galilean moons of Jupiter arises of a structured conducting body. In this schematic the naturally because they have been studied intensely. field would bend back between currents I01 and I12, Other large moons exist in the outer solar system, but it would straighten out partially between I12 and I23, and is only for Titan that observations relevant to the plas- would bend again between I23 and I34 before straight- ma interaction are available at this time. Table 3 gives ening out again. It seems most likely that the structure at some of the dimensionless parameters relevant to the Callisto arises because of localized clouds of pickup plasma interaction at the location and time of the ions, with each individual cloud producing a localized Voyager 1 encounter in 1980. As for the Galilean Alfven wing signature. Control of part of the low fre- moons, the flow was sub-magnetosonic and no shock quency decametric emissions at the orbital period of was present. The large Alfvenic Mach number implied a Callisto gives evidence that the field-aligned current as- large bendback angle of the Alfven wing and the inter- sociated with the Alfven wing closes in Jupiter’s iono- action was further complicated by the interactions of the sphere (Menietti et al., 2001). flowing plasma with the atmosphere and extended cloud M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2075

Table 3 the magnetospheric plasma with a dense, multi-ion Selected dimensionless parameters (Voyager 1 encounter) ionosphere. Cassini’s passes will tell us if this moon has MA ¼ u=vA ¼ 1:9 an intrinsic magnetic field and will provide evidence of hA° ¼ 62 loss processes that may be relevant to understanding the 2 2 1=2 Ms ¼ u=cs ¼ 0:57Mf ¼ u=ðvA þ cs Þ ¼ 0:55 2 long-term evolution of its atmosphere. b ¼ p=ðB =2l0Þ¼11 _ 2 M=qivrT < 600 surf 6 Bint =Bbg 2:3; ð0:8Þ The entries are based on data obtained by Voyager 1 during a 12. Summary single encounter in 1980. u is flow speed, cs is the sound speed. Here M , M , and M , are the Alfvenic, sonic, and magnetosonic Mach A S f In simple pictures of the solar system, moons move numbers, hA is the bendback angle of the Alfven wing, p is the plasma thermal pressure, M_ represents the mass addition rate at Titan and is passively in their orbits, governed by the gravitational given relative to the rate at which mass flows into the cross-section of force that binds them to the planet that they circle. Here Titan from the ambient plasma flow and rT is Titan’s radius. Nu- we have examined a different class of processes linking a _ merical values are taken from Neubauer et al. (1984) with M from planet with the satellite bodies that it controls. Even for Strobel and Shemansky (1982). The surface field estimate is that of Israelevich in Kabin et al. (2000), which should be contrasted with the the Galilean moons, the interactions have not yet been smaller estimate (shown in brackets) given by Neubauer et al. (1984). fully characterized and we are challenged to extend our interpretation of the host of electrodynamic effects im- of pickup ions near Titan (Ness et al., 1982). Voyager 1 portant in planetary systems. Such studies are relevant observations showed field draping around Titan in a to our understanding of the evolution of small bodies configuration much like that seen at Venus (Kivelson and their atmospheres. Further insight can be antici- and Russell, 1983). An equatorial current sheet formed pated in the near future as Galileo data from multiple in the downstream wake of Titan, a feature that was not passes by the moons of Jupiter is more fully exploited observed for any of the passes by the Galilean moons. and as measurements from Titan’s environment become At Titan, the external field and plasma conditions are available. likely to vary more than at any Jovian moon. The magnetospheric field and the plasma environment change with time at the Galilean moons, but the changes Acknowledgements are not dramatic. Titan, whose orbit, lies near 17RS (RS is the radius of Saturn), may at times enter the dayside The author acknowledges the valuable support of the magnetosheath or even the solar wind as shown in Galileo Project and particularly of the UCLA magne- Fig. 25; major changes of field and plasma environment tometer team: Krishan K. Khurana, Christopher T. must occur quite often. Cassini, in orbit about Saturn, Russell, Raymond J. Walker, Steven Joy, Todd King, will repeatedly encounter Titan and will acquire mea- Joe Mafi, Martin Volwerk and Christopher Zimmer. surements within the interaction region for qualitatively This paper was prepared with partial support from different parameter regimes of the surrounding plasma. NASA under Grant JPL 1238965 and from the Division In anticipation of Cassini’s exploration of Titan, simu- of Atmospheric Sciences of the National Science lations are being developed to address the interaction of Foundation under Grant NSF ATM 0205958.

References

Acuna, M.H., Neubauer, F.M., Ness, N.F. Standing Alfven wave current system at Io: voyager 1 observations. J. Geophys. Res. 86, 8513–8521, 1981. Bigg, E.K. Inﬂuence of the satellite Io on Jupiter’s decametric emission. Nature 203, 1008–1010, 1964. Blanco-Cano, X., Russell, C.T., Strangeway, R.J. The Io mass loading disk: wave dispersion analysis. J. Geophys. Res. 106, 26261–26275, 2001. Carbary, J.F., Krimigis, S.M., Ip, W.-H. Energetic particle microsignatures of Saturn’s satellites. J. Geophys. Res., 888,947–888,958, Fig. 25. The Saturn magnetospheric system showing the bow shock 1983. (narrow line), the magnetopause (heavy solid line), the orbit of Titan, Clarke, J.T., Ajello, J., Ballester, G., et al. Ultraviolet emissions from and the planet ring system (not to scale). To the left, in a low dynamic the magnetic footprints of Io, Ganymede and Europa on Jupiter. pressure solar wind, Titan’s orbit remains within the magnetopause. In Nature 415, 997–1000, 2002. the center, for a larger solar wind dynamic pressure, Titan’s orbit Delamere, P., Bagenal, F., Ergun, R., Su, Y. Momentum transfer moves into the magnetosheath, and to the right, for an exceptionally between the Io plasma wake and Jupiter’s magnetosphere. J. high solar wind dynamic pressure, Titan’s orbit enters the solar wind. Geophys. Res. 108(A6) Jun 18, Art. No. 1241, 2003. 2076 M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077

Drell, S.D., Foley, H.M., Ruderman, M.A. Drag and propulsion of Kivelson, M.G., Warnecke, J., Bennett, L., Joy, S., Khurana, K.K., large satellites in the ionosphere: an Alfven propulsion engine in Linker, J.A., Russell, C.T., Walker, R.J., Polanskey, C. Gany- space. J. Geophys. Res. 70, 3131–3145, 1965. mede’s magnetosphere: magnetometer overview. J. Geophys. Res. Frank, L.A., Ackerson, K.L., Lee, J.A., English, M.R., Pickett, G.L. 103, 19,963–19,972, 1998. The plasma instrumentation for the Galileo mission. Space Sci. Kivelson, M.G., Khurana, K.K., Stevenson, D.J., Bennett, L., Rev. 60, 283–304, 1992. Joy, S., Russell, C.T., Walker, R.J., Zimmer, C., Polanskey, C. Frank, L.A., Paterson, W.R., Ackerson, K.L., Vasyliunas, V.M., Europa and Callisto: induced or intrinsic fields in a period- Coroniti, F.V., Bolton, S.J. Plasma observations at Io with the ically varying plasma environment. J. Geophys. Res. 104, Galileo spacecraft. Science 274, 394–395, 1996. 4609–4625, 1999. Frank, L.A., Paterson, W.R., Ackerson, K.L., Bolton, S.J. Low- Kivelson, M.G., Khurana, K.K., Russell, C.T., Joy, S.P., Volwerk, energy electron measurements at Ganymede with the Galileo M., Walker, R.J., Zimmer, C., Linker, J.A. Magnetized or spacecraft: probes of the magnetic topology. J. Geophys. Res. 24, unmagnetized: ambiguity persists following Galileo’s encounters 2159, 1997. with Io in 1999 and 2000. J. Geophys. Res. 106, 26,121–26,135, Frank, L.A., Paterson, W.R. Plasma observed with the Galileo 2001a. spacecraft during its flyby over Io’s northern polar region. J. Kivelson, M.G., Khurana, K.K., Russell, C.T., Walker, R.J. Magnetic Geophys. Res. 107 (A8), 2002, doi: <10.1029/2002JA009240>. signature of a polar pass over Io Eos Trans. AGU, 82 (47), Fall Goertz, C.K. Io’s interaction with the plasma torus. J. Geophys. Res. Meet. Suppl., Abstract P11A-01, 2001b. 85, 2949–2956, 1980. Kivelson, M.G., Khurana, K.K., Volwerk, M. The permanent and Goldreich, P., Lynden-Bell, D. Io a Jovian unipolar inductor. inductive magnetic moments of Ganymede. Icarus 157 (2), 507– Astrophys. J. 156, 59–78, 1969. 522, 2002. Gurnett, D.A., Kurth, W.S., Shaw, R.R., Roux, A., Gendrin, R., Krimigis, S.M. et al. A nebula of gases from Io surrounding Jupiter. Kennel, C.F., Scarf, L.F., Shawhan, S.D. The Galileo plasma wave Nature 415, 994–996, 2002. system. Space Sci. Rev. 60, 341–355, 1992. Linker, J.A., Kivelson, M.G., Walker, R.J. An MHD simulation of Gurnett, D.A., Kurth, W.S., Roux, A., Bolton, S.J., Kennel, C.F. plasma flow past Io: Alfven and slow mode perturbations. Galileo plasma wave observations in the Io plasma torus and near Geophys. Res. Lett. 15, 1311–1314, 1988. Io. Science 274, 391–392, 1996a. Mauk, B.H., Williams, D.J., Eviatar, A. Understanding os space Gurnett, D.A., Kurth, W.S., Roux, A., Bolton, S.J., Kennel, C.F. environment interaction: recent energetic electron measure- Evidence for a magnetosphere at Ganymede from plasma-wave ments from Galileo. J. Geophys. Res. 106 (26), 195–208, observations by the Galileo spacecraft. Nature 384, 535–537, 2001. 1996b. McGrath, M.A. Hubble Space Telescope observations of Galilean Gurnett, D.A., Persoon, A.M., Kurth, W.S., Roux, A., Bolton, S.J. satellites, Highlights of Astronomy IAU Symposia, 12, 618, Electron densities near Io from Galileo plasma wave observations. 2002. J. Geophys. Res. 106, 26,225–26,232, 2001. Mendillo, M., Baumgardner, J., Flynn, F., Hughes, W.J. The extended Hill, T.W., Dessler, A.J., Goertz, C.K. Magnetospheric models, sodium nebula of Jupiter. Nature 348, 312–314, 1990. in: Dessler, A.J. (Ed.), Physics of the Jovian Magneto- Menietti, J.D., Gurnett, D.A., Kurth, W.S., Groene, J.B. Control of sphere. Cambridge University Press, New York, pp. 353– Jovian radio emission by Ganymede. Geophy. Res. Lett. 25 (23), 394, 1983. 4281–4284, 1998. Hill, T.W., Vasyliunas, V.M. Jovian auroral signature of Io’s Menietti, J.D., Gurnett, D.A., Christopher, I. Control of Jovian radio corotational wake. J. Geophys. Res. 107 (A12), 2002, SMP 31, emission by Callisto. Geophys. Res. Lett. 28, 3047–3050, 2001. doi:10.1029/2002JA009514. Ness, N.F., Acuna, M.H., Behannon, K.W., Neubauer, F.M. The Hinson, D.P., Kliore, A.J., Flasr, F.M., et al. Galileo radio occultation induced magnetosphere of Titan. J. Geophys. Res. 87, 1369–1381, measurements of Io’s ionosphere and plasma wake. J. Geophys. 1982. Res. 103, 29343–29357, 1998. Neubauer, F.M. Non-linear standing Alfven wave current system at Kabin, K., Israelevich, P.L., Ershkovich, A.I., Neubauer, F.M., Io: theory. J. Geophys. Res. 85, 1171–1178, 1980. Gombosi, T.I., Zeeuw, D.L.D., Powell, K.G. Titan’s magnetic Neubauer, F.M., Gurnett, D.A., Scudder, J.D., Hartle, R.E. wake: atmospheric or magnetospheric interaction. J. Geophys. Res. Titan’s magnetospheric interaction, in: Gehrels, T., Matthews, 105, 10761–10770, 2000. M.S. (Eds.), Saturn. Univ. Ariz. Press, Tucson, pp. 760–787, Khurana, K.K. et al. The origin of Ganymede’s polar cap, 2003 (in 1984. preparation). Neubauer, F.M. The sub-Alfvenic interaction of the Galilean satellites Kivelson, M.G. Pulsations and magnetohydrodynamic waves, in: with the Jovian magnetosphere. J. Geophys. Res. 103, 19,843– Kivelson, M.G., Russell, C.T. (Eds.), Introduction to Space 19,866, 1998. Physics. Cambridge University Press, New York, pp. 330–355, Neubauer, F.M. Alfven wings and electromagnetic induction in the 1995. interiors: Europa and Callisto. J. Geophys. Res. 104, 28,671– Kivelson, M.G., Russell, C.T. The interaction of flowing plasmas with 28,684, 1999. planetary ionospheres: a Titan-Venus comparison. J. Geophys. Piddington, J.H., Drake, J.F. Electrodynamic effects of Jupiter’s Res. 88, 49–57, 1983. satellite Io. Nature 217, 935–937, 1968. Kivelson, M.G., Khurana, K.K., Means, J.D., Russell, C.T., Snare, Russell, C.T., Kivelson, M.G. Detection of SO in Io’s exosphere. R.C. The Galileo magnetic field investigation. Space Sci. Rev. 60, Science 287, 1998–1999, 2000. 357–383, 1992. Russell, C.T., Kivelson, M.G. Evidence for sulfur dioxide, sulfur Kivelson, M.G., Khurana, K.K., Walker, R.J., Warnecke, J., Russell, monoxide, and hydrogen sulfide in the Io exosphere. J. Geophys. C.T., Linker, J.A., Southwood, D.J., Polanskey, C. Io’s interaction Res. 106, 33,267–33272, 2001. with the plasma torus: Galileo magnetometer report. Science 274, Russell, C.T., Wang, Y.L., Blanco-Cano, X., Strangeway, R.J. The Io 396–398, 1996a. mass loading disk: constraints provided by ion cyclotron wave Kivelson, M.G., Khurana, K.K., Russell, C.T., Walker, R.J., War- observations. J. Geophys. Res. 106, 26233–26242, 2001. necke, J., Coroniti, F.V., Polanskey, C., Southwood, D.J., Schu- Schilling, N., Khurana, K.K., Kivelson, M.G. Limits on an intrinsic bert, G. Discovery of Ganymede’s magnetic field by the Galileo dipole moment in Europa. J. Geophys. Res. (Planets), 2004 spacecraft. Nature 384, 537–541, 1996b. (submitted). M.G. Kivelson / Advances in Space Research 33 (2004) 2061–2077 2077

þ Smyth, W.H., Marconi, M.L. An initial look at the Iogenic SO2 source Williams, D.J., Mauk, B.H., McEntire, R.E., Roelof, E.C., during the Galileo ﬂyby of Io. J. Geophys. Res. 103, 9083–9089, 1998. Armstrong, T.P., Wilken, B., Roederer, J.G., Krimigis, S.M., Strobel, D.F., Shemansky, D.E. EUV emission from Titan’s upper Fritz, T.A., Lanzerotti, L.J. Electron beams and ion compo- atmosphere: Voyager 1 encounter. J. Geophys. Res. 87, 1361–1368, sition measured at Io and in its torus. Science 274, 401, 1982. 1996. Su, Y., Ergun, R., Bagenal, F., Delamere, P. Io-related auroral arcs: Williams, D.J., Mauk, B.H., McEntire, W. Properties of modelling parallel electric ﬁelds. J. Geophys. Res. 108 (A2), Art. Ganymede’s magnetosphere as revealed by energetic par- No. 1094, Feb 28, 2003. ticle observations. J. Geophys. Res. 103, 17523–17534, Wang, Y., Russell, C.T., Raeder, J. The Io mass loading disk: model 1998. calculations. J. Geophys. Res. 106, 26,243–26260, 2001. Zimmer, C., Khurana, K.K., Kivelson, M.G. Subsurface oceans on Williams, D.J., McEntire, R.W., Jaskulek, S., Wilken, B. The Galileo Europa and Callisto: constraints from Galileo magnetometer energetic particles detector. Space Sci. Rev. 60, 385, 1992. observations. Icarus 147, 329, 2000.