Outer magnetospheric structure: Jupiter and Saturn compared

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116,, doi: /2010ja016045, 2011 Outer magnetospheric structure: Jupiter and Saturn compared D. R. Went, 1 M. G. Kivelson, 2,3 N. Achilleos, 4,5 C. S. Arridge, 5,6 and M. K. Dougherty 1 Received 27 August 2010; revised 11 January 2011; accepted 10 February 2011; published 20 April [1] The Jovian dayside magnetosphere is traditionally divided into three different regions with the outermost region, colloquially referred to as the cushion region, existing between the outer edge of the magnetodisk and the magnetopause. Magnetometer and plasma data from 6 different spacecraft are used to determine the average properties of this region, including its characteristic thickness at the subsolar point, and these observations are compared with data from the Saturnian magnetosphere obtained using the Pioneer, Voyager, and Cassini spacecraft. Significant differences are found in the structure of the two rotationally driven magnetospheres with the Saturnian system showing little evidence for the cushion region seen at Jupiter. These differences are discussed in terms of the parameter regimes pertinent to each planet, and the potential effect of magnetodisk warping at Saturn is discussed. It is tentatively suggested that while the Jovian magnetodisk typically breaks down several tens of planetary radii inside the magnetopause, thus allowing plasmadepleted flux tubes beyond it to relax into the cushion region configuration, the Saturnian magnetodisk may persist until much closer to the magnetospheric boundary. A number of observational tests of this hypothesis are proposed, and the need for improved observations at both planets is stressed. Citation: Went, D. R., M. G. Kivelson, N. Achilleos, C. S. Arridge, and M. K. Dougherty (2011), Outer magnetospheric structure: Jupiter and Saturn compared, J. Geophys. Res., 116,, doi: /2010ja Blackett Laboratory, Imperial College London, London, UK. 2 Department of Earth and Space Sciences, University of California, Los Angeles, California, USA. 3 Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California, USA. 4 Department of Physics and Astronomy, University College London, London, UK. 5 Centre for Planetary Sciences at UCL/Birkbeck, London, UK. 6 Mullard Space Science Laboratory, Department of Space and Climate Physics, University College London, Dorking, UK. Copyright 2011 by the American Geophysical Union /11/2010JA Introduction [2] The structure and dynamics of the Saturnian magnetosphere are often described as being intermediate between those of Jupiter and the Earth [Gombosi et al., 2009]. Plasma in the terrestrial magnetosphere is forced into large scale motion by the solar wind driven Dungey cycle [Dungey, 1961] beginning with dayside reconnection between magnetospheric and interplanetary magnetic fields. The open flux tubes thus produced are then convected over the polar regions of the planet by the fast flowing solar wind, during which time they lose much of their mass content, before meeting again in the magnetotail and reconnecting to form closed flux tubes. These closed flux tubes can then return to the dayside and complete the circulation. [3] In contrast, despite some evidence for Dungey cycle operation [Cowley et al., 2003], the overall dynamics of the Jovian magnetosphere appear to be dominated by internal sources of angular momentum. Plasma released by the volcanically active moon Io is continually picked up by the Jovian magnetic field and, through field aligned currents linked to the ionosphere, rapidly accelerated to near corotational velocities [Bagenal and Sullivan, 1981]. Vasyliũnas [1983] then described the resulting circulation as plasma is driven radially outward by the combined action of the centrifugally driven interchange (b < 1) and ballooning (b < 1) instabilities [Kivelson and Southwood, 2005] where b (beta) is the ratio of plasma to magnetic pressure. Despite distorting the Jovian magnetic field [Smith et al., 1974] this outward transportation and ballooning is thought to be restricted on the dayside by the dynamic pressure of the solar wind acting on the magnetopause. Only on the nightside (where confinement by the magnetopause becomes negligible) can plasma loaded flux tubes expand without restriction. Eventually a critical point is reached beyond which magnetic curvature can no longer support the outward acting forces associated with expansion [Goertz, 1983] and a plasmoid is released downtail in a burst of reconnection [Woch et al., 2002]. The resulting plasmadepleted flux tubes, still anchored to the planet, are then able to dipolarize [Kivelson, 2005] and return to the inner magnetosphere [Kivelson and Southwood, 2005] where they become reloaded with iogenic plasma so that the so called Vasyliũnas cycle can repeat. [4] Saturn, like Jupiter, is also a fast rotator with significant sources of plasma in its magnetosphere. Moreover, the inner moon Enceladus is known to have a highly dynamic atmosphere with active geysers that play a similar role at Saturn to the volcanism seen on Io [Dougherty et al., 2006; Porco et al., 2006; Pontius and Hill, 2006]. Consequently, 1of14

2 contribution from those involved in the Dungey cycle [Cowley et al., 2003] being significant only during periods of enhanced solar wind activity [Badman and Cowley, 2007]. The resulting circulation pattern is described in detail by Kivelson and Southwood [2005] and well illustrated by Figure 7 of that paper. [6] Saturnian equivalents of the inner magnetosphere and magnetodisk have recently been reported by Arridge et al. [2008a] but are observed only when the magnetosphere is in an expanded configuration. In this paper we complete the comparison with Jupiter by considering evidence for an outer magnetospheric cushion region at Saturn, paying particular attention to those expanded passes where the magnetodisk, which forms a vital component of the cushion region s definition at Jupiter, is expected to be present. We begin with a review of the extensive observations made at Jupiter. Figure 1. Schematic representation of the Jovian magnetosphere with (top) the noon meridian viewed from dusk and (bottom) the equatorial plane viewed from above. In both cases the Sun is to the left. The inner (blue), middle (yellow), and outer (green) magnetospheres are not shown to scale. Figure adapted from Smith et al. [1976]. there is now growing evidence for a Saturnian circulation pattern consistent with the Vasyliũnas cycle [André et al., 2007; McAndrews et al., 2009] although, in contrast to the Jovian dynamics discussed above, the solar wind appears to retain considerable influence on the outer magnetosphere [McAndrews et al., 2008], magnetotail [Bunce et al., 2005] and aurora [Bunce et al., 2008]. The complex interaction between solar wind and internally driven transport processes at Saturn is thus a topic of much research. [5] Returning our attention to Jupiter, the dayside magnetosphere is traditionally divided into three qualitatively distinct spatial regions (Figure 1) as discussed by Smith et al. [1976]. The inner magnetosphere (R ] R J ) is dominated by Jupiter s strong internal dipole with a smoothly varying, southward directed field close to the equator. Further out, the increasing significance of centrifugal forces leads to a radially stretched magnetodisk (20 ] R ] 60 R J ) where the ballooning instability dominates the magnetic field geometry. Finally, between the magnetodisk and magnetopause, the more dipolar yet disordered outer magnetosphere or cushion region (described by Kivelson [1976] as a layer of magnetic turbulence ) is suggestive of reduced centrifugal stresses and plasma depleted flux tubes. Kivelson and Southwood [2005] interpret these flux tubes as those involved in the final (mass release) stages of the Vasyliũnas cycle with a secondary 2. The Cushion Region at Jupiter 2.1. Typical Characteristics: Ulysses [7] A typical inbound pass through the Jovian magnetosphere was made by the Ulysses spacecraft [Wenzel et al., 1992] along a low latitude (<15 ) prenoon meridian [Haynes, 1995]. To illustrate the qualitative structure of the observed magnetic field we consider changes in two parameters derived from the magnetic field data. The first of these is the ratio of the poloidal field components to the total field magnitude ( B / B, B R / B ) and the second is the angle, INT, between the observed magnetic field, B OBS, and the magnetic field associated with the Jovian internal dipole, B INT [Randall, 1998]. These observations are presented in Figures 2b and 2c, respectively, between distances of R J with the magnetopause located at 88 R J. The data is plotted with increasing radial distance (decreasing UT time for an inbound pass) on the x axis. [8] Close to the magnetic equator, field lines with a dipolar configuration are southward directed with B > B R while radially stretched, nondipolar field lines are associated with B R > B. Away from the equator the changing direction of dipolar magnetic field lines invalidates this interpretation and the angular parameter INT must be used to characterize the field instead. A critical angle ( INT = 50 ) is defined above which the magnetic field will be considered significantly nondipolar and if, in addition to this, B R > B, we describe the resulting field configuration as a radially stretched magnetodisk. [9] With these points in mind the Ulysses data of Figure 2 can be seen to show three distinctly different magnetospheric regions. A quasi dipolar inner magnetosphere, shaded blue, exists close to the planet where INT < 50. Between distances of R J Ulysses explores a radially stretched region, shaded yellow, where B R > B and INT reaches values close to 90. This is the Jovian middle magnetosphere or magnetodisk which, on closer inspection, is found to consist of a relatively thin current sheet (R C <2R J [Mauk and Krimigis, 1987], where R C is the radius of curvature of the field lines on the magnetic equator) surrounded by two radially stretched regions of opposite polarity. Crossings of this current sheet, observed as reversals in B R, occur whenever the magnetic equator passes over the spacecraft. In the spacecraft rest frame the magnetic equator flaps up and down, with approximately the planetary rotation period, 2of14

3 Figure 2. Ulysses observations in the Jovian magnetosphere. (a) Jovian System III magnetic field components (B R,red;B, blue or white; B,green)and± B (black). (b) Normalized poloidal field components ( B / B, blue or white; B R / B, red). (c) Angle INT between the observed magnetic field, B OBS,and the internal magnetic field, B INT. Horizontal dashed lines denote the critical magnetodisk angles of 50 and = 130. (d) Thirty minute normalized magnetic field RMS fluctuation. (e) SWOOPS thermal electron density (blue or white) and temperature (red). Vertical dashed lines denote local minima in absolute magnetic latitude, l M,which beyond 50 R J corresponds to l M = 0. The inner magnetosphere (blue), magnetodisk (yellow), transition region (white), cushion region (green), boundary layers (cyan), magnetopause crossings (red), and magnetosheath (grey) are shaded. The radial distance, planetocentric latitude, and local time of the spacecraft are shown along the x axis. due to the rocking motion of the rotating planetary dipole which is inclined by 10 to the rotation axis. Beyond the magnetodisk the field direction rotates slowly and, for R > 83 R J, Ulysses explores a third magnetospheric region which is once again dipolar. This is the outer magnetosphere or cushion region, shaded green, separated from the magnetodisk by a roughly 14 R J transition region of intermediate properties, shaded white. In the transition region the disturbed magnetic field rotates through 90 and the two poloidal field components are comparable in magnitude. [10] The cushion region electron density (measured by the SWOOPS instrument [Bame et al., 1992]) is 10 2 cm 3 while, in contrast, the first crossing of the magnetodisk plasma sheet in the middle magnetosphere is associated with an electron density almost an order of magnitude higher. At the radial distance of this crossing ( 68 R J ) the electron density observed in the magnetodisk lobes is comparable to that seen previously in the overlaying cushion region. Both the plasma sheet and magnetodisk lobe density decrease upon approaching the planet, a counterintuitive observation interpreted by Phillips et al. [1993] as an effect associated with Ulysses unusual trajectory. In order to launch Ulysses into a polar orbit around the sun, the planetocentric latitude of the spacecraft increased significantly on approaching Jupiter. This is clearly evident from the top right panel of Figure 3 where the meridional projection of Ulysses s Jupiter flyby trajectory is shown in red. The high planetocentric latitude also explains the B R dominated field seen in the quasi dipolar ( < 50 ) inner magnetosphere. As can be seen from Figure 2e, where density is plotted in blue/white and temperature in red, a transient density (temperature) increase (decrease) is observed at roughly 24 R J. This feature is interpreted by Phillips et al. [1993] as a spacecraft traversal of high latitude open field lines. [11] The highly disturbed nature of the transition region field is clear when the RMS fluctuation (calculated from the 3of14

4 Figure 3. In situ exploration of (top) Jovian and (bottom) Saturnian magnetospheres with spacecraft trajectories projected onto the (left) equatorial and (right) instantaneous meridional planes. The inner and outer dashed lines represent the nominal locations of the magnetopause and bow shock, respectively. Pioneer 10, blue; Pioneer 11, gold; Voyager 1, pink; Voyager 2, black; Ulysses, red; Galileo, cyan; Cassini, grey; Cassini Rev 20 inbound, dashed purple. Pythagorean sum of the standard deviations of each field component computed over a 30 min time interval; Figure 2d) is examined for this region at 1 min resolution. The thermal electron density appears to correlate with some of these fluctuations (dn e /n e 2 10) and was studied in detail by Southwood et al. [1993]. Some of the density enhancements were found to be associated with magnetic field rotations reminiscent of the current sheet crossings seen in the magnetodisk, however their occurrence in the transition region appears to be unrelated to the spacecraft s magnetic latitude and they are often accompanied by simultaneous reversals in both B R and B. These are interpreted as spacecraft encounters with a highly warped magnetodisk, strongly tilted in the meridional plane. Other density enhancements are not associated with a clear rotation of the magnetic field but are, instead, associated with a large decrease in the magnitude of the B dominated background magnetic field. These phenomena are termed magnetic nulls and are discussed in detail by Haynes et al. [1994] and Southwood et al. [1995]. [12] A consideration of the radial force balance condition for an isotropic (sub)corotating plasma [Southwood and Kivelson, 2001] allows us to relate the observed magnetodisk and cushion region field configurations to the ambient population of magnetospheric plasma: ^n B2 = 0 ¼ r P þ B2 þ N i ðm i þ m e ÞW 2 r: ð1þ R C 2 0 [13] Equation (1) describes the first order balance between the magnetic curvature force (left), pressure gradient force (right) and centrifugal force (far right). Here R C is the local radius of curvature of the field, B 2 /2m 0 is the magneticpressure, P is the plasma pressure (assumed to be isotropic), N i is the number density of ions, m e and m i are the electron and mean ion masses, respectively, W is the angular frequency of plasma rotation and r is the perpendicular distance from the spin axis of the planet about which the plasma rotates. The unit vector ^n points in the direction of the outward normal to the field line. According to this equation, higher density plasmas will tend to stretch out the magnetic field (decreasing the radius of curvature in order to increase the stabilizing tension force) whereas lower density plasmas, 4of14

5 at a given r and w, can be successfully constrained by a less stretched configuration. [14] The effect described above is negligible in the low b inner magnetosphere (dominated by Jupiter s strong internal dipole) however it becomes increasingly important at larger radial distances where the plasma and magnetic pressures become comparable. Since the cushion region itself is found at large (>50 R J ) radial distances, the quasi dipolar nature of the cushion region field suggests, in light of equation (1), that the Jovian outer magnetosphere is depleted of magnetospheric plasma relative to the radially stretched magnetodisk. In this picture the cushion region consists of flux tubes which have recently lost much of their mass content as a result of Vasyliũnas and Dungey cycle reconnection while the transition region corresponds to the distorted outer edge of the magnetodisk. Dense clumps of plasma occasionally break off the outer edge of the magnetodisk and move through the overlaying cushion region where they are observed in magnetic field data as sharp decreases in the total magnitude of the magnetic field. These are the magnetic nulls of Haynes et al. [1994] Spatial and Temporal Variability [15] A total of six magnetometer carrying spacecraft have explored the Jovian magnetosphere to date (Pioneer 10, Pioneer 11, Voyager 1, Voyager 2, Ulysses and Galileo; see Figure 3) covering all local times in the equatorial plane out to distances of 100 R J. New Horizons was not equipped with a magnetometer and, while Cassini skimmed the dusk magnetosphere en route to Saturn, it did not penetrate far enough for the full magnetospheric structure to be determined. Considered together, the available observations reveal significant temporal and spatial variability in the properties of the cushion region described above. [16] Considering the temporal variability first, the expansion and contraction of the Jovian magnetosphere (usually an equilibrating response to changes in solar wind dynamic pressure) is thought to result in magnetopause and, by extension, cushion region motion, relative to the planet, at velocities comparable to or greater than those of an exploring spacecraft [Sonnerup et al., 1981; Cowley and Bunce, 2003]. In the rest frame of the planet this motion results in the cushion region sweeping back and forth over the exploring spacecraft at the same time as the spacecraft itself moves relative to Jupiter. The inbound leg of the Pioneer 10 flyby illustrates this point well with the spacecraft crossing the cushion region to magnetodisk boundary 3 times in the space of just 3 days, covering a radial distance of over 40 R J while doing so. Such dynamical considerations act to modulate the time a spacecraft spends inside the cushion region and makes the true inertial thickness of the region impossible to determine from single spacecraft data. The same unpredictable boundary motion also has consequences for the stability and thickness of the underlying plasma sheet and is likely to control the probability of plasma blobs breaking off from the magnetodisk [Southwood and Kivelson, 2001]. This may, in turn, introduce a temporal variability to the nature and extent of cushion and transition region field fluctuations. Other potential sources of variability, such as the bursty nature of magnetotail reconnection [Woch et al., 2002] and the variable activity levels seen at Io [Bagenal et al., 2004] are probably of secondary importance. [17] From a spatial perspective, Kivelson and Southwood [2005] identified a local time asymmetry in the cushion region with the region of quasi dipolar field being more evident in the morningside magnetosphere as opposed to afternoon. In the predusk sector Kivelson and Southwood [2005] found the B R and B components to be more comparable than in the cushion region and, while the B R component reversed sign rather irregularly, clear spectral peaks were found at the rotation period of Jupiter. These observations were interpreted as a result of the plasma sheet thickening as it rotates toward dusk, reducing the contrast between centrifugally stressed and plasma depleted flux tubes, combined with a gradual refilling of the cushion region by plasma that has broken off the outer edge of the dynamically unstable plasma sheet. Such complex variability is difficult to quantify in the absence of multispacecraft observations and, as a result, it is beyond the remit of this paper to consider such variability in detail Average Properties [18] Previous studies of the cushion region [Smith et al., 1976; Balogh et al., 1992; Kivelson et al., 1997] considered spacecraft data on an individual basis only and, as we have seen above, such a study tells us little about the average properties of the region. We address this problem for the first time by considering observations made by all Jupiter exploring spacecraft to carry a magnetometer to date. We do this by defining the average thickness of the cushion region at Jupiter to be the average separation between the cushion region s inner boundary (projected to the subsolar point using the Joy et al. [2002] magnetopause model) and the mean subsolar location of the Joy et al. [2002] magnetopause. Here we use the Joy et al. [2002] magnetopause location determined from a single gaussian fit to the spacecraft observations. Determining the instantaneous location of the cushion region s inner boundary does not require knowledge of the speed at which the boundary itself is moving though considerable ambiguity is often involved in its determination due to the gradual nature of the transition between the cushion region and magnetodisk. To reduce this ambiguity, passes on which a stable magnetodisk configuration ( B R > B, INT > 50 ) could not be identified were excluded from the analysis due to the resulting ambiguity in distinguishing adjacent magnetospheric regions. A total of 13 transition points could be identified in this way (between 1973 and 2003) and their distribution in the equatorial plane is shown in Figure 4. [19] The mean location of the cushion region s inner boundary maps to a subsolar location of 54 R J which suggests a mean cushion region inertial subsolar thickness of order L CR 20 R J. The 16 R J standard deviation in the location of the inner boundary is similar to that seen in the location of the Joy et al. [2002] magnetopause and is consistent with the effects of variable solar wind dynamic pressure modulating the size of the magnetospheric cavity. The range of observations is, of course, larger than the standard deviation quoted above and, once again, the Pioneer 10 inbound pass provides a good illustration of the variability. [20] Uncertainties in both the mean and standard deviation of the cushion region s inner boundary were estimated using a Monte Carlo method similar to that of Achilleos et al. [2008]: 500 random subsamples, each comprising 75% of the total number of observations used in this investigation, were used 5of14

6 the absolute rotational voltage across a 20 R J thick cushion region with a center 65 R J from the planet is then approximately 6 MV. A comparison between this value and the mean Dungey cycle reconnection voltage of 0.25 MV [Badman and Cowley, 2007] led Badman and Cowley [2007] and Kivelson and Southwood [2005] to conclude that the Dungey cycle contribution to the cushion region flux content is negligible under typical solar wind conditions. The outer magnetosphere of Jupiter is thus, predominately, a rotational phenomenon Scaling Jupiter to Saturn [23] For comparative purposes both the mean inertial subsolar thickness of the cushion region, L CR, and the mean rotational voltage across the cushion region at the subsolar point, V CR, must be appropriately scaled to the smaller Saturnian magnetosphere. Here we perform this scaling using the mean subsolar standoff distance of the magnetopause, R SS, and the total rotational voltage across the magnetosphere, V ROT, as shown below using values from Table 1: L CR ðsþ L CR ðjþ R SSðSÞ 6R S ð3þ R SS ðjþ Figure 4. (top) Nominal location of the Joy et al. [2002] magnetopause (dashed line) with the 1 s variability shaded. Circles represent individual spacecraft observations of the cushion region inner boundary: Pioneer 10, blue; Pioneer 11, gold; Voyager 1, pink; Voyager 2, black; Ulysses, red; Galileo, cyan. (bottom) The single fit gaussian distribution of magnetopause locations (dashed line) and cushion region inner boundary observations (solid line) projected onto the +ve X JSMAG axis using the Joy et al. [2002] magnetopause model. to calculate the mean and standard deviation of the subsolar location of the cushion region s inner boundary. The standard deviation of the resulting distributions was then used as a measure of the uncertainty in each parameter. The size of these uncertainties (3 R J and 2 R J for the mean and standard deviation, respectively) is probably indicative of the small number of observations available for analysis. [21] The mean rotational voltage across the cushion region at the subsolar point is a useful parameter for quantifying the relative contribution of different magnetospheric processes to its formation and evolution. It can be estimated by integrating the motional electric field, E CR = v CR B CR where v CR and B CR are the cushion region s plasma velocity and magnetic field, respectively, across the subsolar cushion region as shown below: jv CR j¼je CR L CR jjv CR jjb CR jjl CR j: Here we have assumed that v CR, B CR and L CR are mutually orthogonal. Assuming that cushion region plasma rotates at roughly 50% of the rigid corotation speed, the mean azimuthal velocity may be written as v CR 0.5WR CR where R CR is the mean radial distance to the center of the cushion region. With an outer magnetospheric field strength of order 10 nt, ð2þ V CR ðsþ V CR ðjþ V ROT ðsþ 200kV: ðjþ V ROT If the thickness of the cushion region scaled linearly with the typical subsolar distance to the magnetopause, one would expect a typical subsolar width of order 6 R S. Such a thickness should be readily apparent in Pioneer, Voyager and Cassini magnetometer observations. However, it is important to note that there is no a priori reason to believe that cushion region properties will scale linearly with these parameters. A more sophisticated scaling would take into account differences in the upstream solar wind dynamic pressure, dayside reconnection voltage, internal mass loading rate, planetary rotation rate and magnetospheric flux content at each planet. However our poor understanding of how these parameters interact to form the cushion region prevents us from constructing a scaling constant with greater physical significance at this time. The value of the above scaling comes instead from the simple and intuitive comparison between the Jovian and Saturnian magnetospheres that the scaling results permit. A Saturnian cushion region with a mean subsolar inertial thickness of roughly 6 R S will occupy approximately the same fraction of its parent magnetosphere as the cushion region seen at Jupiter. Similarly, a Saturnian cushion region associated with a mean subsolar rotational voltage of roughly 200kV would contain the same fraction of the magnetosphere s total rotational voltage as the cushion region seen at Jupiter. 3. The Cushion Region at Saturn 3.1. Typical Characteristics: Cassini Rev 20 [25] A typical pass through the (expanded) Saturnian magnetosphere is represented by the inbound leg of Cassini s Rev 20 orbit (9 January to 17 January 2006, dashed purple in Figure 3) along a low latitude dawn meridian. Spacecraft data are presented in Figure 5 with increasing radial distance ð4þ 6of14

7 Table 1. Comparison of Physical and Magnetospheric Parameters for Jupiter and Saturn Parameter Jupiter Saturn Definite Value Equatorial radius, R P [Cox, 2001] km km Magnetic moment [Cox, 2001] Tm Tm 3 Rotation period, t 9.925h [Cox, 2001] 10.66h [Cox, 2001] Rotation frequency (W =2p/t) s s 1 Axial tilt [Cox, 2001] Dipole tilt [Cox, 2001] 10 <1 Total rotational voltage, V ROT 400 MV [Badman and Cowley, 2007] 12 MV [Badman and Cowley, 2007] Long Term Average Outer magnetosphere B a 8nT 3 nt Plasma ion mass, m i 24 amu [Thomas et al., 2004] 18 amu [Khurana et al., 2007] Active moon source rate, S 10 3 kgs 1 [Vasyliũnas, 2008] 10 2 kgs 1 [Vasyliũnas, 2008] Equatorial radius of curvature, R C >2R J [Mauk and Krimigis, 1987] < 2 R S [Achilleos et al., 2010a] Magnetopause subsolar 75 R J [Joy et al., 2002] 24 R S [Achilleos et al., 2008] Standoff distance, R SS Magnetospheric flux tube 10 3 KgWb 1 [Pontius and Hill, 1989] 10 3 KgWb 1 [McAndrews et al., 2009] Mass content, s Dungey Cycle reconnection 0.25 MV [Badman and Cowley, 2007] MV [Badman and Cowley, 2007] Voltage, V D a Average outer magnetospheric field strengths were obtained from this study. Figure 5. Cassini Rev 20 observations in the Saturnian magnetosphere. (a) KRTP magnetic field components (B R,red;B, blue or white; B,green)and± B (black). (b) Normalized poloidal field components ( B / B, blue or white; B R / B, red). (c) Angle between the observed magnetic field, B OBS, and the magnetic field of the Burton et al. [2009] internal dipole, B DIP. Horizontal dashed lines denote the critical magnetodisk angles of 50 and = 130. (d) Thirty minute normalized magnetic field RMS variance. (e) CAPS/ELS thermal electron density (blue or white) and temperature (red). Vertical dashed lines denote points of K = 100 Kurth longitude [Kurth et al., 2008], separated by roughly one planetary rotation. The inner magnetosphere (blue), transition region (white), magnetodisk (yellow), magnetopause (red), and magnetosheath (grey) are shaded. The radial distance, planetocentric latitude, and local time of the spacecraft are shown along the x axis. 7of14

8 (decreasing UT time for an inbound pass) on the x axis. The final magnetopause crossing was made at a radial distance of 45.6 R S (R SS = 28.5 R S ) on the equatorial dawn meridian. The spacecraft then proceeded inward, moving through the dayside magnetosphere over a period of 10 days, to a 5.6 R S periapsis near the dusk terminator. As in section 2 for Jupiter, the qualitative structure of the observed magnetic field is most readily apparent when we consider the ratio of the poloidal field components to the total field magnitude ( B / B, B R / B ) and the angle, INT, between the observed magnetic field and the field associated with the Burton et al. [2009] internal dipole. [26] Two magnetospheric regions are apparent on this pass (c.f. the three observed at Jupiter) with a quasi dipolar region close to the planet, shaded blue, and a radially stretched ( INT > 50, B R > B ) region, shaded yellow, from 25 R S out to the magnetopause. These two regions are interpreted as Saturnian equivalents to the Jovian inner magnetosphere and magnetodisk, respectively, separated by a roughly 7 R S transition region, shaded white in Figure 5, where the mean field rotates through 30 and has properties intermediate between the two adjacent regions. It should be noted, however, that this is a very different type of transition region to the one described at Jupiter as it involves two qualitatively different magnetospheric regions. The absence of an equivalent transition region for the Ulysses pass at Jupiter is most likely due to differences in the spacecraft trajectories and the speed at which each spacecraft made transition into the quasidipolar region. Unlike the Ulysses observations made at Jupiter there is no evidence for a third, quasi dipolar region between the magnetodisk and magnetopause which might be interpreted as a Saturnian equivalent of the cushion region. [27] Both the inner magnetosphere and magnetodisk show variability with a period close to the approximately 10h30m planetary rotation period, however, unlike the periodic phenomena seen in the Jovian magnetosphere, these variations cannot be interpreted as a rotational flapping of the magnetosphere due to the small (<1 ) tilt of the Burton et al. [2009] internal dipole. They instead constitute the famous yet enigmatic rotational modulation (or camshaft signal ) first identified in magnetic field data by Espinosa and Dougherty [2000]. The superposition of the associated camshaft field onto Saturn s internal dipole creates a signature, near the equatorial plane beyond 15 R S, very similar to that of a rotating, tilted current sheet [Southwood and Kivelson, 2007]. However, for many of the low latitude orbits considered in this study the mean location of the current sheet is well above the equatorial plane such that crossings of the magnetodisk center (with B R reversing sign) are rare. The vertical displacement of the magnetodisk is primarily a result of the ambient solar wind which, due to Saturn s significant (26 ) axial tilt, impacts the magnetosphere from above (southern winter) or below (southern summer) the equatorial plane. The resulting asymmetry in magnetic pressure north and south of the equator then acts to warp the magnetodisk into a bowl shape [Arridge et al., 2008b]. [28] Observations of the thermal electron density and temperature are made by the CAPS/ELS instrument [Young et al., 2004; Lewis et al., 2008] throughout the Rev 20 orbit. In the quasi dipolar inner magnetosphere the thermal electron number density reaches values in excess of 1 cm 3 and the temperature is very low with a minimum, close to the planet, of just a few kiloelectron volts. The plasma density (temperature) decreases (increases) rapidly with distance and, in the radially stretched magnetodisk, the density often approaches values comparable with the noise level of the ELS instrument [Arridge et al., 2009]. At some points along the trajectory the count rate of thermal electrons drops below the threshold above which reliable plasma moments can be generated and thus the moments presented in Figure 5 are assumed to come from the denser regions of the Saturnian plasma sheet. Here the number density of thermal electrons is found to be of order 10 3 cm 3 while the corresponding temperature is of order 100 ev Saturnian Survey [29] During the period 1 July 2004 to 1 March 2009 Cassini made a total of 212 passes (106 orbits) through the Saturnian magnetosphere; six additional passes (3 flybys) were made collectively by Pioneer 11, Voyager 1 and Voyager 2 in the preceding interval of Of these 218 passes, 26 were associated with a low latitude, dayside trajectory and an innermost magnetopause crossing satisfying R SS 23 R S. The majority (16) of these were made through the postdawn (0600 < LT <1000) sector of the magnetosphere where, in analogy with Jupiter, the cushion region should be most evident, with five more made in the noon (1000 < LT < 1400) and predusk (1400 < LT < 1800) sectors, respectively. The general properties of these passes are shown in Table 2 and, as at Jupiter, only those passes associated with a stable, well defined magnetodisk ( INT > 50, B R > B ) were inspected for evidence of a cushion region. [30] For the 12 passes on which an unambiguous magnetodisk was identified, spacecraft observations reveal an extremely dynamic magnetosphere varying over multiple time scales of order minutes to days. The position and/or orientation of the magnetodisk appear to vary both during and between spacecraft passes at Saturn, as does the mean angle by which the magnetic field deviates from that of the Burton et al. [2009] dipole. In general, however, the magnetosphere is characterized by a two layered geometry (as observed on Cassini Rev 20; Figure 5) with a highly dipolar, B dominated region close to the planet, analogous to the Jovian inner magnetosphere, and a radially stretched magnetodisk ( > 50, B R > B ) at large radial distances out to the magnetopause. There is no evidence for a third magnetospheric region, between the magnetodisk and magnetopause, which might be interpreted as a Saturnian equivalent to the cushion region seen at Jupiter. Rotational periodicities are often present at Saturn, both in the inner magnetosphere and magnetodisk, but the physical origin of these periodicities is different to those seen at Jupiter and crossings of the magnetodisk are rare. Finally, the field in the Saturnian magnetodisk is also less stretched than that at Jupiter, as evidenced by the two poloidal field components lying closer together in value. These observations are consistent with the magnetodisk modeling work of Achilleos et al. [2010a] which shows that field lines in the Saturnian magnetodisk have a much larger radius of curvature (>2 R S ) than their Jovian equivalents. [31] The remaining passes can be divided into two categories. In the noon sector the reduced distance to the 8of14

9 Table 2. Summary of Low Latitude (l < 20 ), Expanded (R SS 23 R S ) Dayside Passes at Saturn as Identified Using the Arridge et al. [2006] Magnetopause Model and the Innermost Magnetopause Crossing for Each Pass Pass Direction R SS /R S LT/Decimal Hours l/degree Magnetodisk? Cushion? Voyager 2 OUT YES None observed Cassini Rev 00B OUT YES None observed Cassini Rev 020 IN YES None observed Cassini Rev 015 OUT YES None observed Cassini Rev 009 OUT YES None observed Cassini Rev 016 OUT YES None observed Cassini Rev 019 IN YES None observed Cassini Rev 003 OUT YES None observed Cassini Rev 018 IN YES None observed Cassini Rev 000 IN NO N/A Cassini Rev 005 OUT YES None observed Cassini Rev 017 IN YES None observed Cassini Rev 013 IN NO N/A Cassini Rev 008 IN NO N/A Cassini Rev 003 IN NO N/A Cassini Rev 012 IN NO N/A Cassini Rev 057 OUT NO N/A Cassini Rev 048 OUT NO N/A Cassini Rev 051 OUT NO N/A Cassini Rev 052 OUT NO N/A Cassini Rev 049 OUT YES None observed Cassini Rev 044 OUT NO N/A Cassini Rev 052 IN NO N/A Cassini Rev 051 IN NO N/A Cassini Rev 050 IN NO N/A Cassini Rev 048 IN NO N/A magnetopause prevents a magnetodisk from forming for all but the most expanded magnetospheric conditions (Cassini Rev 49 outbound, R SS >29R S ) and plasma depleted flux tubes, if present, are difficult to distinguish from their plasmaloaded counterparts. This fundamental ambiguity could not be resolved using CAPS/ELS plasma moments as the lack of clear magnetodisk crossings (discussed above for Rev 20) prevented us from obtaining the central plasma sheet density necessary for comparison. Here it is important to remember that, at Jupiter, the cushion region and magnetodisk lobes have essentially the same density; it is only the density at the magnetic equator (synonymous with the center of the plasma sheet in the magnetodisk) that changes upon entering the cushion region. The dusk sector field, in contrast, is typically disturbed such that a stable magnetospheric configuration, either magnetodisk like or dipolar, is difficult to define. Similar observations were made in the dusk sector of the Jovian magnetosphere where the plasma sheet becomes so thick [Kivelson and Southwood, 2005] that exploring spacecraft rarely leave the disturbed central region. In such cases a disk like interpretation of the magnetic field geometry is no longer applicable. 4. Discussion [32] The lack of a persistent, unambiguous region of quasidipolar, B dominated field between the dayside magnetopause and magnetodisk at Saturn is in stark contrast to the 20 R J thick, local time dependent cushion region typically observed at Jupiter. The apparent implication of this discovery is that the Saturnian dayside magnetosphere typically lacks an outermost layer of plasma depleted flux tubes and that dynamical processes in the Saturnian outer magnetosphere manifest themselves in a very different way to those observed at Jupiter The Importance of Plasma Return Flows [33] Both the Jovian and Saturnian magnetospheres are associated with a constant time averaged magnetospheric mass content and, as a result, the time averaged rate at which plasma is generated in the inner magnetosphere (primarily by processes related to Io and Enceladus) must equal the timeaveraged rate at which plasma is lost through reconnective processes occurring in the magnetotail. The difference between the Jovian and Saturnian inner magnetosphere source rates is typically assumed to be around an order of magnitude [Vasyliũnas, 2008]; however, caution must be applied when using this figure for the following important reason. Inner magnetosphere source rates are typically derived from a measurement of the momentum loading of magnetospheric field lines and, as a result, involve a significant contribution from charge exchange processes which change the momentum of magnetospheric plasma while not significantly altering its total mass. The inclusion of such charge exchange processes in the momentum loading calculation thus results in an overestimation of the true magnetospheric source rate by an unknown factor that is difficult to determine through experimental means. [34] The modeling work of Delamere et al. [2007] suggests that charge exchange processes may be an order of magnitude more important at Saturn than at Jupiter such that the true difference between the Jovian and Saturnian inner magnetosphere source rates may then approach a factor of 100. From an order of magnitude perspective, the average mass content of flux tubes in the Jovian and Saturnian magnetospheres (Table 1) is comparable. This fact, combined with the idea that average mass input equals average mass output, then implies that flux tubes in the Jovian magnetotail lose mass roughly 100 times faster than their Saturnian equivalents. This is similar to the factor of 30 difference in the total 9of14

10 rotational voltage across each magnetosphere (Table 1) and suggests that Vasyliũnas cycle reconnection is of a similar dynamical importance in each magnetosphere. Both systems should therefore contain a similar number of plasma depleted flux tubes (relative to their total flux content) and, all other things being equal, a Saturnian equivalent to the cushion region seen at Jupiter should be evident in the dayside outer magnetosphere. [35] Badman and Cowley [2007] estimate the mean reconnection voltage associated with the Dungey cycle to be 0.25 MV for Jupiter and MV for Saturn. The difference between these two values is smaller than for the Vasyliũnas reconnection voltage however theoretical calculations [Badman and Cowley, 2007] suggest a layer of plasma depleted flux tubes, adjacent to the dawn magnetopause at Saturn, with typical thickness of R S if the Vasyliũnas cycle is neglected. No evidence for such a layer was observed in this study. [36] Considering the Dungey and Vasyliũnas cycles as separate and distinct processes may be an oversimplification of reality as the microphysics of magnetotail reconnection at Jupiter and Saturn are poorly understood. Kivelson and Southwood [2005] suggest that the Dungey and Vasyliũnas cycles may close during the same reconnective episode in the magnetotail and, in this case, the total number of plasmadepleted flux tubes moving back around to the dayside will be altered from that expected when the two cycles are considered separately. Regardless of these details the above discussion makes it clear that plasma cycle return flows are likely to play an important role in the dynamics of Saturn s outer magnetosphere and that plasma depleted flux tubes returning to the dayside should be ubiquitous in spacecraft observations Depleted Flux Tube Configuration [37] If there are plasma depleted flux tubes in the Saturnian outer magnetosphere, why are these flux tubes not associated with the quasi dipolar magnetic field configuration seen in the Jovian cushion region? The warped nature of the Saturnian magnetodisk will change the magnetic field configuration expected near the equator (particularly at large distances from the planet) in a way that is sensitive to the solar wind dynamic pressure and the instantaneous location of the spacecraft. However, the fact that the Jovian cushion region is observed even when the magnetodisk is tilted more than 10 away from the equator suggests that such warping is unlikely to account for the observed lack of a cushion region at Saturn. [38] Of more importance may be the dynamical motion of the magnetodisk at each planet. At Jupiter the magnetodisk flaps up and down (in the rest frame of the spacecraft) due to the 10 tilt between the planets dipole and rotation axes. The regular transition into different magnetospheric regions that results from this motion may make differences in magnetic field topology easier to detect. At Saturn the equivalent motion is far more complicated and, at present, poorly understood. However, should the magnetodisk at Saturn move in a more restricted fashion than at Jupiter, differences in topological regions may be less obvious in spacecraft observations. [39] Our analysis of Saturnian spacecraft data was limited to passes associated with a relatively expanded magnetosphere (R SS >23R S ) in order to ensure that the interior magnetodisk configuration, so essential for defining entry into the cushion region, was present. When the magnetosphere is in a compressed state the magnetodisk configuration vanishes and the entire dayside magnetosphere is associated with a quasi dipolar geometry very similar to that of the extensively studied terrestrial magnetosphere. Under such conditions it is very difficult to identify plasma depleted flux tubes based on their magnetic field configuration alone and a detailed study of the associated plasma data becomes important. However, because of the very low densities observed in the outer magnetosphere, the collection and interpretation of such measurements is fraught with difficulties and beyond the scope of this investigation. [40] The Dungey cycle contribution to the cushion region s flux content is expected to increase during compressed magnetospheric conditions [Badman and Cowley, 2007] however the response of the dominant Vasyliũnas cycle is less certain. Although the steady state contribution of the Vasyliũnas cycle to the cushion region s flux content must depend only on the internal mass loading rate, the physical mechanism by which this flux is added to the region may be controlled by the upstream solar dynamic pressure [Zieger et al., 2010] in a complicated fashion. However, even if the Vasyliũnas cycle is neglected, Badman and Cowley [2007] still predict a cushion region thickness of order 0.5 R S under expanded magnetospheric conditions. This suggests that restricting our analysis to the expanded magnetosphere will not significantly bias our conclusions in this study. [41] One potential explanation for the observations described above lies in the physics of the magnetodisk itself. The disk is created as corotating magnetospheric plasma, strongly confined to the centrifugal equator, balloons outward under the influence of centrifugal forces. The frozen in magnetic field gets dragged out with the expanding plasma and eventually adopts the classic magnetodisk configuration. This expansion is generally thought to be restricted on the dayside by the dynamic pressure of the solar wind acting on the magnetopause such that only on the nightside of the planet can the expansion continue to the point of reconnection. Such reconnection typically removes mass from the system and keeps the total mass content of the magnetosphere in a state of long term quasi equilibrium. [42] Consider, however, a situation in which the dynamic pressure of the solar wind is particularly low such that the magnetopause retreats to large distances on the dayside. Is it possible, under these conditions, for the stretching of the magnetodisk to continue to the point of reconnection on the dayside too? And if so, what are the likely implications of this reconnection for the large scale structure and dynamics of the system? In the compressed state, where the magnetodisk extends all the way to the magnetopause, any plasmadepleted flux tubes present on the dayside will be draped over the magnetodisk and, as a result, retain a significant radial component. The north south thickening of the current sheet due to this draped magnetic flux is likely to be negligible and probably undetectable. In the expanded state the radial distance at which magnetodisk reconnection occurs also represents the maximum distance at which a stable magnetodisk like configuration can be maintained. Beyond this distance corotation can no longer be enforced and plasmadepleted flux tubes will be able to relax into a more dipolar 10 of 14

11 Figure 6. The critical ion density, N C, required for magnetodisk breakdown in the (top) Jovian and (bottom) Saturnian magnetospheres is shown in red. The measured thermal electron density is shown in blue, and the 1 s variability in the mean location of the Joy et al. [2002] (Jupiter) and Arridge et al. [2006] (Saturn) magnetopause is shaded in grey, mapped to the local time of the magnetopause crossing. Vertical black lines denote the actual location of the magnetopause observed by each spacecraft. configuration reminiscent of the cushion region. Small clumps of plasma containing closed loops of magnetic field may detach from the outer edge of the magnetodisk but, for the most part, this plasma will remain inside the magnetopause and eventually diffuse onto adjacent magnetospheric field lines. Eventually, this plasma will be lost in the magnetotail. [43] Can this effect explain the formation of the cushion region at Jupiter and, similarly, the absence of a cushion region at Saturn? The radial distance at which the magnetodisk breaks down can be estimated from the magnetospheric force balance condition (equation (1)) if we make the simplifying assumptions that m i m e and that the poorly constrained outward acting pressure gradient forces are equal to some multiple, k, of the centrifugal force. [44] Rearranging equation (1) for the field line radius of curvature, R C, under these assumptions we obtain B 2 R C ðk þ 1Þ 0 N i m i W 2 ; ð5þ from which it is immediately apparent that, for increasing outward forces, the field line radius of curvature will decrease. This decrease cannot continue indefinitely, however, as the radius of curvature will eventually approach values comparable with the ion gyroradius. At this point the MHD assumptions upon which equation (1) is based will begin to break down and, for a low energy plasma, the magnetic field is likely to develop an x line. Reconnection at this x line may explain many of the magnetic nulls observed in the transition region at Jupiter by Haynes [1995]. [45] Assuming that ions have a kinetic energy E KE (3/2) k B T i, where T i is the mean ion temperature, the mean ion gyroradius, R g, for equatorially mirroring particles may be expressed as R g ¼ m i 2k B T 1=2 i ; ð6þ jq i jb m i where q i is the mean electromagnetic charge on an ion. Setting R C = R g in equation (5), substituting for equation (6) and rearranging for N i gives jq i jb 3 m 1=2 i N C ; ð7þ ðk þ 1Þ 0 m 2 i W2 2k B T i where N C is now the critical number density required for the ion gyroradius and field line radius of curvature to be equal. To estimate the distance at which this critical density is reached in each magnetosphere the value of N C must be compared with the observed number density measured by in situ spacecraft. This comparison is presented in Figure 6 for the Ulysses (inbound) pass at Jupiter and Cassini Rev 20 (inbound) pass at Saturn where it has been assumed that ion and electron number densities and temperatures are equal. This assumption is qualitatively consistent with the results of Thomsen et al. [2010]. 11 of 14