Saturn s mysterious magnetism

Transcription

1 Saturn s mysterious magnetism In his 2013 RAS Presidential Address, David Southwood looks at the surprises that have come from the exploration of Saturn s deceptively simple magnetic field. 1: This false-colour image, made from Cassini near-infrared data, shows green aurorae streaking out about 1000 km above the cloud tops of Saturn s south polar region. Note the alignment of the aurora with the planet s banded clouds and rotation axis. Blue indicates reflected sunlight at 2 3 µm, green, light from hydrogen ions at 3 4 µm, and red, thermal emission at 5 µm. The dark spots and banded features in the image are silhouettes of clouds and small storms that outline the deeper weather systems and circulation patterns. (NASA/JPL/ASI/Univ. Arizona/Univ. Leicester) Writing a proposal to NASA from Imperial College to put a magnetometer on the Cassini spacecraft, 25 years ago, I never dreamt that the basic result promised would remain unresolved today. The key goal of the instrument was to determine the internal magnetic asymmetry of Saturn. We had no idea of the surprises to come. So far, Saturn remains an aligned rotator, with planetary rotation and field symmetry axes aligned within our present capacity to measure. However, magnetic axial symmetry is broken globally outside the planet s surface. A complex web of exterior fields associated with the planet s magnetosphere and ionosphere obscures the apparently simple aligned rotator within; how this happens is the primary topic of this paper. Saturn s abiding axisymmetry remains a troubling problem for planetary dynamo theorists. As Cao et al. (2011, 2012) point out, Cowling s anti-dynamo theorem (Cowling 1933), which states that an axisymmetric magnetic field cannot be maintained by an axisymmetric motion, made an aligned rotator with a dynamo field unlikely. Following the first reports of Saturn s axisymmetric field, Stevenson (1982) suggested that there was a conducting layer above the dynamo region in which a skin effect in the stratified rotation suppressed the non-axisymmetric terms. However, no dynamo model has yet successfully produced the effect envisaged. Alternatives are being sought; Saturn s field is substantially weaker than the highly asymmetric field of Jupiter, which may be a clue. Another clue is the observed polar intensification of the field; recently, Cao et al. (2011, 2012) have introduced differential rotation in the dynamo region to produce this and interest is centring on Couette flow models where the flow that makes the field occurs between inner and outer shells and where the field becomes very substantially wound up. Models currently produce energy in the poloidal field that is perhaps 0.05 of that in the toroidal field; in a convection driven dynamo, the ratio is close to unity. Seven years of Cassini data (from orbit insertion in 2004 to 2011) have shown no firm deviation from axisymmetry. Nonetheless, Cassini has not flown in close orbits; it was only during Saturn orbit insertion (SOI) in June 2004 that Cassini braved the rings and passed only 0.3 R S (R S = Saturn radius, km) above the cloud tops. The field detected during the SOI pass does appear to show a deviation from the standard model. However, the single pass had very limited planetary longitude coverage and the data preclude any definitive statement. However, it does whet one s appetite for the later stages of the Cassini mission when Cassini will repeatedly pass close to the planet in what are called the proximal orbits. Near periapsis the orbit will be taken inside the innermost ring (the D ring) passing only a few thousand kilometres above the planet s cloud tops. Before that time, we have to complete the challenge of calibrating and removing efficiently, the rotating external fields that the Saturn system creates. Saturn s rotation During the Voyager spacecraft flybys of Saturn in 1980 and 1981, the planetary radio astronomy instrument gave the first indications of A&G February 2014 Vol

2 2: Data from inbound passes of Pioneer 11 and Voyager 2 plotted against time but also radial distance from Saturn (in units of Saturn radii, R S = km). CA marks closest approach. Both approaches were in the morning sector of local time. The component shown is the azimuthal component B f. The continuous lines show sinusoids adjusted for the Doppler shift associated with spacecraft angular motion on the assumption that the signals are rotating with the planet. In other words, the signals appear to be stationary in what was believed at the time to be the planetary frame. 3: Magnetic field data from the Imperial College magnetometer on the Cassini spacecraft for the period 2 11 April 2010 inclusive. 1 min resolution data averaged over 1 hr are shown. The coordinate system is based on the planetary radial (r), co-latitudinal (q) and azimuthal (f) directions. The upper panel shows the radial (B r black trace) and southward (B q red trace) components of the field. Near periapsis, the B q (red) component goes off scale peaking at 457 nt. The lower panel shows the residual radial and southward components after subtracting the standard offset planetary dipole field model along with the azimuthal component B f (green trace). rotational asymmetry (Desch and Kaiser 1981). They discovered the Saturn kilometric radiation (SKR) pulsing with a period near 10.7 hr (639 m 24±7 s). The period of the radio pulses was immediately interpreted as the deep interior rotation period of the planet. Puzzlingly, the direct measurements of the planetary magnetic field during Pioneer and Voyager flybys of Saturn by separate teams known for their competitive streak revealed an internal planetary magnetic field, axisymmetric to octupole order (represented often as a dipole offset along the rotation axis to the north by 0.04 R S ). It was assumed that there must be a high order and probably high latitude magnetic anomaly ordering the radio data that the flyby orbits near the ecliptic had not detected. Saturn s kilometric radio signals could not be detected at Earth. The Ulysses solar polar orbiting spacecraft was, however, equipped with a very sensitive radio receiver and had its perihelion near Jupiter s orbit, roughly half way between the Sun and the orbit of Saturn. Where Ulysses and Saturn were on the same side of the Sun, the SKR could be detected. Using measurements from the Ulysses spacecraft from 1994 to 1996, Galopeau and Lecacheux (2000), found a small variation, ~1% per year, in the period of the SKR radio emission. This seems too rapid to arise from secular changes in the rotation period of so massive a body as Saturn and the first doubts about the use of pulsed radio emissions for determining an object s rotation rate were sown. Doubts even arose over whether the pulsed signals were generated by a rotating source although this assumption, unlike the former, has survived. At Imperial we found ourselves facing a more complicated task at Saturn than we d envisaged a decade before. In preparation, we looked at Voyager and Pioneer magnetic data from two decades earlier. Re-analysis of the Pioneer and Voyager spacecraft magnetic data by Espinosa and Dougherty (2000) and Espinosa et al. (2003a, b) indeed showed features pulsing with a period of 10.7 hr. However, the polarization of the magnetic pulsations showed that they came from a source outside the planet. Figure 2 shows the sinusoidal transverse fields detected it was as if there was a rotating pressure enhancement or cam rotating far out in the planet s magnetosphere. Moreover, although it took years to be sure that the radio signal sources were rotating, it was proved right away that the magnetic signals had an azimuthal wave number of 1 and they were rotating in the same sense as the planet. The term cam has stuck, at least for the field in the dipolar region of the magnetosphere. A cam removes rotation symmetry and that is what Saturn s external fields do. Cassini arrives The Imperial College group s discovery was vindicated when the Cassini spacecraft arrived in orbit in mid-2004 and found that the 10.7 hr oscillations were more or less ubiquitous in Saturn s magnetosphere. If Espinosa et al. s (2003a) polarization argument for the external force was subtle to some, what made its external nature abundantly clear was that the amplitude did not fall off as the cube or higher order power of radial distance from the planet. Indeed, the signal amplitude was remarkably constant over radial distances of tens of R S, as shown in figure 3. In 11 days, the spacecraft comes from a distance of 36.6 R s from the planet, passes through periapsis at 3.5 R S and then returns to 34.3 R S. The lower panel shows the field after the model (axisymmetric offset dipole) field has been subtracted, eliminating the background field. A large depression in the B q component remains around closest approach. This is the effect of the ring current, the energetic charged particles trapped in the field and drifting about the planet. There was a surprise in that the ring current extends in to periapsis. Initially, charged 1.14 A&G February 2014 Vol. 55

3 mat erial from the rings was suspected as the primary source. However, in a separate shock delivered by the Saturn system, geysers were discovered on the tiny moon Enceladus, first through the magnetic perturbation (Dougherty et al. 2006) and then subsequently by imaging. Enceladus, whose orbit is at ~4 R S, just outside the main rings, is the primary source of positively charged material, water group ions, in the inner magnetosphere. The data in figure 4 are from the same time as the data in figure 3, but the ring current depression has been removed from the B q component. Moreover, the ordinate is no longer time but a rotating phase with period matching the SKR radio period. Although the signals are not entirely sinusoidal, they are periodic and also well fit by the rotating phase at all local times and over an enormous range of radial distance. Within R ~ R S, the transverse DB r and B f components are in quadrature with DB r leading B f. The rotation of the fields is in the same sense as the planetary rotation. Beyond R ~ R S, the background field switches fairly rapidly to being radially in or outwards (as can be seen in figure 3). DB q and B f are then in quadrature with DB q leading and DB r and B f in antiphase. Figures 3 and 4 have been chosen when the spacecraft orbit is both in the equatorial plane of the magnetosphere and also near equinox. At equinox, solar illumination of the northern and southern polar caps is about equal and the ionosphere and magnetosphere are at their most symmetric. But the signals are not symmetrical. During cycles , in figure 4, the spacecraft is on closed dipolar field lines, i.e. field lines that extend from one hemisphere to another in the inner magnetosphere. It is surprising that all three components are present in a plane of symmetry the equatorial plane of the system. Were the signals also to be north south symmetric one would expect at least one field component to have a node, i.e. equal zero. The lack of symmetry implies transport of momentum and energy across the equator. The explanation is that the signals are made up of two separate northern and southern components, oscillating at slightly different frequencies, another one of Saturn s surprises. A further surprise The discovery of north south asymmetry in the 10.7 hr signals came very early in Gurnett et al. (2009) showed that radio signals originating in the northern and southern hemisphere were pulsing at slightly different rates: the southern hemisphere signals were pulsing at 648 minutes period, whereas the northern signals pulsed at 636 minutes. I wondered whether the magnetic signal had a similar asymmetry and fate allowed me to find out. In late 2008, the Cassini orbit had been adjusted to adopt a very high inclination with a short orbital period of 4: The data from the same time (2 11 April 2010) as in figure 3. The ring current signature has been removed from the B q component. The abscissa is no longer time but the phase of a signal rotating about the planet at the SKR pulse period (the phase used is that of the mean frequency). The annotation is in cycles from 1 January : The three panels show data from two successive orbits of Cassini. The red trace is shifted by 281 hr (roughly one orbit period) with respect to black trace. Data have been smoothed to remove variations on timescale of less than one hour and detrended. Black trace data are from days (2 February 20:00 to 28 February 13:00) and red trace data are from days (14 February 13:00 to 12 Mar 11:00 UT). The labelling and shading indicate the hemisphere in which the spacecraft is. The phase difference between signals an orbit apart is zero when the spacecraft is in the south for all three field components. In the north, the phase difference between orbits is close to p, again for all three components. around 10 days. Moreover, the orbit was more or less symmetric north south, ideally suited for the discovery that northern and southern magnetic signals were indeed also oscillating at slightly different periods (Southwood 2011). Figure 5 shows the three field components for three identical orbits in total, shown by dint of the red trace being shifted 281 hours back in time, very close to one orbital period. Shading indicates when the spacecraft is in the southern hemisphere, where it is evident that the red and black components are in phase. In the north, the red and black components one orbit apart are in antiphase. There must be a slight difference in period between north and south, just as in the radio data. The 1.7 2% difference in radio period meant that the expected beat cycle between northern and southern periods would be about cycles, giving a half beat period of about 12 days. The happy and remarkable coincidence of the relation between the orbital period in and the half beat period provided the most unambiguous phase differences possible. Leicester leaps in The Leicester group, primarily Gabby Provan and David Andrews working with Stan Cowley, had done a lot of work surveying and codifying the 10.7 hr signals since They set to work to identify northern and southern signals and their time evolution (Andrews et al. 2010, Provan et al. 2011) throughout the mission. Importantly it was established that the northern and southern compressional signals (i.e. the component along the background field) had different phase relationships with the transverse components. In the closed field regions, the northern/ southern B q component is in phase/antiphase with the transverse component in the meridian. This can then be fed into the decoding of the signals. An excellent analysis resulted (Andrews et al. 2012) which traced separately the northern and southern signals almost continuously from A&G February 2014 Vol

4 6: The figure shows the traces of the peak normalized amplitudes (R) of south (A) and north (B) signals resulting from using a demodulation process that then homes in on the signal using requirements of phase coherence, level of agreement between components and continuity. The ordinate is period in hours. The horizontal dashed line shows the guide periods for initial demodulation in each case (10.75 and hr respectively). Full details of the signal processing methods used are given in Andrews et al. (2012). The abscissa is marked in days since 1 January The start of each year is marked along the top. (Provided by D Andrews, adapted from fig. 7 of Andrews et al. 2012) 7: Schematic illustrating the Andrews et al. (2012) determination of the distribution of northern and southern signals. Cassini s arrival at Saturn. Data from Andrews et al. (2012) is displayed in figure 6. Cassini orbit insertion took place in southern Saturnian mid-summer. By 2011, it was past equinox (August 2009). Andrews et al. (2012) found that the ratio of southern to northern amplitude reduced from around 3 in mid-summer to about 1 around the time of equinox. Tantalizingly, the northern and southern periods do approach a mean period of ~642 min around equinox. Could this mean that solar illumination of the polar cap was at the root of the seasonal variation? Probably not. The ionospheric seasonal illumination hypothesis has a basic flaw. If a seasonal north south period difference were associated with changes in the solar illumination of the polar caps, the ionization (and therefore the ionospheric conductivity) of the summer cap would be the greater in summer. Prima facie, one expects the summer cap then to be more tightly bound to the neutral atmosphere and so to rotate faster not slower in mid-summer. For the moment, the seasonal variation remains a mystery to be resolved. Work continues on tracking the evolution of northern and southern periods (Provan et al. 2013). Post-equinox, the system seems to have adopted a different behaviour with abrupt (~1 Cassini orbital period) switches between southern and northern period. The periods themselves are reported to be relatively stable, min (north) and (south), but one can note that the southern rotation remains the slower although we are now well past equinox. Theories, speculation The magnetic oscillations modulate nearly all Saturn magnetospheric processes and it is thus likely that they arise from a global process. Gurnett et al. (2007) and Goldreich and Farmer (2007) proposed a two-cell magnetospheric circulation rotating with the planet transporting material from the Enceladus orbit out to interplanetary space as the cause. The idea had first been put forward by Hill et al. (1981). As an author on the Gurnett paper, I keep the idea in my mind (and suspect that a variant of it will be in the final answer), but, as presented, it doesn t work for a global circulation. Notably, in the dipolar regions of the magnetosphere, the heavy Enceladus material is observed to rotate at only about half the planetary rotation speed. In parallel, Southwood and Kivelson (2007) picked on the north south asymmetry of the signals and proposed a rotating system of fieldaligned cam currents connecting northern and southern ionospheres, roughly at the limit of the dipole field s dominance. The strongest currents in the Saturn system have been detected in the vicinity of the magnetic shells in question, (see, for example, Talboys et al. 2009). Southwood and Kivelson (2009) developed the idea further, suggesting that the rotating currents and their shear fields interact with a field pattern fixed with respect to the Sun due to the solar wind stress beyond the dipole region. The peak solar-wind-induced stress is in the mid-morning hours, consistent with the radio observations. Nonetheless, ideas developed before 2009, however suggestive, had not come to terms with the dual-period nature of the signals. Why does Saturn s external magnetic field seem to break its symmetry separately, north and south? The most important clue lies in a further result from Andrews et al. s (2012) survey. When the Cassini spacecraft was on flux tubes in either polar cap, the signal detected was, within the ~10% uncertainty, pristine, i.e. pure southern or pure northern. On the other hand, when the spacecraft was in the closed field regions the signal was a mixture. The result, summed up in the schematic in figure 7 suggests that the sources are indeed directly linked to the motion of the caps (Southwood and Cowley, in prep.). Unfortunately, the speed at which the spacecraft passes through the polar cap boundaries (at between invariant latitude) is so high that the location of the change from pristine to mixed signals is not precise. In the central polar caps the flux tubes extend far out into interplanetary space; the plasma there is likely to be tenuous. Very little torque is needed from the ionosphere and these tubes will probably rotate at close to 10.7 hr. On the flanks of the cap, however, there is likely to be solar wind material which has entered either directly through the dayside cusp or through magnetic reconnection on the dayside as part of the Dungey cycle (Cowley et al. 2004). Details of the high-latitude circulation of this material may well yield the answer to our problem. Can a rotating circulation be set up in the regime between the permanently open field of the polar cap proper and the permanently closed field lines which contain the heavy ionized material from Enceladus and the rings in the deep interior of the magnetosphere? Jia et al. (2012) 1.16 A&G February 2014 Vol. 55

5 8: Aurorae on Saturn can last for days very different from those on Earth, which last only about 10 minutes. And on Saturn it is the pressure of the solar wind that appears to drive auroral storms, whereas on Earth the main driver is the Sun s magnetic field, carried in the solar wind. (NASA, ESA, J Clarke [Boston Univ.], Z Levay [STScI]) and Jia and Kivelson (2012) have provided a simple modification of a University of Michigan global magnetospheric computer simulation that was calibrated to fit the magnetic fields seen, but has turned out to then explain many other observed features of the magnetosphere. Their model imposes twin vortices (azimuthal wave number m = 1, Southwood and Kivelson 2007) in the neutral atmosphere in each hemisphere centred near the expected open closed field lines boundary. The neutral motion then drives plasma vortices and results in a global magnetospheric response. The proposal of a neutral atmosphere source (originally made by Smith 2006) is not unique, but the success of the Michigan work in modelling many observed aspects of the system is a big step forward. It shows what magnetospheric consequences can follow from a simple twinvortex rotating ionospheric flow pattern. However, because the Michigan model treats the ionosphere magnetosphere coupling along the field direction in a quasi-static limit, very similar results arise for vortices set up by motions driven in the ionized medium. The location of the northern and southern vortices not only at magnetically conjugate sites but also on the magnetic shells where the dipole field is ceasing to be dominant make this seem an appealing line of enquiry. An example of the 10.7 hr magnetic oscillations deep in the northern polar cap and tail lobe of Saturn is shown in figure 9, from a distant high-latitude Cassini pass in late Barely perceptible in the top panel is a compressional oscillation in field strength which shows up in the radial component shown as second (dotted) trace oscillation plotted in the middle panel. The oscillations look unremarkable until one recalls not only that the spacecraft is km from the planet, but also that the signals imply a very large-scale length. An order of magnitude estimate of the spatial displacement of the plasma (from frozen-in field theorem) yields x ~ l b B where l is a parallel scale length. A minimum value for l would be approximately the spacecraft radial distance, 30 R S, although it is likely to be larger (e.g. the parallel wavelength of an Alfvén wave, ~600 R S ). Taking the latter value is a reductio ad absurdum the scale of motion would then be ~100 R S, exceeding the transverse dimensions of the system. Magnetohydro dynamics deals with this by introducing a compression magnetic component to limit transverse motion, seen here as the small compressional B r oscillation. A reasonable deduction for the plasma displacement amplitude might be x ~ 5 10 R S. In fact, as the transverse field components can be seen to be nearly in quadrature and of similar magnitude, the local plasma is not moving just back and forth but is executing a circular motion, transverse to the field direction with radius of order x. Most, if not all, of the polar cap flux tubes are rotating similarly. At the author s request, a simple computational test was recently undertaken by Xianzhe Jia of the University of Michigan. Jia (personal communication, 2013) looked at an axisymmetric cylindrical column of collision-free plasma threaded by a magnetic field with the field line feet embedded at one end in a perfect conductor. The plasma was forced into rotation about the symmetry axis by setting the conductor rotating. The plasma not only started to rotate but slewed back and forth transverse to the background field at the rotation rate. The latter motion, introduced by the plasma, is called a kink motion in solar physics (Edwin and Roberts 1983). The slewing removes the original rotational symmetry and the kink motion sets up compressional oscillations in the rotating tube. If this simple simulation can be extended to fit the more complex environment of the open field lines of the Saturn polar cap mag netosphere, one has a solution for the origin of the global 10.7 hr Saturn magnetic field oscillations. In fact, the ionosphere is not a perfect conductor. Nonetheless, at ionospheric levels the travel time for MHD signals around the planet is certainly much less than the 10.7 hr period. The signal will be quasi-static and be described A&G February 2014 Vol

6 Planets and pulsars 9: 10.7 hr magnetic oscillations deep in the polar cap/tail lobe of Saturn. The data is taken from the Imperial College magnetometer on the Cassini spacecraft. 10 rotation cycles and around 110 hr of data are shown. The top panel shows the strength of the mainly radial background field, the two lower panels show at larger scale the meridional (q) and azimuthal (f) components. Also shown in the middle panel is the oscillation in radial component (filtered to zero mean). by an electrostatic potential: equipotentials would be instantaneous streamlines for horizontal motion. The observed m = 1 (longitudinal wave number) nature of the Saturn signals (Southwood and Kivelson 2007) means the potential varies around the polar cap boundary and the electric field cannot be zero there; in consequence, the rotating disturbance moves the polar cap boundary north and south. At any instant, there should be poleward motion in one (planetary) longitude sector and equatorward in the other, i.e. an equatorward bulge on one side of the polar cap boundary and a depression or poleward displacement on the other. The auroral zone does in fact rock in just this way (Nichols et al. 2008, 2010). Why are the signals present? The computation by Jia shows that they are associated with the transmission of torque along the field to maintain the polar cap in rotation. Hitherto it had been assumed that a torque would be uniformly exerted on the open polar cap, putting a local time-independent twist into the field (Isbell et al. 1984, Milan et al. 2005). There is no observational evidence of that field. Concluding remarks This report is one of work in progress. The magnetic field of Saturn has provided surprise after surprise. We ve come a long way in sorting out the external field but the story is not over. The Imperial College Cassini magnetometer was expected to identify the manner in which the internal planetary field departs from axisymmetry; the final multiple orbits of Cassini should provide a dénouement. In the meantime, we hope to wrap up our understanding of the external field prior to this new phase of exploration. However, in time-honoured fashion, Saturn s external field has displayed a new mode of behaviour post-equinox and has started switching between northern and southern dominance in an almost chaotic manner (Provan et al. 2013). We re not out of the woods yet. D J Southwood, Blackett Laboratory, Imperial College, London, UK. Acknowledgments. The author expresses particular thanks to Xianzhe Jia (University of Michigan) for his computational work; to David Andrews, particularly for provision of figure 6 to the author s specifications; and Michele Dougherty for her superb stewardship of the Cassini magnetometer since Cassini SOI. He also acknowledges useful discussions with many colleagues. Stan Cowley, Emma Bunce, Margaret Kivelson, Gabby Provan and Vytenis Vasyliunas deserve particular mention. Support is also acknowledged from the University of Michigan Dept of Ocean Atmospheres and Space Science general fund as well as a Senior Research Investigator position at Imperial College. Cassini magnetometer data reduction at Imperial College has been supported by UK Space Agency through a contract from the European Space Agency. References Andrews D J et al J. Geophys. Res. 115 A Andrews D J et al J. Geophys. Res. 117 A Arons J 1979 Space Sci. Rev Cao H et al Earth and Planetary Science Letters Cao H et al Icarus Cowley S W H et al Annales Geophysicae Cowling T G 1933 Mon. Not. R. Astron. Soc Desch M D and Kaiser M L 1981 Geophys. Res. Lett Dougherty M K et al Science There is an interesting immediate spin-off for pulsar astrophysics in our recent work. A neutron star need not have a dynamo and one might expect such a collapsed object to have aligned magnetic and rotation axes. The simplest model (Goldreich and Julian 1969) of a pulsar is based on an aligned rotator. One of the conundrums of the model (see e.g. Arons 1979, Kennel et al. 1979, Michel 2004, Timokhin 2006) is that the inherent axisymmetry appeared to mean the system would not pulse. Yet Saturn has found a way to do it. Saturn s field and rotation axis unquestionably closely align like Goldreich and Julian s ideal pulsar, and Saturn manages to emit pulses of radio waves at close to its rotation rate. The plasma environment around Saturn, its magnetosphere and ionosphere, break the angular symmetry imposed by the central object. The angular variation of the magnetosphere electrodynamics leads to pulsing field-aligned currents in the auroral zones of the planet and a byproduct is that radio waves are emitted in the kilometric band. Given that it is very likely that a rotating neutron star would be embedded in a plasma environment, could not the plasma around a pulsar break the inherent symmetry, just as at Saturn? Edwin P M and Roberts B 1982 Solar Phys Espinosa S A and Dougherty M K 2000 Geophys. Res. Lett Espinosa S A et al. 2003a J. Geophys. Res Espinosa S A et al. 2003b J. Geophys. Res Galopeau P H M and Lecacheux A 2000 J. Geophys. Res Goldreich P and Farmer A J 2007 J. Geophys. Res. 112 A Goldreich P and Julian W H 1969 Astrophys. J Gurnett D A et al Science Gurnett D A et al Geophys. Res. Lett. 36 L Hill T W et al J. Geophys. Res Isbell J et al J. Geophys. Res Jia X et al J. Geophys. Res. 117 A Jia X and Kivelson M G 2012 J. Geophys. Res. 117 A Kennel C F et al Space Science Reviews Michel F C 2004 Adv. Space Res Milan S E et al J. Geophys. Res. 110 A Nichols J D et al J. Geophys. Res. 113 A Nichols J D et al Geophys. Res. Lett. 37 L Provan G et al J. Geophys. Res. 116 A Provan G et al J. Geophys. Res. Space Physics Smith C G A 2006 Ann. Geophys Southwood D J 2011 J. Geophys. Res. 116 A Southwood D J and Cowley S W H submitted to J. Geophys. Res. Southwood D J and Kivelson M G 2007 J. Geophys. Res. 112 A Southwood D J and Kivelson M G 2009 J. Geophys. Res. 114 A Stevenson D J 1982 Geophys. Astrophys. Fluid Dyn Talboys D L et al Geophys. Res. Lett. 36 L Timokhin A N 2006 Mon. Not. R. Astron. Soc A&G February 2014 Vol. 55