The dynamics of Saturn's magnetosphere are driven by two major features of the system: First, the dominant source of magnetospheric plasma is the icy moon Enceladus, located deep within the magnetosphere; and second, like Jupiter, Saturn is a very fast rotator, with a rotational period of only 10.7 h. The dynamical imperative is to rid the magnetosphere of the continuously supplied plasma, and the fast rotation provides the mechanisms to do so. Thus, magnetospheric dynamics are intimately related to mass transport processes, including radial diffusion, flux tube interchange, magnetic reconnection, and plasmoid formation. We review recent progress and new questions relating to these processes at Saturn.
 Dynamical processes in the Earth's magnetosphere are largely driven by the coupling of mass, momentum, and energy from the solar wind. Outflow of material from the upper atmosphere introduces significant modifications to the processes, but the dominant driver remains the solar wind. By contrast, while there is certainly evidence for dynamical interaction with the solar wind, Saturn's magnetospheric dynamics appear to be dominantly driven by two major factors: (1) The dominant source of plasma lies deep within the inner magnetosphere itself, and (2) the planet rotates very rapidly. The problem addressed by this review is how Saturn manages to remove its internally produced plasma. As we will see, the rapid rotation of the planet provides the dynamical means to accomplish that task.
 Flybys of Saturn by Pioneer 11 (1979) and the two Voyager spacecrafts (1980 and 1981), followed by the treasure trove of data from the Cassini spacecraft, which has been in orbit around Saturn since 2004, have demonstrated that Saturn's magnetospheric plasma is dominated by water-group ions (O+, OH+, H2O+, H3O+) and H+ that are now known to originate in the water jets that emanate from Saturn's tiny icy moon Enceladus, located only 4 Rs (Saturn radii) away from the center of the planet. Those jets produce a toroidal cloud of neutral water molecules centered on the orbit of Enceladus. Some of the water dissociates into constituent neutrals, and then water-group plasma is produced through a combination of photoionization and electron impact ionization. Estimates of the ionization and recombination rates within the inner magnetosphere [e.g., Sittler et al., 2008] reveal that a rough steady state to the magnetospheric content requires that plasma mass be shed through outward radial transport, with eventual loss to the solar wind at the magnetopause or down the magnetospheric tail. The estimated plasma production rate is in the range of 12–250 kg/s [Bagenal and Delamere, 2011], and the dominant presence of water-group plasma in the outer magnetosphere confirms that outward transport does occur.
 The rapid rotation of the planet, coupled through the ionosphere by magnetic-field-aligned currents into the magnetosphere, produces fast plasma flows in the corotational direction, which dominate plasma motion throughout the magnetosphere right up to the magnetopause [e.g., Thomsen et al., 2010]. The fact that these flow speeds lie significantly below the value of full corotation testifies to the ongoing mass and momentum loading by new ionization and charge exchange in the inner magnetosphere, combined with outward transport of the new plasma [e.g., Pontius and Hill, 2009]. These rapid flows at large distances from the planet (up to 20 or 30 Rs or more) produce very strong centrifugal forces, which provide the means for dynamic processes to enable outward mass transport.
 Our understanding of the general configuration of Saturn's magnetosphere and the dynamical processes operating there as of the end of Cassini's prime mission (June 2008) has been comprehensively reviewed [e.g., André et al., 2008; Gombosi et al., 2009; Mitchell et al., 2009b; Mauk et al., 2009; Kurth et al., 2009; Arridge et al., 2011]. The purpose of the present paper is to update this understanding with recent progress in the specific area of transport-related dynamical processes and to identify new questions that have been raised. Necessarily, a number of very important topics will not be addressed at all, including the role of dust, the distribution of high-energy particles, periodicities, moons, waves, aurorae, plasma instabilities, and radio emissions. Each of these topics is worthy of its own review, but space limitations preclude addressing them here. In particular, the issue of periodicities observed throughout the magnetosphere is certainly relevant to the magnetospheric dynamics, but that topic has recently been thoroughly reviewed by Carbary and Mitchell . Since the cause of these periodicities is not yet agreed upon [Carbary and Mitchell, 2013], it is not yet clear how, why, or even if they are manifested in the processes we focus on in this review. A future review should certainly explore that linkage.
 The following discussion addresses two principal issues that arise from the dynamical imperative to shed plasma produced in the inner magnetosphere: First, how does plasma leave the source region? Second, how does that plasma leave the magnetosphere?
2 Inner Magnetosphere Transport: How Does Plasma Leave the Source Region?
 Cassini observations from the prime mission convincingly established that the principal process initiating outward transport of inner magnetospheric plasma is a centrifugally driven interchange instability [e.g., Mauk et al., 2009, and references therein]. This is a Rayleigh-Taylor-like instability that occurs when a “heavy” fluid overlies a “lighter” fluid. In Saturn's case, the heavy fluid consists of the dense, cold, water-dominated plasma produced in the inner magnetosphere, the light fluid consists of the hot, low-density outer magnetospheric plasma, and the role of gravity is played by the outward centrifugal force associated with Saturn's rapid rotation. Figure 1a illustrates schematically the structure that is thought to develop at the outer edge of the internal plasma source region, where the density gradient is unstable to interchange motions [see, e.g., Southwood and Kivelson, 1987]. Fingers of dense inner source plasma extend radially outward into the low-density, hotter outer magnetosphere, while similar channels of hot, tenuous material drive deep into the cold plasma source region.
 The most detectable dynamic features attributable to this instability are sudden, brief incursions of hot plasma into the inner magnetospheric cool plasma [e.g., Burch et al., 2005; Hill et al., 2005]. The hot plasma typically exhibits an energy dispersion, such that more-energetic ions are seen first, followed by lower energy ions, then lower energy electrons, and finally the more energetic electrons, a signature attributable to energy- and species-dependent gradient and curvature drifts (on top of the E × B drift associated with corotation) from an “injection” location some longitudinal distance away from the spacecraft. The amount of the energy dispersion is greater for “older” injections, i.e., those that occurred a larger distance away from the spacecraft. The degree of dispersion has been exploited to estimate the injection locations [e.g., Hill et al., 2005]. Interchange events with little dispersion are then understood to be “local” or “young”, i.e., they correspond to injections that occurred very recently, very near the longitude of the spacecraft [e.g., Burch et al., 2005].
 Interchange events also have detectable magnetic field signatures [e.g., André et al., 2005, 2007; Leisner et al., 2005] and distinct plasma wave signatures [e.g., Menietti et al., 2008; Rymer et al., 2009; Kennelly et al., 2013].
 An initial study of 48 selected interchange events showed them to be typically less than 10–20 h old and less than 1 Rs wide [Hill et al., 2005], important clues to the causative dynamic mechanism (c.f., Yang et al. , who used the RCM to study interchange at the outer edge of the Io torus at Jupiter). Using the observed energy dispersion to trace the injections back to their undispersed origin, Hill et al.  found no particular clustering in either local time or Saturn longitude (SLS, the Saturn Longitude System derived from observations of the observed periodicity of Saturn Kilometric Radiation (SKR) by Voyager [Carr et al., 1981]). A local time dependence and/or a longitude dependence could potentially link interchange onset to other periodically varying magnetospheric processes. In contrast to the earlier work, a subsequent, more comprehensive study of 429 events [Chen and Hill, 2008] found that the occurrence seems to peak between local times of ~4 and 12. Chen and Hill also found a possible dependence on SLS longitude, but perhaps surprisingly, no dependence on the SLS2 longitude, which is an updated longitude system based on the Cassini observations of SKR periodicities [Kurth et al., 2007] that has been found to organize other magnetospheric phenomena. The events are only found between L ~ 5 and L ~ 10 (L is the equatorial crossing distance of a dipole field line).
Chen and Hill  further found that the hot injected flux tubes only occupy about ~5–10% of the full 360° of longitude, demonstrating that the interchange occurs via rapid injection of narrow channels of hot, outer magnetospheric material separated by much broader, slower outflow of cold, inner magnetospheric material. Based on this observation, combined with a select set of plasma outflow measurements in the inner magnetosphere [Wilson et al., 2008], the inflow speed of the injected material was estimated to be up to 40 km/s near L ~ 9–10, declining to a few km/s at L < 7 [Chen et al., 2010]. By comparison, an in-depth study of a single interchange event observed at L ~ 7 yielded an estimated inflow speed of 71 km/s [Rymer et al., 2009, as corrected by Chen et al., 2010]. A similar analysis by Burch et al.  found an average inflow speed of ~25 km/s between 10 and 5 Rs.
 A more recent analysis of 249 very “young” injections, identified on the basis of their plasma wave signature [Kennelly et al., 2013], found a similar radial distribution to that found by Chen and Hill , but a significantly different local time distribution. The occurrence of such local events peaked near midnight, with a secondary peak in the afternoon. Further, the near-midnight events were found to be associated with SLS4 longitudes [Gurnett et al., 2011], the time-extended version of the SLSx (representing the various SLS versions) system, which separately tracks the periodicity of the northern and southern SKR sources. Curiously, the association had a seasonal dependence, with the pre-equinox occurrence rather strongly centered between ~100° and ~200° of SLS4 north longitude and the post-equinox occurrence peaking between ~100° and ~200° of SLS4 south longitude. No definitive longitude relationship could be identified for the post-noon set of events. The different longitude dependences derived by Chen and Hill and Kennelly et al. may be due to the fact that the Chen and Hill study did not simultaneously sort the events by local time or season, as Kennelly et al. did. It may also be due to uncertainties in establishing the longitude of onset for the older events, which requires back-tracing the dispersive signatures. However, it would be well worth additional studies to establish more firmly whether or not longitude (and which longitude system) actually influences the onset of the interchange instability.
 In addition to these studies of the occurrence statistics of interchange events, considerable progress has been made in delineating the more detailed properties of individual interchange injections. Comparison of hot electron phase space densities within injections with those further out in the magnetosphere provides strong evidence that the injections do indeed correspond to the near-adiabatic transport of outer magnetospheric hot plasma deeper into the magnetosphere [Rymer et al., 2009]. Modeling the observed pitch angle distributions within injections by adiabatic transport from an isotropic distribution at larger radial distances, Rymer et al.  concluded that the injected flux tubes appear to have traveled between ~3 and 5 Rs during the injection, arriving to be seen between L ~ 7 and 8.5.
DeJong et al.  observed that while very young injections appear empty of lower energy plasma [e.g., Burch et al., 2005], older ones show a very intense low-energy electron component. Spectrally similar but distinctly hotter than the ambient cool electron population of the inner magnetosphere, these electrons are strongly field aligned, leading DeJong et al.  to suggest that they are ionospheric in origin, drawn into the magnetosphere as part of the field-aligned current system associated with the interchange. The hotter temperature of this population causes enhancements in the average energy flux in the energy range of 12–100 eV, which DeJong et al. use as a marker for injections. They found that in the range of peak occurrence between 6 and 9 Rs, enhanced fluxes in this energy range seem to be organized in SLS3, with largest energy fluxes between about 335° and 90°. Subsequent work showed a day-night asymmetry in the energy fluxes of trapped (~90° pitch angle) fluxes in this energy range, with the nightside fluxes notably higher than those on the dayside [DeJong et al., 2011]. De Jong et al. attributed this asymmetry to an inhibition of interchange on the dayside. Curiously, the field-aligned population in this energy range, which was earlier shown to be the marker of interchange activity, showed less day/night asymmetry. Further, it should be noted that this low-energy population occurs in older interchange events, so the point of observation is not the same as the point of origin, and a more sophisticated back-tracing like that of Chen and Hill may be needed to draw conclusions about the local time distribution of events. Moreover, as discussed further below, there is now a different explanation for higher temperatures on the night side than on the dayside [e.g., Thomsen et al., 2012].
 Progress in observations of interchange events has been paralleled by progress in numerical modeling of the interchange instability. Multifluid simulations of the evolution of a system with a relatively dense torus of plasma centered on Enceladus' orbit [Kidder et al., 2009] showed the development of fingers of dense, cool, torus plasma moving out into the surrounding less dense, hotter, outer magnetosphere. The instability developed more quickly for higher mass loading, and it appeared to be suppressed if the interplanetary magnetic field in the simulation was parallel to the planetary field. To balance the cold, outflowing fingers, regions of hot, inflowing material formed between them. However, they did not appear to penetrate very deeply into the dense torus, and the outflow fingers were comparable in width to the inflow regions, unlike the observational conclusions of narrow inflow channels [e.g., Chen and Hill, 2008].
 Simulations using the Rice Convection Model [Liu et al., 2010; Liu and Hill, 2012] similarly show the development of outflowing fingers of cold, inner magnetospheric plasma, separated by fingers of inflowing empty flux tubes (illustrated in Figure 1c). For those simulations, a continuous source of inner magnetospheric plasma was assumed. In the initial stages of the instability, the inflow and outflow fingers had comparable azimuthal widths, but in the late stages, the cold material began to dominate the content of the outer magnetosphere, so the inflow channels covered only a small fraction of the longitude space, and inflow velocities were typically a factor of 10 higher than outflow velocities, based on arguments regarding the net magnetic flux balance in the inner magnetosphere. While the authors noted this similarity to the observations [Chen and Hill, 2008], it also seems possible that the late time preponderance of slow outflow regions in the simulation is due to the diminishing availability of empty flux in the outer magnetosphere, as the outflow continues to expand and fill more and more of the simulation space. Moreover, the early time results may be a better reflection of the actual process at Saturn, which is clearly characterized by sudden, rapid, discrete injections (i.e., “young”) as opposed to quasi-steady channels of inflow. Thus, while the numerical models have successfully reproduced many of the observed features of the interchange process at Saturn, they are not yet telling the whole story.
 In this discussion of interchange injections in Saturn's inner magnetosphere, it is important to distinguish explicitly between these and another class of “injections” that are observed in the same region of space, with some similar properties [see also Mitchell et al., 2009b]. In contrast to the relatively frequent, small-scale injections of hot plasma (<~50 keV) that are identified as interchange related, there exist less frequent (~once every few days) [Paranicas et al., 2007] particle injections that extend to hundreds of keV [e.g., Mauk et al., 2005]. These injections are most prominent in electrons and exhibit the same sort of energy-dependent dispersion seen at lower energies, but a single injection can persist for more than 20 h and can be seen repeatedly by Cassini as the injected particles drift around the planet [Paranicas et al., 2007]. Associated ion injections are also seen, but they tend to die out more quickly, presumably due to the effectiveness of charge exchange with ambient neutrals. They may be related to the ion intensifications seen as rotating energetic neutral atom (ENA) blobs when Cassini is at high latitudes [Carbary et al., 2008]. The latter are seen between ~5 and ~15 Rs, with peak occurrence near ~9 Rs, possibly because that is the ENA-producing sweet spot in the overlap between the neutral cloud density, which decreases with radial distance, and the injected ion flux, which is higher at larger L. Brandt et al.  have modeled the ENA observations of such an observed rotating region, which over the course of roughly 13 h from its initial injection near midnight, formed a distinct, energy-dependent spiral form. To successfully model the formation of the spiral, Brandt et al. used an initial injection of energetic ions between ~8 and ~16 Rs, extending over several hours of local time centered at midnight. This location is in accord with the observations of nightside ENA injections described by Mitchell et al.  and addressed further below. It is also in accord with the preferred source region of a set of 52 electron injections seen in the Magnetospheric Imaging Instrument (MIMI) energy range (20–200 keV) and analyzed by Müller et al. . Such less frequent, broader, and higher-energy particle injections thus appear to be associated with magnetotail dynamics as discussed more extensively below. It is unfortunate that the term “injection” has been used for both these and the smaller-scale, interchange-related events, as this has caused considerable confusion in the literature.
 In addition to the somewhat stochastic radial transport provided by the interchange instability, Saturn's inner magnetosphere (between at least 4 Rs and 10 Rs) also apparently features a global convection pattern, characterized by inflow toward Saturn on the duskside of the magnetosphere and outflow away from Saturn on the dawn side. This pattern was first identified in systematic offsets in the radial location of satellite absorption microsignatures in energetic particle fluxes [Roussos et al., 2005, 2007; Andriopoulou et al., 2012] but was also inferred from day/night asymmetries in plasma temperatures and energetic particle phase space densities [Thomsen et al., 2012]. It is also apparent in systematic radial velocities of ENA blobs associated with injected high-energy ions [Carbary et al., 2008]. Moreover, the corresponding flow velocities have now been directly measured with in situ plasma observations [Wilson et al., 2013]. Peak radial flow speeds are ~few to 10 km/s and do not appear to vary much with L. The flow pattern results in zero-energy drift paths that are ~1 Rs closer to Saturn on the nightside than on the dayside. It is not at this point clear what the cause of this circulation is, nor what its consequences might be for the overall transport of plasma out of the inner magnetosphere.
 As noted earlier, interchange injection channels only extend to L ~ 5. Inside that distance, the magnetosphere is interchange stable, and radial transport must proceed by other means. While there is some evidence [Gurnett et al., 2007] for a possible corotating convection pattern in the inmost magnetosphere (3 < L < 5) [e.g., Gurnett et al., 2007; Goldreich and Farmer, 2007], the cause and persistence of such a pattern are not known. Like the local-time-fixed pattern at larger radial distances, it is also not clear how it might affect the movement of plasma out of the source region. Moreover, inside of L ~ 5–6, recombination turns out to be a more effective loss process than transport [Sittler et al., 2008], so here we will not further address transport processes in this region.
3 Outer Magnetosphere Transport: How Does the Plasma Leave the Magnetosphere?
 Plasma transport from the well-established interchange region (L ~ 5–11) out to the final loss region at the magnetopause or down the tail is not well understood. The continued preponderance of corotational flow throughout this region [e.g., Kane et al., 2008; McAndrews et al., 2009; Thomsen et al., 2010, 2013] suggests that centrifugal forces may continue to be important. To remove the estimated 12–250 kg/s of ion mass produced in the inner magnetosphere, an average outflow velocity ~3–70 km/s of mass-loaded flux tubes would be needed beyond L ~ 15 [Bagenal and Delamere, 2011]. A limited analysis of energetic ion anisotropies near dawn and midnight found an average radial flow consistent with zero, but the calculation addressed only the velocity, not the mass flux [Kane et al., 2008]. Relatively high-density intervals studied in the post-midnight tail region [McAndrews et al., 2009] commonly exhibited an outward flow component with radial speeds typically exceeding ~20 km/s beyond L ~ 20 (H. J. McAndrews, unpublished data, 2009). In the pre-midnight region, data from Cassini orbits during 2010 show both inflows and outflows with radial speeds in the range of tens of km/s, although most of the flows are not strongly deviated from corotation, except at large L near the magnetopause [Thomsen et al., 2013]. Radial flow velocities through the rest of the middle-to-outer magnetosphere have not yet been analyzed. This is clearly an important topic for further study.
 By whatever means the plasma is transported outward through the middle/outer magnetosphere, it must ultimately be lost to the solar wind, either through the magnetopause or down the tail. At the Earth, an active process of magnetic reconnection occurs at the magnetopause, allowing the loss of magnetospheric plasma and the entry of magnetosheath plasma. However, at Saturn only a very few such in situ episodes have been observed [Huddleston et al., 1997; McAndrews et al., 2008; Lai et al., 2012; Badman et al., 2013], and reconnection there should be much less common than at the Earth because of the relatively high plasma beta at Saturn's dayside magnetopause [e.g., Scurry et al., 1994; Masters et al., 2012]. Nonetheless, observations of the dayside auroral oval suggest the occurrence of magnetopause reconnection that opens up a significant amount of Saturn's magnetic flux [Radioti et al., 2011], and the existence of the dark polar cap is itself suggestive evidence of open field lines [e.g., Badman et al., 2005; Belenkaya et al., 2008, 2011], although the existence of open field lines does not speak to the rate at which they are produced. In situ observations combined with nearly simultaneous auroral imaging have led to the conclusion that depending on the plasma beta and magnetic shear, reconnection can indeed occur at different locations on the magnetopause, and under some conditions, it can result in significant solar wind-driven flux transport in Saturn's outer magnetosphere [Badman et al., 2013; Desroche et al., 2013]. However, at this stage, a quantitative estimate of the rate at which plasma could be lost from the magnetosphere through dayside reconnection is still lacking. The present review focuses on the shedding of magnetospheric plasma, but we note that to the extent that magnetopause reconnection occurs at Saturn, there is a concomitant imperative to keep the magnetospheric magnetic flux in rough balance, a requirement that also has a bearing on magnetospheric dynamics, as will be mentioned below.
 Another possibility for loss through the magnetopause is via a Kelvin-Helmholtz (KH) instability [e.g., Delamere et al., 2013, and references therein]. In the nonlinear phase, the KH instability allows the entry of magnetosheath plasma into the magnetosphere, with the concomitant loss of magnetospheric plasma into the magnetosheath. The instability appears to be quite common at Saturn's magnetopause, especially in the afternoon [Delamere et al., 2013], so it is possible that any magnetospheric plasma that gets transported out to magnetopause distances can be shaved off into the magnetosheath through KH processes. Like magnetopause reconnection, the contribution of KH losses to the overall mass balance problem at Saturn has not yet been quantitatively assessed.
 The alternative to mass loss through the magnetopause is mass loss down the magnetospheric tail. As mentioned above, the large azimuthal flow velocities found in the outer magnetosphere imply that centrifugal forces continue to be substantial. On the dayside, these forces can be countered by the restraining dynamic pressure of the solar wind, but as flux tubes rotate onto the nightside, this confinement is removed, and flux tubes tend to stretch out radially. Eventually, the planetary field becomes too weak to confine the plasma, and it is released into the downtail direction. Vasyliunas  visualized this rotationally driven mass loss process as a global-scale pattern of magnetic tail reconnection, characterized by a neutral line extending across a portion of the tail. Plasma in the disconnected portion of a flux tube would flow away downtail on the tailward side of the neutral line, while the remaining flux tube content would flow back toward the planet on the sunward side of the neutral line. Figure 1b schematically illustrates this disconnection of the equatorial portion of mass-loaded flux tubes.
 The Vasyliunas process can also proceed via nonsteady and smaller-scale events: a stream of bubbles of plasma down the dusk magnetopause or the formation of plasmoids through something like a ballooning instability [e.g., Kivelson and Southwood, 2005]. Some evidence exists for the “planetary wind” along the dusk magnetopause [Thomsen et al., 2013], and MHD simulations exhibit similar features [Jia et al., 2012].
 Plasmoid formation and release has been seen in the tail in both magnetic field data [Jackman et al., 2007, 2009b, 2011, 2013] and plasma data [Hill et al., 2008]. In addition to the departing plasmoids, in situ signatures of dipolarization of reconnected flux returning to the inner magnetosphere have been reported in magnetic field data [Russell et al., 2008]. A large-scale plasma energization event observed by Cassini on the outbound leg of its initial orbit insertion was attributed to such a dipolarization caused by a major compression of Saturn's magnetosphere [Bunce et al., 2005]. Recently, a probable example of the plasma signature of a smaller dipolarization event in the pre-midnight tail has been identified [Thomsen et al., 2013].
 Plasmoid formation and dipolarizations are also identified by remote-sensing techniques. An ENA brightening observed Saturnward of the plasmoid described by Hill et al. was attributed to accelerated ions in the dipolarizing flux tubes inward of the reconnection region [Mitchell et al., 2005; Hill et al., 2008]. This very important observation thus connected plasmoid formation with the ENA brightenings related to energetic particle injections into the inner magnetosphere, as discussed above. Thus, the frequency and morphology of remotely sensed ENA brightenings, as well as the associated SKR bursts [e.g., Mitchell et al., 2005, 2009a] and energetic particle injections can be used to monitor magnetotail dynamics. A similar relationship between plasmoid magnetic field signatures and SKR bursts [Jackman et al., 2009b] leads to the same conclusion. One important insight from such studies [Mitchell et al., 2009a] is that there exist two classes of probable tail reconnection events: One is attributable to large-scale tail disruption associated with strong magnetospheric compression following the passage of interplanetary shocks [Bunce et al., 2005; Mitchell et al., 2005, 2009a], and the other is a less dramatic but much more common type that occurs roughly periodically with the SLS3 Saturn rotation period, apparently associated with the rotation of a particular longitude sector into a particular local time sector [Mitchell et al., 2009a]. Although Mitchell et al. do not report the favored longitude sector, their results show clearly that the midnight-to-dawn region is the favored local time sector.
 Global MHD simulations of Saturn's mass-loaded magnetosphere also show the occurrence of tail reconnection and plasmoid formation [e.g., Zieger et al., 2010; Jia et al., 2012]. The simulations of Zieger et al.  show a periodic pinching off of large-scale plasmoids that extend across a significant fraction of the tail. The period of release is inversely related to the solar wind dynamic pressure and also depends on the mass-loading rate in the inner magnetosphere. The recent work of Jia et al.  shows that the behavior of a newly formed plasmoid (and hence its in situ signature) depends on whether or not the reconnection, which starts on closed, stretched field lines, ultimately proceeds into open lobe field lines. If it does, the magnetic tension of the reconnected lobe field causes rapid tailward acceleration of the plasmoid, producing the classic plasmoid signatures [e.g., Jackman et al., 2007, 2011; Hill et al., 2008]. This situation is illustrated in Figure 1d, which shows a large plasmoid departing tailward under the magnetic tension of reconnected lobe field lines. But if the lobe does not become involved in the reconnection, the magnetic tension of the remaining overlying closed field lines restrains the plasmoid from departing downtail, and it moves with much the same velocity that it had prior to the reconnection onset (c.f., Figure 1b). This result yields the important insight that the in situ signature of a newly formed plasmoid may not be the classic signature of one that is escaping downtail [see also Thomsen et al., 2013; and Mitchell et al., submitted manuscript, 2013].
 While plasmoids are indeed observed at Saturn, the estimated frequency at which they are formed (0.4 per day based on the magnetic field signatures [Jackman et al., 2011] or 1–2 per day based on ENA injection events [Mitchell et al., 2009a]), combined with an estimate of their size and content, suggests that they are far (~1–2 orders of magnitude) from capable of accounting for the loss of plasma produced in the inner magnetosphere [Bagenal and Delamere, 2011]. A similar finding emerges from the MHD simulations [Zieger et al., 2010; Jia et al., 2012], which seem to show that a large fraction of planetary plasma may be lost through the magnetotail near the flanks. Bagenal  came to the same conclusion regarding plasmoids as a mechanism for mass loss at Jupiter. Thus, if downtail loss accounts for a significant fraction of what is produced in the inner magnetosphere, it appears that we need to seek other processes than large-scale plasmoid creation and loss. Possible evidence for a quasi-steady Vasyliunas-type reconnection region in the pre-midnight tail has recently been presented [Thomsen et al., 2013], which could potentially account for a large amount of plasma removal, but more work is needed to confirm or refute this suggestion.
 Finally, it is of interest to consider the possible interrelationship of the Dungey- and Vasyliunas-type convection patterns. Cowley et al.  proposed a global circulation pattern that combined the two types of cycles. In their model, Vasyliunas-type reconnection of closed flux tubes, heavily loaded with magnetospheric plasma, would occur primarily in the pre-midnight region, with reconnected flux tubes returning to the inner region through the post-midnight sector. Tailward of the region of returning flux would lie the Dungey-cycle x-line, feeding reconnected lobe flux tubes sunward along the dawn flank.
 The reason the Dungey-cycle reconnection of lobe field lines is restricted to the post-midnight magnetosphere in this picture is that the presence of the stretching, loaded, closed field lines in the dusk sector inhibits the reconnection of open lobe field lines. Not until Vasyliunas cycle, centrifugally driven reconnection releases some of the internally produced plasma downtail and the unloaded flux tubes begin to return toward Saturn can lobe reconnection occur (see also the discussion of this point by Jia et al.  and Thomsen et al. ). Thus, the occurrence of Vasyliunas reconnection (allowing mass loss from the system) enables Dungey-cycle reconnection (allowing return of magnetic flux previously opened through magnetopause reconnection).
 The simulations of Jia et al.  are consistent with the picture proposed by Cowley et al. , but so far, the Cassini observations show that if a large-scale, quasi-steady pattern of that sort exists at Saturn, the neutral lines (both V type and D type) must generally be located tailward of the region explored by the spacecraft. Jackman et al. [2009a] showed that even beyond ~40 Rs in the midnight region, the dominant magnetic field direction is southward (i.e., in the direction expected for closed field lines), with brief recurrent northward excursions associated with plasma sheet encounters that they attributed to large-scale wave-like motion of the current sheet. McAndrews et al.  found tailward flows in the post-midnight plasma sheet, but they were not associated with the northward field one would expect for the region downtail from a reconnection line (of either type). Moreover, the flows all contained a significant amount of water-group ions, marking the plasma as magnetospheric in origin, as opposed to the solar wind composition to be expected downstream of a Dungey neutral line (except within released plasmoids). In the pre-midnight region, the flows appear to be dominantly in the corotation direction, with little evidence of the outflows implicit in the Cowley et al.  picture, although so far only fairly early local times have been systematically surveyed [Thomsen et al., 2013].
 On the other hand, magnetic field observations do show evidence of lobe reconnection following plasmoid departure, in the form of the “post-plasmoid plasma sheet” (PPPS) [Jackman et al., 2011]. The PPPS is a prolonged interval of northward field after a plasmoid passes by, which Jackman et al. interpreted as a region of previously open (i.e., lobe) flux, now reconnected with both ends in the solar wind. Thus, the PPPS is taken to be the signature of the downtail exhaust from an ongoing reconnection process that produced the plasmoid and then proceeded to reconnect lobe field lines as well. Thus, whether or not there is a global pattern of the type envisioned by Cowley et al. , smaller-scale Vasyliunas-type reconnection that produces plasmoids in the tail can proceed to involve the open field of the lobes and ultimately allow the return of some of that flux to the inner magnetosphere. It may also be that strong solar wind pressure enhancements may compress the lobes and force the onset of large-scale Vasyliunas-type reconnection that then evolves, at least temporarily, into a large-scale Dungey cycle.
4 Outstanding Questions
 There has been considerable recent progress in understanding the dynamical processes that enable the transport and loss of plasma produced in Saturn's inner magnetosphere, but numerous important questions remain:
Inner Magnetosphere: Does the plasma production rate vary significantly over time? How and why does the depth of penetration of interchange injections vary? What controls the rate at which they occur? What controls their scale size? Does the injection speed vary? Is there or is there not a favored local time and/or longitude sector for the initiation of interchange, and what does that imply about the relationship to other periodic phenomena? By what process do fresh interchange injections subsequently get filled with field-aligned thermal electrons that are hotter than the ambient pickup plasma? Is there a corresponding ion signature? What is the source of the radial circulation pattern? What are the consequences (e.g., does it affect interchange instability)? What is the frequency of occurrence of discrete energetic particle injection events, and what does that tell us about outer magnetosphere dynamics? What is the best way to quantitatively describe the macroscopic convective transport (interchange, injections, and global radial circulation)? Does a diffusion approach make any sense in this region?
Outer magnetosphere: What is the dominant transport mechanism in the middle/outer magnetosphere (10–18 Rs)? Does interchange continue to operate? What other transport process might bring inner magnetosphere plasma out to potential loss regions? What is the source of the hot, low-density outer magnetospheric plasma (e.g., what is its composition)? Is there evidence for plasma of solar wind origin? How much magnetospheric mass can be lost through dayside reconnection? How much through a nonlinear Kelvin-Helmholtz instability? What evidence is there for quasi-stationary tail reconnection (either Dungey cycle or Vasyliunas cycle)? What do plasmoids look like if they do not involve lobe reconnection? Is there a range of plasmoid scale sizes? Can small-scale plasmoids “slip through” overlying magnetic field, allowing a sort of drizzle loss of plasma? What is the scale size of inflows and outflows in the tail, and what is the implied mass loss rate? And finally, given the very prominent, large-scale, periodic variations of magnetospheric configuration [e.g., Carbary and Mitchell, 2013], what role do these periodicities in magnetospheric behavior play in transport and loss?
 The Cassini mission continues to expand our understanding of the global production, transport, and loss of magnetospheric plasma and the dynamic processes that are involved. We are now beginning to glimpse the big picture of how these various processes relate to each other, but there remain important and elusive areas where new investigations will prove very fruitful.
 The author is grateful to Los Alamos National Laboratory for the support provided her as a guest scientist. This work was funded by the NASA Cassini program through JPL contract 1243218 with Southwest Research Institute. The Cassini project is managed by the Jet Propulsion Laboratory for NASA.
 The Editor thanks Christopher Arridge and an anonymous reviewer for their assistance in evaluating this paper.