the extended saturnian neutral cloud as revealed by global ena simulations using cassini/mimi...

15
JOURNAL OF GEOPHYSICAL RESEARCH: SPACE PHYSICS, VOL. 118, 3027–3041, doi:10.1002/jgra.50295, 2013 The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements K. Dialynas, 1,2 P. C. Brandt, 3 S. M. Krimigis, 1,3 D. G. Mitchell, 3 D. C. Hamilton, 4 N. Krupp, 5 and A. M. Rymer 3 Received 12 February 2013; revised 8 April 2013; accepted 23 April 2013; published 7 June 2013. [1] We show that the neutral gas vertical distribution at Saturn must be 3–4 times more extended than previously thought for the >5 R S regions, while the neutral H distribution is consistent with H densities that reach up to 150/cm 3 close to the orbit of Titan. We utilize a technique to retrieve the global neutral gas distribution in Saturn’s magnetosphere, using energetic ion and energetic neutral atom (ENA) measurements, obtained by the Magnetospheric Imaging Instrument (MIMI) onboard the Cassini spacecraft. Our ENA measurements are consistent with a neutral cloud that consists of H 2 O, OH, H, and O, while the overall shapes and densities numbers concerning the neutral gas distributions are constrained according to already existing models as well as recent observations. The neutral gas distribution at Saturn is determined by simulating a 24–55 keV hydrogen image of the Saturnian magnetosphere, measured by the Ion and Neutral Camera (INCA), averaged over the time period from 1 July 2004 to 23 August 2005. The ionic input of the model includes a proton distribution of combined Charge Energy Mass Spectrometer (CHEMS, 3–230 keV/e), Low Energy Magnetospheric Measurements System (LEMMS, 30.7 keV to 2.3 MeV), and INCA (5–300 keV) in situ measurements. These measurements cover several passes from 1 July 2004 to 10 April 2007, at various local times over the dipole L range 5 < L < 20 R S . A parameterized neutral gas distribution is changed until agreement between the simulated and average INCA image is obtained. Citation: Dialynas, K., P. C. Brandt, S. M. Krimigis, D. G. Mitchell, D. C. Hamilton, N. Krupp, and A. M. Rymer (2013), The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements, J. Geophys. Res. Space Physics, 118, 3027–3041, doi:10.1002/jgra.50295. 1. Introduction [2] Neutral gas distributions around giant planets are important indicators of the source, loss, and transport pro- cesses that redistribute material from the planet and its moons and rings, through interaction with the ambient plasma of the planetary magnetosphere. Energetic neutral atoms (ENAs) are produced by charge exchange between energetic ions and neutral gases resident in the magneto- sphere and can be imaged by remote sensing systems on board spacecraft, to produce a picture of the magnetosphere 1 Office of Space Research and Technology, Academy of Athens, Athens, Greece. 2 Department of Astrophysics, Astronomy and Mechanics, Faculty of Physics, National and Kapodistrian University of Athens, Athens, Greece. 3 Applied Physics Laboratory, Johns Hopkins University, Laurel, Maryland, USA. 4 Department of Physics, University of Maryland at College Park, College Park, Maryland, USA. 5 Max-Planck-Institut für Sonnensystemforschung, Katlenburg-Lindau, Germany. Corresponding author: K. Dialynas, Office of Space Research and Tech- nology, Academy of Athens, Athens, Greece. ([email protected]) ©2013. American Geophysical Union. All Rights Reserved. 2169-9380/13/10.1002/jgra.50295 and in turn to provide a marker for plasma-neutral processes. This technique has been widely used at Earth to image the ring current and the plasma sheet remotely [Roelof et al., 1985; Brandt et al., 2002]. [3] Before the Voyager 1 encounter at Saturn, Kirsch et al. [1981] using Low Energy Charged Particle (LECP) data and in order to explain the shape of the spectrum, attributed the escaping radiation from the vicinity of Saturn to either soft X-rays or energetic neutrals emanating from the planet. In their study, the ENA emissions were favored in contrast to X-rays, thus providing the first evidence of charge exchange at Saturn between trapped energetic ions with the planet’s neutral cloud, which was believed to consist of neutral hydrogen at that time. [4] More than two decades later, the Ion Neutral Camera (INCA) on board the Cassini spacecraft [Krimigis et al., 2004] provided mass-discriminated global observations of Saturn’s plasma environment [Krimigis et al., 2005; Mitchell et al., 2005], revealing an extensively large magnetosphere and energetic ion populations that essentially corotate with the planet [Paranicas et al., 2005]. Compositional analy- ses at both the thermal [Young et al., 2005], using Cassini Plasma Spectrometer (CAPS) measurements, and suprather- mal [Krimigis et al., 2005] energy range with Charge Energy Mass Spectrometer (CHEMS) in situ observations showed 3027

Upload: a-m

Post on 07-Feb-2017

213 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

JOURNAL OF GEOPHYSICAL RESEARCH: SPACE PHYSICS, VOL. 118, 3027–3041, doi:10.1002/jgra.50295, 2013

The extended Saturnian neutral cloud as revealed by global ENAsimulations using Cassini/MIMI measurementsK. Dialynas,1,2 P. C. Brandt,3 S. M. Krimigis,1,3 D. G. Mitchell,3 D. C. Hamilton,4N. Krupp,5 and A. M. Rymer3

Received 12 February 2013; revised 8 April 2013; accepted 23 April 2013; published 7 June 2013.

[1] We show that the neutral gas vertical distribution at Saturn must be �3–4 times moreextended than previously thought for the >5 RS regions, while the neutral H distribution isconsistent with H densities that reach up to�150/cm3 close to the orbit of Titan. We utilizea technique to retrieve the global neutral gas distribution in Saturn’s magnetosphere,using energetic ion and energetic neutral atom (ENA) measurements, obtained by theMagnetospheric Imaging Instrument (MIMI) onboard the Cassini spacecraft. Our ENAmeasurements are consistent with a neutral cloud that consists of H2O, OH, H, and O,while the overall shapes and densities numbers concerning the neutral gas distributionsare constrained according to already existing models as well as recent observations. Theneutral gas distribution at Saturn is determined by simulating a 24–55 keV hydrogenimage of the Saturnian magnetosphere, measured by the Ion and Neutral Camera (INCA),averaged over the time period from 1 July 2004 to 23 August 2005. The ionic input of themodel includes a proton distribution of combined Charge Energy Mass Spectrometer(CHEMS, 3–230 keV/e), Low Energy Magnetospheric Measurements System (LEMMS,30.7 keV to 2.3 MeV), and INCA (5–300 keV) in situ measurements. Thesemeasurements cover several passes from 1 July 2004 to 10 April 2007, at various localtimes over the dipole L range 5 < L < 20 RS. A parameterized neutral gas distribution ischanged until agreement between the simulated and average INCA image is obtained.Citation: Dialynas, K., P. C. Brandt, S. M. Krimigis, D. G. Mitchell, D. C. Hamilton, N. Krupp, and A. M. Rymer (2013), Theextended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements, J. Geophys. Res.Space Physics, 118, 3027–3041, doi:10.1002/jgra.50295.

1. Introduction[2] Neutral gas distributions around giant planets are

important indicators of the source, loss, and transport pro-cesses that redistribute material from the planet and itsmoons and rings, through interaction with the ambientplasma of the planetary magnetosphere. Energetic neutralatoms (ENAs) are produced by charge exchange betweenenergetic ions and neutral gases resident in the magneto-sphere and can be imaged by remote sensing systems onboard spacecraft, to produce a picture of the magnetosphere

1Office of Space Research and Technology, Academy of Athens,Athens, Greece.

2Department of Astrophysics, Astronomy and Mechanics, Faculty ofPhysics, National and Kapodistrian University of Athens, Athens, Greece.

3Applied Physics Laboratory, Johns Hopkins University, Laurel,Maryland, USA.

4Department of Physics, University of Maryland at College Park,College Park, Maryland, USA.

5Max-Planck-Institut für Sonnensystemforschung, Katlenburg-Lindau,Germany.

Corresponding author: K. Dialynas, Office of Space Research and Tech-nology, Academy of Athens, Athens, Greece. ([email protected])

©2013. American Geophysical Union. All Rights Reserved.2169-9380/13/10.1002/jgra.50295

and in turn to provide a marker for plasma-neutral processes.This technique has been widely used at Earth to image thering current and the plasma sheet remotely [Roelof et al.,1985; Brandt et al., 2002].

[3] Before the Voyager 1 encounter at Saturn, Kirsch et al.[1981] using Low Energy Charged Particle (LECP) data andin order to explain the shape of the spectrum, attributed theescaping radiation from the vicinity of Saturn to either softX-rays or energetic neutrals emanating from the planet. Intheir study, the ENA emissions were favored in contrast toX-rays, thus providing the first evidence of charge exchangeat Saturn between trapped energetic ions with the planet’sneutral cloud, which was believed to consist of neutralhydrogen at that time.

[4] More than two decades later, the Ion Neutral Camera(INCA) on board the Cassini spacecraft [Krimigis et al.,2004] provided mass-discriminated global observations ofSaturn’s plasma environment [Krimigis et al., 2005; Mitchellet al., 2005], revealing an extensively large magnetosphereand energetic ion populations that essentially corotate withthe planet [Paranicas et al., 2005]. Compositional analy-ses at both the thermal [Young et al., 2005], using CassiniPlasma Spectrometer (CAPS) measurements, and suprather-mal [Krimigis et al., 2005] energy range with Charge EnergyMass Spectrometer (CHEMS) in situ observations showed

3027

Page 2: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

that the most numerous species are H+. Additionaly, thesignificant presence of water group ions (O+, H2O+, OH+),which become predominant in the inner parts of the magne-tosphere, indicate that their sources are more likely to be theplanetary rings and satellites than the Kronian atmosphere.

[5] Several observations using the Ultraviolet Spectrom-eter (UVS) instrument onboard Voyager [Shemansky andHall, 1992] and also by the Hubble Space Telescope (HST)[Shemansky et al., 1993; Ip, 1997] showed that water groupneutrals exist in the magnetosphere of Saturn. Moreover,although those neutrals dominate over the ion populations,the gas cloud distribution is not very well understood[Esposito et al., 2005]. Subsequent observations by theCassini spacecraft confirmed that Enceladus is in fact anefficient source of water group neutrals that emanate fromits polar plumes [Dougherty et al., 2006; Hansen et al.,2006; Porco et al., 2006], a fact that makes Enceladus theprimary source of plasma in the magnetosphere. The massoutflow rate from Enceladus was estimated to be �300 kg/s[Jurac and Richardson, 2005], while a more recent esti-mation of this rate is �200 kg/s (with a 15% standarddeviation), as shown by Hansen et al. [2011]. However,recent analyses concerning the neutral gas cloud at Saturnindicate that at the inner to middle part of the magneto-sphere, other neutral sources than Enceladus are also activein this region. For example, Shemansky et al. [2009] hasrecently shown that the density of neutral H at the rings issignificantly high, 104/cm3, and originates from Saturn’s topatmospheric layers. The ring system and other moons, otherthan Enceladus, were characterized as minor sources [Juracand Richardson, 2007; Jurac et al., 2001a, 2001b].

[6] The investigation of the structure and compositionof the neutral gas cloud is critical, as it contributes tothe modulation of the Saturnian plasma environment. Thesignificantly dense neutral cloud at Saturn that fills upthe whole planetary system can, in fact, drive part of themagnetospheric dynamics or even dominate the dynami-cal processes at Saturn as noted by previous authors [e.g.,Kivelson, 2006]. In the present study, we use all avail-able Cassini/Magnetospheric Imaging Instrument (MIMI) insitu measurements to produce energetic proton and singlyionized oxygen equatorial distributions for various energyranges (24–55 keV, 55–90 keV, and 90–149 keV for H+ and90–170 keV for O+) and local times over the dipole L rangeof 5 < L < 20 RS. Furthermore, by utilizing all availableINCA images in the time period 183/2004 to 235/2005 andselecting those times during which the INCA imager waslooking at Saturn’s magnetosphere from the same vantageposition, we were able to produce a 24–55 keV H averageimage of the magnetosphere that corresponds to�2.7 Saturnrotations (total integration time of 29.5 h). By combiningthe ion and ENA information, we obtain the neutral gas dis-tribution around Saturn after simulating the aforementionedaverage H image.

[7] The ENA images presented in this study showenhanced ENA emissions at both the dayside and night-side magnetosphere that reach just below Titan’s orbit(�20.3 RS), and consequently, an extensive neutral cloudaround the planet is revealed. Furthermore, the ion equato-rial distributions exhibit local time asymmetries (day-nightas well as dusk-dawn) that can be explained by the multipleinjection events in conjunction to the azimuthal flow

properties of the plasma populations inside the Saturnianmagnetosphere. A forward modeling approach is then incor-porated in order to simulate the aforementioned observedENA image of the magnetosphere which leads in retrievingthe global neutral gas distribution around Saturn.

[8] Principal results of this study are consistent with aneutral cloud that is primarily formed at Enceladus andconsists of H, O, H2O, and OH. Our results are in gen-eral agreement with the Richardson et al. [1998] andJurac and Richardson [2005] models as well as the recentresults by Melin et al. [2009], but a �3 to 4 times moreextended neutral cloud in the vertical direction than previ-ously thought is needed in order to explain the observedENA intensities which originate from the vicinity of theplanet. Taking into consideration recent observations pre-sented by Shemansky et al. [2009], where the neutral Hdistribution shows an excess in densities (up to �150/cm3)close to the orbit of Titan, the simulations are significantlyimproved.

[9] The paper is organized as follows: In section 2, wedescribe the instrumentation and data sets used in this study.In section 3, we provide information on the ion spectral anal-yses that are performed in order to obtain ion equatorialdistributions in several energy ranges in terms of integratedintensities. These ion distributions are then simulated usinga parametric ion model, while their shapes and general prop-erties are discussed. Section 4 presents all available ENAimages of the Saturnian magnetosphere obtained by theINCA instrument from a particular vantage position for anextended imaging time period, while the properties of theseobserved ENA distributions are discussed. The model con-cerning the neutral gas cloud around Saturn, together withdetails on the resulting neutral density distributions afterthe ENA simulations, are thoroughly explained in section 5.Section 6 starts with summarizing the principal results of thisstudy and is then divided into two subsections that provideinformation on the model limitations (the effects of mag-netospheric dynamics on the simulations performed) anda general discussion of our results concerning the neutralgas distributions in accordance to relevant studies found inthe literature.

2. Instrumentation and Data Set Information[10] We have produced H+ and O+ equatorial distribu-

tions using all available Cassini/MIMI data for the timeperiod from the SOI (DOY 183/2004) to DOY 100/2007.The in situ ion data used in this study are a combination ofCHEMS 3 keV/e to 226 keV/e channels for H+ and 9 keV/eto 226 keV/e for O+, LEMMS A0–A7 channels (assumed tobe mostly counting protons) that cover the energy range of30.7 keV to 2.3 MeV, and INCA time-of-flight (TOF) energychannels which occupy the 5.2 keV to 360 keV energy rangefor H+ and 39 keV to 677 keV for O+, as explained byDialynas et al. [2009].

[11] Figure 1 shows the ion data coverage plot withrespect to local time (LT) and L-shell. Each point corre-sponds to a 30 min accumulation time, combined CHEMS-LEMMS-INCA spectrum (for H+), or a CHEMS-INCAspectrum (for O+), and the resulting spectra (O+ spectra notshown here) present the same shape and properties as dis-cussed by Dialynas et al. [2009]. The data set was binned

3028

Page 3: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Figure 1. Ion data coverage plot in local time and L-shellthat corresponds to the time period from the SOI (DOY183/2004) to DOY 100/2007.

into a grid with dimensions of 1 RS in dipole L and 30ı inLT (2 h). We have used all available information (ion spectrafound at each defined bin) to produce one average combinedCHEMS-LEMMS-INCA ion spectrum at each different bin(e.g., Figure 2, column c).

[12] During the period 265/2005 to 182/2006, the Cassinispacecraft traveled Saturn’s magnetosphere very close tothe nominal magnetic equatorial plane, while orbits duringlate 2006 to 2007 sampled higher latitudes up to � ˙50ı.These high-latitude orbits mapped mostly the dayside mag-netosphere of Saturn (LT 12:00 to 18:00). According toKrimigis et al. [2007] and Sergis et al. [2007], the daysideplasma sheet extends up to ˙45ı latitude, while the night-side plasma sheet is less extended and slightly elevated fromthe equatorial plane. We should note that this northwarddisplacement of the plasmasheet on the nightside is causedby the geometry of the solar wind flow with respect to theplanet’s magnetospheric tilt for the time period examinedand as a consequence is a seasonal effect and not a perma-nent characteristic of Saturn’s magnetosphere, as thoroughlyexplained by Arridge et al. [2008]. Thus, in order to producethe equatorial ion distributions (described in section 3), boththe equatorial and high-latitude orbits were taken into con-sideration for the dayside magnetosphere, up to ˙45ı, andalso orbits within the range of ˙10ı on the nightside.

[13] Energetic neutral atom (ENA) images were obtainedby the INCA camera onboard Cassini spacecraft [Krimigiset al., 2004]. The INCA sensor is a large geometry factor(G�0.6 cm2 sr for H) camera that analyzes separately thecomposition (H and O) and direction of the incident ENA.The detection of the ENA is based on the TOF technique,and several discrete energy passbands are defined within theenergy range of �5.4 to >220 keV. INCA has a wide field ofview (FOV) of 90ı in the nominal spin direction (azimuthaldirection) and 120ı in the direction perpendicular to the spinplane (elevational direction), while presenting a good angu-lar resolution of 8ı�4ı for hydrogen at energies greater than�50 keV.

[14] We have analyzed all available hydrogen images inthe 24–55 keV energy range, where INCA was looking

at Saturn, imaging the entire magnetosphere from approx-imately the same vantage position (–10 < XSZS < –3 RS,–31 < YSZS < –28 RS, –3 < ZSZS < –2 RS) during Cassini’snominal mission from the SOI (DOY 183/2004) to year2008, and from these images we have produced an averageimage of the magnetosphere. This procedure revealed 440approximately 4 min accumulation time hydrogen images,where the spacecraft was located at the aforementioned XSZS,YSZS, ZSZS location ranges, at the Saturn Equatorial Sys-tem (SZS), in which the X axis points roughly toward theSun, Z axis is parallel to the planetary rotation axis, andY axis completes the right-hand system. Although the timeperiod examined is, as already noted, quite extended (SOI toDOY 200/2008), the orbits that fulfilled the aforementionedrequirements refer to the time period from the SOI to late2005 (DOY 235/2005).

[15] As the spacecraft rotates about its spin axis, or ingeneral changes its orientation, the INCA FOV changesalso. This results in a variety of FOV orientations for theaforementioned 4 min images, which correspond to differentpixel orientations and prohibit averaging the images pixelby pixel. Although consecutive images usually refer to thesame FOV orientation and can be averaged in a routine man-ner (pixel by pixel), this cannot be applied for all imagessampled in different times, and furthermore, since we aimfor an average image of the entire magnetosphere, we haveconverted all individual images to the same frame by usingmotion compensation. As a consequence, the initial set ofimages was combined into 16 approximately 2 h accumu-lation time images which will be shown and discussed insection 4 of the present study.

3. Ion Equatorial Distributions at Saturn[16] Figure 2a illustrates the equatorial H+ and O+ distri-

butions using data obtained by the MIMI suite. All spectrawere numerically integrated over the energy ranges of inter-est to obtain the ion equatorial distributions in terms ofintegrated intensities. These energy ranges should corre-spond to INCA energy passbands for the ENAs. There-fore, the ion equatorial distributions are produced in the24–55 keV, 55–90 keV, and 90–149 keV ranges for H+

and 90–170 keV range for O+. Having the ion distributionsin integrated intensities means that we do not need a fullenergy-dependent charge-exchange cross section distribu-tion for completing the simulations which will be explainedin section 5. The cross sections which correspond to the geo-metrical mean of each channel can be used instead. As canbe seen in the McEntire and Mitchell [1989] publication, thecross sections for all species present a rather slow decreasewith energy up to �100 keV, while a significant dropoutin the cross sections data occurs for energies greater than�100 keV. For example, in the 24–55 keV energy range,the cross sections for protons impacting on neutral oxygenrange from �3.3 � 10–16/cm2 to �1.2 � 10–16/cm2, whilefor 36 keV (geometrical mean of the 24–55 keV channel),the equivalent cross section is �2.2 � 10–16/cm2.

[17] The ion distributions illustrated in Figure 2a (for bothspecies at all energy ranges) show a day-night asymmetry,with the nightside being more enhanced in terms of inte-grated intensities, as well as a dusk-dawn asymmetry. Asthoroughly explained in the literature [e.g., Nakamura et al.,

3029

Page 4: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Figure 2. (a) Ion equatorial distributions by using �3 years of in situ combined CHEMS-LEMMS andINCA measurements in the region of 5< L < 20 RS. The dashed circles denote the L-shells shown fromthe center of Saturn per 2 RS. The sun is to the left, and the local times are indicated. The 24–55 keV H+

distribution was used for the simulations shown in Figures 4 and 7. (b) The equivalent simulationson the ion distributions as described in the text. (c) Combined CHEMS-LEMMS-INCA H+ spectra(in differential intensities) over selected magnetospheric regions. For 36 keV protons, the maximum fluxranges from �200 to �500/(cm2 sr s keV), while the slope of the 24 to 55 keV H+ distribution is �2.5.

2002; Nosé et al., 2010; Mauk et al., 2005; Hill et al.,2005; Mitchell et al., 2009], dynamic planetward injectionsof energetic particles over anti-sun magnetospheric regionsare very common within the space environments. Theseinjections, particle energization, are usually driven by therelease of magnetic energy stored in the tail in the formof stretched field lines that relax planetward. Mauk et al.[2005], using a combination of CHEMS and LEMMS mea-surements, showed that the nightside inner magnetosphereof Saturn could well be dominated by a series of injections,a fact that possibly explains the enhanced ion intensitiesin this region as seen in Figure 2a. Large-scale energeticion injections have been also observed by the INCA imageron board Cassini, as shown by Mitchell et al. [2005] andCarbary et al. [2008]. Moreover, these observations suggestthat injections within the nightside magnetosphere of Saturnoccur preferably in the post-midnight sector and extend from

the orbit of Rhea (�8.7 RS) out beyond the orbit of Titan(�20.3 RS).

[18] This nightside enhancement in the ion intensities(Figure 2a) is likely to be the result of a continuousheating/acceleration process associated with the transition inthe current sheet from a broad region on the dayside to athin region on the nightside as the distributions of particlesrotate around the planet. Moreover, the injected ions aretransported planetward for several planetary radii over atime period that is rapid when compared to the planetaryrotational period. As a consequence, the enhancement ofparticles is possible to have secondary dependence on theradial transport velocities, not in a sense of an accumula-tion of particles but rather a dynamic heating (dusk) andsubsequent cooling (dawn).

[19] For example, the radial propagation speed of aninjection lies within the range of �100–200 km/s [Rymer

3030

Page 5: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

et al., 2009] for magnetospheric regions <11 RS, while thecorotation speed is significantly lower. Ideally, this radialtransport of the particles preserves their adiabatic invari-ants of magnetic moment and bounce, but as shown byDialynas et al. [2009] this is more likely to be the casefor protons rather than singly ionized oxygen. On the otherhand, inward transport of particles has been also recentlyobserved in the electron data using the LEMMS sensor byParanicas et al. [2010], who modeled the injections showingthat the energization process is consistent with the con-servation of these two adiabatic invariants for the case ofenergetic electrons.

[20] Apart from the day-night asymmetry seen inFigure 2a, a slight dusk-dawn asymmetry is also observed atthese equatorial ion distributions. The rapid rotation of Sat-urn and the generally weak solar wind conditions near theplanet result in a large region (�20 RS) where the drift ofions due to the corotation electric field dominates the trans-port of particles even up to 100 keV [Brandt et al., 2008].The injected ion distributions drift around Saturn subjectto corotation and gradient curvature of the magnetic fieldlines [Thomsen and Van Allen, 1980]. Consequently, whilecorotation as well as these drift drivers are in the samedirection [Paranicas et al., 2005], this will result in drivingthe freshly nightside injected particles to rotate around theplanet from the pre-dawn sector while experiencing charge-exchange decay at subsequent times. This mechanism possi-bly explains the aforementioned dusk-dawn asymmetry seenin Figure 2a. On the other hand, recent MHD simulations[e.g., Jia et al., 2012] have shown that this dusk-dawn asym-metry could be a result of the large azimuthal componentin the tail, so that after a reconnection event occurs, theplasma, associated flow channels, and dipolarization frontspropagate toward the dawn sectors.

[21] We do not see the spiral patterns expected from driftdispersion [Brandt et al., 2008; Mauk et al., 2005] becausethese spirals drift around in magnetic local time (MLT), andwe present average distributions. Although recent observa-tions have shown that the azimuthal velocity of cold plasmatends to lag corotation at Saturn’s magnetosphere [Sauret al., 2004; Kane et al., 2008], the drift of the injected par-ticles is dominated by the corotation electric field and thegradient and curvature drift inside 20 RS. The latter causesmore energetic ions to drift faster than corotation (while theopposite is valid for electrons, i.e., energetic electrons driftagainst corotation) [Mitchell et al., 2009]. Moreover, as thegradient and curvature drift speed increases with energy andL-shell, the azimuthal flow patterns will eventually presentenergy dispersed signatures, resulting in a spiral spatial pat-tern for the energetic ion distributions [Brandt et al., 2008]which we do not see in our equatorial distributions, sincethese describe an average situation which refers to �3 yearsof observations.

[22] In order to simulate Saturn’s ion (H+ and O+) equa-torial distributions (Figure 2b), we have used a modifiedversion of the parametric model presented by Roelof andSkinner [2000] (also used in studies by Brandt et al. [2008,2010]). In this model, the ion flux is defined in the equato-rial plane with separable functions in azimuthal angle (LT),L-shell, and pitch angle. For the azimuthal dependence, themodel uses a two-harmonic expansion that allows us to mod-ulate the ion flux in local time so that we can obtain a region

of maximum flux, i.e., a day-to-night asymmetry and, at thesame time using the second term, a dusk-to-dawn asym-metry, as seen in Figure 2a. Although the model providesthe framework for a separable function for the ion pitchangle distribution (PAD), we have assumed isotropic PADsfor both species in the present study. Figure 2b shows theresulting simulated equatorial ion distributions for differ-ent energy ranges, produced by the use of our parametricion model.

4. INCA Hydrogen Image Observations[23] The time period from the SOI (DOY 183/2004) to

2007 provided a unique opportunity for the INCA imager toobtain “edge-on” images of the entire Saturnian magneto-sphere, thus revealing the vertical extent of Saturn’s neutralcloud (the determination of which is one of the primarygoals of the present study). The ENA imaging is a techniquesensitive to very low neutral densities that cannot be mea-sured by in situ techniques, which was also demonstrated inBrandt et al. [2012] for Titan’s exosphere. When imagingSaturn’s magnetosphere, we are able to indirectly observethe effects of neutral H (detecting neutral densities in theorder of several 10/cm3) at radial distances that reach outbeyond 20 RS.

[24] Figure 3 illustrates a set of 16 approximately 2 haccumulation time hydrogen images in the 24–55 keVrange that resulted after motion compensation (explainedin section 2) was applied. At this point, we should notethat in general, for a given energy, hydrogen ENAs typi-cally scatter less in the instrument’s front foil than oxygenENAs, so hydrogen images present considerably better res-olution than the equivalent oxygen ones [Paranicas et al.,2005]. Thus, for the purpose of this study, we have usedthe 24–55 keV hydrogen images which generally presentsufficient resolution (see section 2 of the present study forinstrument details).

[25] The images in Figure 3 have not been corrected forthe point spread function (PSF) resulting from scatteringin the front foil of INCA [Mauk et al., 2003]. The PSFis instead taken into account in the simulations describedin section 5. For 36 keV protons, the full width at halfmaximum (FWHM) of the PSF is �11ı in the elevationangle and �9ı in the azimuthal angle. Likewise, we havenot applied a Compton-Getting correction [Compton andGetting, 1935] to these images as we intend to apply thecorrection (as used by Forman [1970] and Paranicas et al.[2005]) to the ENA simulations. According to the Compton-Getting effect, the ENA intensities that are caused by ionsconvected with a plasma that flows toward INCA will behigher than in the case in which the plasma flows away fromthe detector.

[26] The H images in Figure 3 are shown in the SZSsystem while Cassini is located at the pre-dawn sectorslightly below the equatorial plane at a radial distance�30 RS from the planet, with X � –8.2 RS, Y � –28.9 RS,and Z � –2.7 RS. The solid circles around Saturn representits main rings and the orbit of Rhea (�8.7 RS), while theoutermost circle denotes the orbit of Titan which is drawnfor reference (�20.3 RS).

[27] These images show enhanced ENA emissions at boththe dayside and nightside magnetosphere with the dayside

3031

Page 6: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Figure 3. Observed 2 h accumulation time INCA H images in the 24–55 keV energy range over thetime period from the SOI to DOY 235/2005 at selected times where the Cassini spacecraft was lookingat Saturn from approximately the same vantage position. The images are defined at the SZS frame, withthe planet being at the center of the images. The outermost solid circle denotes the orbit of Titan.

being sometimes less extended. The general shape of theENA emissions seen in Figure 3 is consistent with the shapeof the coresponding ion equatorial distributions discussedin the previous section. As explained by Carbary et al.

[2008], these ENA enhancements refer to a rotating sourceof ENA emission, in addition to a steady state component.This steady state component is apparent in our ENA dataseen in Figure 3, while the techniques used in this study are

3032

Page 7: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Figure 3. (continued).

not sufficient to visualize the rotating one. The steady statecomponent seen in Figure 3 presents peak brightness in themidnight sector with a mean value of �60/(cm2 sr s). TheseENA emissions can be attributed to periodic injections thatare re-energized approximately every Saturn rotation that

then gradually drift and decay, a fact that was early realizedby Mitchell et al. [2005] and documented also by Carbaryet al. [2008], and as a consequence, the rotating componentof these ENA emissions present also peak brightness in themidnight sector.

3033

Page 8: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

[28] Note that, in principle, the enhanced nightside inten-sities from this vantage point are also what is expectedfrom the Compton-Getting effect, which will be taken intoaccount in the simulations below. However, INCA imagessampled from selected vantage positions from the dusk-side (not shown here), where the Compton-Getting effect isreversed, show enhanced ENA emissions at both the daysideand nightside magnetosphere. Consequently, as will be alsoexplained in section 5, the enhanced nightside ENA emis-sions shown in Figure 3 are not due to the Compton-Gettingeffect.

5. Model and Simulations[29] In the present study, we use a forward modeling

approach in order to simulate the ENA emissions fromSaturn’s magnetosphere. The ENAs are formed when singlyionized energetic particles undergo charge-exchange colli-sions with neutrals resident in the magnetosphere. Giventhat the ENA intensity is jENA(E), then the charge-exchangeprocess can be written as a line-of-sight integral over theion intensity of the i-species ji(E) and the cold neutralgas exospheric density for k-species nk(S), multiplied bythe charge-exchange cross section between i-ions and thek-atoms. Mathematically, this can be written as follows:

j iENA(E) =

Xk

�ik(E)Z

Sji(E)nk(S)dS (1)

[30] Here, we assume that the ion intensity distributionunderlying the average ENA images shown in section 4 iswell represented by the average 24–55 keV H+ distributionobtained statistically in section 3. Then, the approach is tofit a neutral gas model to the average ENA images whilekeeping the ion distribution fixed. In this way, we retrievethe neutral gas distribution from the average ENA images.The neutral gas distribution at Saturn is assumed to consist ofH, OH, H2O, and O in this study. Therefore, the ion-neutralreactions considered are as follows:

H+ + H– > HENA + H+ (2)

H+ + O– > HENA + O+ (3)

H+ + OH– > HENA + OH+ (4)

H+ + H2O– > HENA + H2O+ (5)

and by taking into account (2)–(5), equation (1) can bewritten as follows:

j HENA(E) = �H+–H(E)

ZS

jH+ (E)nH(S)dS

+ �H+–O(E)Z

SjH+ (E)nO(S)dS

+ �H+–OH(E)Z

SjH+ (E)nOH(S)dS

+ �H+–H2O(E)Z

SjH+ (E)nH2O(S)dS (6)

[31] Following the equations described, the free param-eters in our case are the neutral density distributions pereach species (nH, nO, nOH, nH2O), and furthermore, theseare determined by simulating an observed ENA image.Figure 4a shows the average 24–55 keV H INCA image of

Figure 4. (a) Hydrogen ENA image in the 24–55 keVenergy range, averaged over 448 images that correspondto �2.7 Saturn rotations. (b) Simulated ENA image usingthe 24–55 keV proton distribution shown in Figure 2a andthe neutral density distributions represented by the blacklines in Figures 5 and 6. (c) Azimuthal and elevational pixelaverages in �9 and �11ı respectively for the simulated andobserved INCA images. Error bars are smaller than the sizeof the points.

3034

Page 9: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Figure 5. Neutral gas densities radial profiles used forthe simulations shown in Figure 4 (using black lines) andFigure 7 (using red lines).

the magnetosphere which was used for comparison with oursimulated hydrogen image (Figure 4b). The image shown inFigure 4a is defined by averaging the set of 16 images shownin Figure 3, pixel by pixel, while the total time that thisimage corresponds to is �2.7 Saturn rotations. The geom-etry and axis system for this image is exactly the sameas explained for Figure 3. As is apparent from the image,enhanced ENA emissions exist just below the orbit of Titanon the nightside magnetosphere, while the vertical extent ofthe neutral cloud (along the Z axis) is quite significant.

[32] At this point, we need to be reminded that we applyboth the PSF and Compton-Getting corrections to the sim-ulated image and not the observed ENA images. The PSFthat is used in this study for �36 keV (geometrical mean ofthe 24–55 keV INCA H channel) is �11ı in both imagingdirections, due to the fact that the average image we are sim-ulating contains all possible FOV orientations (nominally,PSF is�11ı in the elevation angle and�9ı in the azimuthalangle, as noted in section 4). The Compton-Getting effectdepends on the azimuthal flow speed (VC) of the ion distribu-tions inside the magnetosphere together with the � exponent(defined as dlogj/dlogE, where j is the ion flux, adoptedfrom the statistical study of Dialynas et al. [2009]) [e.g.,Paranicas et al., 2005]. Although the azimuthal velocitiesof the magnetospheric particles tend to lag corotation by�30% [Kane et al., 2008], we have assumed that the ionazimuthal flow speeds are consistent with rigid corotationfor performing the simulations. This simplification results ina conservative overestimate of the Compton-Getting effectfor the 24–55 keV H+ particles that we use in this study, butthe bias introduced to our results is practically insignificant(as opposed to O+ particles). For example, the first-orderCompton-Getting at L = 10 RS for a 36 keV proton with V�2.6 � 103 km/s goes as V/VC � 3.8 � 10–2 (assuming rigidcorotation). By contrast, the azimuthal flow speed would bemuch more important for lower energy protons.

[33] Figure 4b shows the resulting simulated INCA Himage in the 24–55 keV energy range, obtained by keepingthe model ion distributions from section 3 fixed and vary-ing a parametric neutral gas distribution until the simulatedimages (Figure 4b) agree with the observed ones (Figure 4a).

The criterion for measuring the goodness of the fit canbe expressed by the well-known error function (chi-squarefunction) which measures the error between the model andthe measurements.

[34] Having a set of N measured values, i.e., N pixels(i = 1 : : :N), the fit can be defined using the normalized �2

parameter adopted from Press et al. [1992]:

�2 =1N

NXi=1

(si – ˛i)2

�2i

(7)

where si is the set of simulated pixel values, ˛i represents thecorresponding observed pixel values, and �i is each pixel’sstandard deviation which was calculated from each indi-vidual image that went into the averaging. The numeratoris dominated by Poisson fluctuations, which causes �2 toconverge to unity for a perfect fit [Roelof and Skinner, 2000].

[35] The observed and simulated images are in a verygood agreement in the vertical direction (along Z axis),but the simulated image is slightly less extended in thehorizontal direction. This is highlighted in Figure 4c, whichillustrates an average of pixels in the peak region of�11ı and �9ı at the elevational (horizontal) and azimuthal(vertical) directions, respectively, for the two images. Theoverall �2 is equal to 1.4, and, given the number of pixelsthat we simulate and the number of the free parameters thatare related to the simulation, this �2 number corresponds to afairly successful simulation. Furthermore, the almost perfectagreement between the simulation and the observation in thevertical direction meets well with the primary goal of thisstudy which is the determination of the extent of the neutralcloud, along the Z axis.

[36] The black lines in Figures 5 and 6 show the neutraldensity radial and meridional profiles, respectively, thatwhere used in order to obtain the simulated ENA image(Figure 4b). However, there are two major considerationsrelated to the neutral densities shapes and numbers thatresulted from this analysis:

[37] 1. The non-uniqueness of the solution, i.e., the neutralgas distributions per each species shown in Figures 5 and 6do not describe the only solution that provides the best fitbetween the simulated image (Figure 4b) and the observedone (Figure 4a). The neutral model that is used in this studyis a parametric one and not a physical model. Furthermore,none of the MIMI instruments is able to directly measure theneutral particle densities.

[38] 2. As shown by Paranicas et al. [2008], the energeticions are effectively absorbed inside �5 RS to �6 RS, so thatthis region is almost void of more energetic protons or singlyionized oxygen. Consequently, there are no ENA emissionsin this region that we can simulate.

[39] In order to introduce a physical interpretation to ourresults and confront the aforementioned difficulties, we con-strained our parametric neutral model in the regions <5 RSaccording to the already existing models of Richardson et al.[1998] and Jurac and Richardson [2005], as well as recentobservations by Melin et al. [2009]. Consequently, the dis-tributions shapes and numbers per each different neutralspecies shown in both Figures 5 and 6 (the resulting neutraldensity radial and meridional profiles) are not a result of thisanalysis but are constrained according to the aforementionedstudies results and/or measurements. Therefore, as noted by

3035

Page 10: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Figure 6. Contours of constant densities for the (a) OH,(b) H2O, (c) O, and (d) H neutral species, plotted in themeridional plane. The Richardson et al. [1998] OH contoursare also plotted for comparison. Black lines were used forthe simulation shown in Figure 4, and red lines were usedfor the simulation shown in Figure 7.

Richardson et al. [1998] and Jurac and Richardson [2005],the dominant species at Saturn are OH and H2O with peakdensities �1000/cm3 near Enceladus orbit, while all othersatellites produce smaller density peaks, a fact that makesthem minor neutral sources. Moreover, the densities of allspecies drop sharply below�2 RS, where Saturn’s main ringsystem is located.

[40] The neutral H cloud is quite extended (H is the lighterspecies among the neutrals used in this study) and becomespredominant for radial distances >14 RS. Although the sur-vival of neutral species is a very complicated issue as itinvolves several processes [Smith et al., 2010] (e.g., chargeexchange, electron impact ionization, and photoionization),in principle, the H atoms can in fact survive more effi-ciently inside the magnetosphere than other heavy neutralsused in this study (while H and O lifetimes are compa-rable) [Richardson et al., 1998]. The neutral O forms avery broad torus centered at �4 RS and falls off withradial distance much slower than OH. This is possibly aresult of greater lifetimes for O compared to OH, mainlyfor the <12 RS regions, while for >12 RS, this situationis reversed [Smith et al., 2010], combined with different

sources for the two species [Melin et al., 2009]. The deduceddensity distributions presented here (for the >5 RS mag-netospheric regions) are in very close agreement with theJurac and Richardson [2005] model and Melin et al. [2009]observations.

[41] Figure 6 (black lines) illustrates the distribution ofOH in the meridional plane as well as all other neutralspecies used in the present study, as derived from our anal-ysis. All species present similar signatures in the meridionalplane, apart from neutral hydrogen that is broader at bothZ and radial direction. A different set of constraints in ourneutral gas model concerns the total neutral atoms per eachspecies. We find that 42% of the total O population exists inthe region 10 to 25 RS which is consistent with the recentUltraviolet Imaging Spectrograph (UVIS) observations pre-sented by Melin et al. [2009] (40%). Assuming cylindricalsymmetry for the neutral cloud, we calculate that the totalO particles, according to our parametric neutral model, is�3.9 � 1034 (or equivalent to �1 � 1012 g) in the regionX = ˙10 RS and Z = ˙5 RS, which is consistent with Melinet al. [2009] (�3.1 � 1034 or equivalent �0.8 � 1012 g) andEsposito et al. [2005] (�4 � 1034 O atoms or equivalentto �1 � 1012 g) results. The total H content that resultedfrom our parametrical neutral model is �4.8 � 1034, whichis, again, in reasonable agreement with Melin et al. [2009]results (�9.6 � 1034).

[42] One of the principal findings of this study is illus-trated in Figure 6: Given the equatorial proton distributionwhich corresponds to the 24–55 keV INCA energy channel,we changed the neutral gas density parameters that controlthe vertical extent of the neutral cloud along the Z axis (thefree parameters) until the average INCA hydrogen imageand the simulated one were found in reasonable agreement(see Figure 4). This procedure revealed that the verticaldistribution of the neutral gas must be about �4 timesthat predicted by Richardson et al. [1998] and Jurac andRichardson [2005] for the regions beyond 5–6 RS.

[43] However, as noted before—and can be seen at rel-evant studies referenced throughout this paper—there areseveral uncertainties concerning the neutral gas profiles atSaturn. For example, H is detectable at the orbit of Titan(�20.3 RS) with peak densities that vary from 50 to 150/cm3

[Shemansky and Hall, 1992; Shemansky et al., 2009]. Whilekeeping our neutral gas distribution constrained in orderto produce our results (as explained before), we haveused the lower end of this range in our model (�50/cm3

at �20.3 RS).[44] Consequently, another principal finding of this study

concerns the neutral H distribution at Saturn: Using a neutralgas distribution such as the one found in Figures 5 and 6(red lines), which introduces an excess of H in the outer-most magnetospheric regions, the horizontal discrepanciesbetween the simulated and observed images (Figure 7) areeffectively minimized, while the overall fit is significantlyimproved (�2 = 1.1). Although the O, H2O, and OH radialprofiles are the same as shown in Figure 5, the “increasing”H density profile contributes in creating more ENA particlesat the outermost magnetosheric regions and, overall, formsa resulting simulated ENA image in close agreement to theobserved one. However, the vertical extent of the neutral gasdistributions (meridional plane) required for performing thesimulation is �3 RS.

3036

Page 11: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Figure 7. Same as Figure 4 using a neutral gas distributionshown with the red lines in Figures 5 and 6.

6. Summary and Discussion[45] By utilizing all available MIMI in situ and remote

observations during an extended time period (�3 years),we were able to produce integrated intensity energetic iondistributions for both H+ and O+ inside the Saturnian mag-netosphere, study several hydrogen ENA images, and fromthese images produce an average H image of the Saturniansystem. Moreover, the combined ion and ENA information,

together with recent neutral density measurements and mod-els found in the literature (as constraints), lead to the extrac-tion of the neutral density distributions needed in order tosimulate this average H image. Primary results of this studyare briefly summarized as follows:

[46] 1. The resulting neutral gas distribution at Saturn(after simulating an average ENA image) showed that theneutral species densities and profiles are consistent withrecent models and observations, but, for the >5–6 RS regions,a �3–4 times more extended neutral cloud in the Z direc-tion than previously thought [e.g., Richardson et al., 1998]is needed in order to explain the ENA intensities obtainedby INCA.

[47] 2. Further simulations showed that the neutral Hdistribution is consistent with H densities that reach up to�150/cm3 close to the orbit of Titan, as noted by Shemanskyet al. [2009].

[48] 3. The integrated intensity ion distributions showedprominent day-night as well as dusk-dawn asymmetries,which could be explained by the multiple injections thatoccur at Saturn, as well as the azimuthal energetic ion flowproperties inside the magnetosphere (namely, corotation,together with gradient and curvature drifts).

[49] 4. An average ENA image of the magnetosphere,defined by several consecutive images, revealed a steadystate source of ENA emissions that surrounds the planet in arelatively broad region at both the equatorial and meridionalplanes, with peak brightness in the midnight sector.

[50] These results, in accordance with relevant studiesfound in the literature, are discussed in the paragraphsthat follow.

6.1. Magnetospheric Dynamics Effectson the Simulations

[51] The dynamics at Saturn’s magnetosphere (e.g.,rotating blobs, multiple injections, and dispersed ion distri-butions over time) as discussed by previous studies [e.g.,Brandt et al., 2008; Carbary et al., 2008; Mitchell etal., 2005] can cause several variations on the ENA emis-sions as shown in Figure 3. When dealing with ENAs, wemust always remember that variable ENA emissions reflectall possible variations in their parent ion distributions thatthese ENAs originate from. In addition, temporal variationson the neutral gas distribution could have the same effect.For example, if the Enceladus source rate is indeed constant(e.g., �300 kg/s), then all possible alterations on the ENAemissions could be attributed to the ion dynamics at Saturnand not neutral gas temporal variations. On the other hand,Smith et al. [2010], using a combination of MIMI, CAPS,and INMS data, concluded that Enceladus source is vari-able with rates of �1027 to �1028 water molecules/s. As aconsequense, all variations in the ENA emissions could beattributed to the neutral gas variations together with the iondynamics at the planet, a fact that reveals the complexity andthe dynamical profile of the Saturnian system.

[52] Due to that, a longer exposure time for the averageimage would give a better estimate of the global neutralgas distribution at the planet. Ideally, a total time of morethan several tens of Saturn rotations would result in veryeffectively minimizing possible variations in the ENA emis-sions as detected by INCA. However, as we explained earlierin section 2, to produce this average image we have used

3037

Page 12: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

all available information during the Cassini nominal mis-sion around Saturn, and there exist no more INCA imagesfrom the same (or approximately the same) vantage position,to further improve the total exposure time that this imagecorresponds to.

[53] As explained earlier, despite the fact that the 2 h ENAimages found in Figure 3 show considerable variation thatcan be attributed to a rotating, asymmetric ring current, thefinal image used in the simulation (Figure 4a) illustrates asteady state component in the H ENA emissions on average.Furthermore, given the consistency between the presentedaverage ion distributions (Figure 2) and the average ENAimage and the fact that there appear to be no large-scale,dynamic, active events that can dominate the average ENAimage shown in Figure 4a during the time periods examined,this total exposure time (�2.7 Saturn rotations) is sufficientto practically eliminate the effects that all possible planetarydynamical processes at Saturn (rotating populations, etc.)can cause to the simulations. Given that we aim for theneutral gas distribution at the planet in a global scale, thisfact meets the purposes of this study. Furthermore, the cri-teria used for the comparison of the observed and simulatedimages (�2 fit) suggest that the simulations are successful.

6.2. Neutral Cloud at Saturn[54] Our method, i.e., simulation of remote ENA observa-

tions (H images) using measured ion distributions (in situ)toward obtaining a neutral cloud distribution, suffers fromseveral limitations which are explained earlier in this study.First, the inherent limitation of the model is that we sim-ulate two-dimensional ENA images of a three-dimensionalsource region. Second, we are not able to obtain the compo-sition of the neutral cloud by utilizing the INCA data alone(presented in this study). Instead, we can only demonstratethat the INCA measurements are consistent with a neutralcloud that consists of H, O, OH, and H2O. Third, while noenergetic ions are present in the <5–6 RS regions, chargeexchange cannot occur at these regions, and consequently,we are not able to deduce neutral cloud density distributionsfor these regions.

[55] Strictly speaking, the peak densities and profiles thatare presented through this analysis for all species in theregions <5–6 RS can be thought of as an extrapolation (moreaccurately, constrained as explained in section 5) of the den-sities needed to reproduce the observed ENA fluxes in theregion >5 RS, where charge exchange takes place. Further-more, while the neutral cloud compositional analysis cannotbe performed with our data and method, the overall shapesand densities numbers concerning the neutral distributions inthe >5 RS regions were also constrained whenever available,as explained in section 5.

[56] In the paragraphs that follow, we discuss the ques-tions that emerge by our two main results, i.e., the verticalextent of the neutral particle distributions at Saturn and theneutral H density excess in the middle magnetosphere. How-ever, we note that although we initially pose these questionsof what could be the cause of the extent of the neutral par-ticles distributions around the planet, the answers to thesequestions lie beyond the scope of this study. Here, we haveprovided the observations and related analyses for the globalgas cloud distribution in an average sense, and we will only

discuss the proposed mechanisms, as these are explained inthe literature.6.2.1. Neutral Particles Vertical Extent

[57] Recent observations and results found in the litera-ture [Esposito et al., 2005; Melin et al., 2009; Shemanskyet al., 2009; Perry et al., 2010] combined with results pre-sented in this study suggest that Enceladus is possibly thedominant source of neutrals at Saturn, while, as explainedby Shemansky et al. [2009], the Saturnian atmosphere seemsto be a strong source of atomic hydrogen at low-ish alti-tudes. Although the radial profiles of the neutral gas particlesused in this study are consistent with the neutral cloud stud-ies mentioned earlier, by changing the neutral gas densityparameters in our model, it was found that a �3–4 timesmore extended neutral cloud than previously thought in theZ direction is needed in order to explain the extended verti-cal ENA emissions at Saturn (for the >5–6 RS radial distanceregions).

[58] We should note, however, that although the overallneutral gas distribution at Saturn must be quite extended,as explained above, the true vertical distributions per eachspecies (separately) might be different from that shown inFigure 6. For example, due to the non-uniqueness of thesolution in our method (explained earlier), the hydrogenatoms could conceivably have much broader extension thanthose of the heavy neutral species and thus mask the truevertical distributions of the H2O, OH, and O atoms [e.g.,Ip, 1996]. Nevertheless, while both Esposito et al. [2005]and Melin et al. [2009] show, for example, O emissions thatextend out to 3 RS along the Z axis (but do not explain thisfeature explicitly), our results can be considered representa-tive of the true vertical distribution of the neutral gas cloudat Saturn.

[59] The question that emerges from this result is Wheredo neutrals find the additional energy needed for them toescape the equatorial plane at these high altitudes alongthe Z direction? By adding the energy gained by the dis-sociation of their “parent” neutrals (a few eV’s) to theirorbital energies, the total energy is perhaps insufficient tocreate that effect. For example, heavy products from pho-todissociation have excess energy of <2 eV, while H atomsfrom the dissociation of H2O and OH have �1.5–2.5 eV[Johnson et al., 2008]. Furthermore, the less time these neu-trals remain in the system, the least they spread in radiusand along the Z direction [Jurac and Richardson, 2005].As we have already pointed out, this might make somesense for hydrogen since the energy gained at its creationis sometimes comparable to the Keplerian energy. Since Hcharge-exchange cross sections are relatively small, somehydrogen particles will indeed escape the system, while oth-ers will perhaps remain in the system and form the observedbroad neutral cloud that is not confined to the equatorialplane. However, this cannot be the case for other neutralsas well.

[60] The answer to that question might be the multiplecharge-exchange process that occurs at Saturn together withneutral-neutral reactions as explained by Cassidy and John-son [2010]. Although the neutral particle energies might besmall at the time of their creation, charge exchange canrecreate neutral particles with energies on the order of sev-eral hundred eV’s, which is sufficient to distribute them athigher altitudes along Z axis. Of course, since the plasma

3038

Page 13: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

density peaks at the equatorial plane, the charge-exchangerate should be higher in that region. On the other hand, theextended plasma sheet as seen in the energetic ion mea-surements at Saturn [Sergis et al., 2007; Krimigis et al.,2007] could in fact provide an extensive region throughoutthe Saturnian magnetosphere in which multiple charge-exchange collisions can be important at high latitudesas well.

[61] As Paranicas et al. [2008] have shown, energeticions are efficiently absorbed inside 5 RS. Consequently,charge exchange between energetic ions and neutral speciesis most unlikely to occur near Enceladus (3.95 RS). How-ever, as shown by Young et al. [2005], a significant numberof corotation energy ions exist in this region and if theseisotropize to some degree, then when they charge exchangeto become neutral particles, they will have enough energyto form a broader distribution. Although most of these new-born neutrals will be so energetic they will leave the system,some portion will probably be in a phase of their gyrationthat their Keplerian energy is low enough to remain grav-itationally bound. Consequently, these will make a muchbroader distribution than the initial cold neutral cloud. Still,this mechanism cannot be a stand-alone explanation of thebroad neutral gas distribution at Saturn.

[62] In conjunction to charge exchange, a slower process(due to the small cross sections which indicate long par-ticle lifetimes), but significantly more efficient mechanismthat can provide the broader neutral particles distributions asreported by the present study, is the neutral-neutral reactions[Farmer, 2009; Cassidy and Johnson, 2010]. Although,initially, the neutral-neutral scattering was not thought toplay a significant role in the formation of the neutralgas cloud distribution (compared to charge exchange, forexample), it is now becoming the most probable can-didate for the extended neutral particle distributions asobserved by INCA (this study) and recent UVIS measure-ments [Esposito et al., 2005; Melin et al., 2009; Shemanskyet al., 2009]. Even if the initial neutral particle distribu-tion is relatively narrow when emanated from Enceladus, asexplained by Cassidy and Johnson [2010], the combinationof plasma processes together with neutral-neutral reactionscan indeed explain the observed neutral cloud extent. Thespreading of O and OH neutral distributions was shown byTadokoro et al. [2012], using a H2O model that includeda combination of charge-exchange and neutral-neutralcollisions.6.2.2. Hydrogen Excess

[63] Another question that is raised by our results isrelated to the H excess at the outermost magnetosphericregions (�20 RS). The reason for the H excess shown in thisstudy, also observed by Melin et al. [2009] as an asymmet-ric H torus on the nightside magnetosphere, is most probablyrelated to the solar radiation pressure on H, as explained bySmyth and Marcony [1993] and modeled by Ip [1996]. Dueto the gravitational potential of Saturn and the long H life-times, these particles are able to make a number of orbitsabout the planet (relative to Titan’s location), before theyare lost through ionization, charge exchange, etc. The accel-eration of solar radiation pressure is—indeed—relativelysmall compared to Saturn gravitational acceleration (0.25 %for solar mean conditions), only a perturbation to the nor-mal Keplerian motion of a H atom. But because of H long

lifetimes, the cumulative effects of the radiation pressurecan be substantial, as it acts on H atom over their longlifetime making the orbits more eccentric with a periapsisprogressively closer to Saturn on the dayside and apoapsisprogressively further from Saturn on the nightside. The netresult is a H torus such as Melin et al. [2009] observed.

[64] On the other hand, as shown by Brandt et al. [2012],Titan holds an extended H2 exosphere that is dominated bya satellite distribution above �10,000 km and extends outto �50,000 km altitude, which coincides with the radiusof its Hill sphere. Titan’s exosphere can—in principle—be a source of atomic H for the outermost magnetosphericregions, while the fate of the neutral species that composeTitan’s exosphere will depend upon several processes, suchas photodissociation, charge exchange, and electron impactionization [e.g., Westlake et al., 2012]. However, sincethere is no detectable torus around Titan beyond the Hillsphere [Sittler et al., 2010], the possibility of an extendedH escape from Titan (e.g., from the dissociation of H2)is much less likely to explain the results presented in thisstudy. We note that the hydrogen Deuterium Absorption Cell(HDAC, onboard Cassini) measurements of Titan’s hydro-gen Lyman alpha emission would yield a H escape rate of�1.7 � 1027 atoms/s, as a result of Jean’s escape [Hedeltet al., 2010].

[65] Nevertheless, due to the complexity in modelingneutral H (many factors act on H inside the magnetosphere,as noted previously), global results on neutral species pre-sented in this study are very important, providing an aid forfuture modelers for constraining global physical models.

[66] Acknowledgments. The authors would like to thank M. Kusterer(Johns Hopkins University Applied Physics Laboratory) for assistance withthe INCA data processing. We are grateful to C.P. Paranicas for usefuldiscussions concerning the Compton-Getting effect, T. H. Smith for use-ful discussions that concern the neutral particles physical mechanisms inthe Saturnian magnetosphere, and all colleagues on the MIMI team, whoprovided valuable comments that have improved the presentation. Workat JHU/APL was supported by NASA under contract NAS5 97271 andNNX07AJ69G and by subcontracts at the University of Maryland and theOffice of Space Research and Technology of the Academy of Athens.Gefördert von der Raumfahrt-Agentur des Deutschen Zentrums für Luft-und Raumfahrt e.V. mit Mitteln des Bundesministeriums für Wirtschaft undTechnologie aufgrund eines Beschlusses des Deutschen Bundestages unterdem Förderkennzeichen 50OH1101.

[67] Masaki Fujimoto thanks the reviewers for their assistance inevaluating this paper.

ReferencesArridge, C. S., K. K. Khurana, C. T. Russell, D. J. Southwood, N. Achilleos,

M. K. Dougherty, A. J. Coates, and H. K. Leinweber (2008), Warpingof Saturn’s magnetospheric and magnetotail current sheets, J. Geophys.Res., 113, A08217, doi:10.1029/2007JA012963.

Brandt, P. C., R. DeMajistre, E. C. Roelof, S. Ohtani, D. G. Mitchell,and S. Mende (2002), IMAGE/high-energy energetic neutral atom:Global energetic neutral atom imaging of the plasma sheet andring current during substorms, J. Geophys. Res., 107(A12), 1454,doi:10.1029/2002JA009307.

Brandt, P. C., C. P. Paranicas, J. F. Carbary, D. G. Mitchell, B. H. Mauk,and S. M. Krimigis (2008), Understanding the global evolution ofSaturn’s ring current, Geophys. Res. Lett., 35, L17101, doi:10.1029/2008GL034969.

Brandt, P. C., K. K. Khurana, D. G. Mitchell, N. Sergis, K. Dialynas,J. F. Carbary, E. C. Roelof, C. P. Paranicas, S. M. Krimigis, and B. H.Mauk (2010), Saturn’s periodic magnetic field perturbations causedby a rotating partial ring current, Geophys. Res. Lett., 37, L22103,doi:10.1029/2010GL045285.

Brandt, P. C., K. Dialynas, I. Dandouras, D. G. Mitchell, P. Garnier,and S. M. Krimigis (2012), The distribution of Titan’s high-altitude

3039

Page 14: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

(out to �50,000 km) exosphere from energetic neutral atom (ENA)measurements by Cassini/INCA, Planet Space Sci., 60, 107–114,doi:10.1016/j.pss.2011.04.014.

Carbary, J. F., D. G. Mitchell, P. Brandt, E. C. Roelof, and S. M. Krimigis(2008), Statistical morphology of ENA emissions at Saturn, Geophys.Res. Lett., 113, A05210, doi:10.1029/2007JA012873.

Cassidy, T. A., and R. E. Johnson (2010), Collisional spreading ofEnceladus’ neutral cloud, Icarus, 209(2), 696–703, doi:10.1016/j.icarus.2010.04.010.

Compton, A. H., and I. A. Getting (1935), An apparent effect of galacticrotation on the intensity of cosmic rays, Phys. Rev., 47, 818.

Dialynas, K., S. M. Krimigis, D. G. Mitchell, D. C. Hamilton, N. Krupp, andP. C. Brandt (2009), Energetic ion spectral characteristics in the Saturnianmagnetosphere using C assini/MIMI measurements, J. Geophys. Res.,114, A01212, doi:10.1029/2008JA013761.

Dougherty, M. K., K. K. Khurana, F. M. Neubauer, C. T. Russel, J. Saur,J. S. Leisner, and M. E. Burton (2006), Identification of a dynamicatmosphere at Enceladus with the Cassini magnetometer, Science, 311,1406–1409, doi:10.1126/science.1120985.

Esposito, W. L., et al. (2005), Ultraviolet imaging spectroscopy shows anactive Saturnian system, Science, 307, 1251–1255.

Farmer, A. J. (2009), Saturn in hot water: Viscous evolution of theEnceladus torus, Icarus, 202, 280–286.

Forman, M. A. (1970), The Compton-Getting effect for cosmic ray particlesand photons and the Lorentz invariance of distributions functions, Planet.Space Sci., 18, 25–31.

Hansen, C. J., L. Esposito, A. I. F. Stewart, J. Colwell, A. Hendrix, W. Pryor,D. Shemansky, and R. West (2006), Enceladus water vapour plume,Science, 311, 1422–1425.

Hansen, C. J., et al. (2011), The composition and structure of the Enceladusplume, Geophys. Res. Lett., 38, L11202, doi:10.1029/2011GL047415.

Hedelt, P., Y. Ito, H. U. Keller, R. Reulke, P. Wurz, H Lammer, H. Rauer,and L. Esposito (2010), Titan’s atomic hydrogen corona, Icarus, 210,424, doi:10.1016/j.icarus.2010.06.012.

Hill, T. W., et al. (2005), Evidence for rotationally-driven plasma trans-port in Saturn’s magnetosphere, Geophys. Res. Lett., 32, L14S10,doi:10.1029/2005GL022620.

Ip, W.-H. (1996), The asymmetric distribution of Titan’s atomic hydrogencloud as a function of local time, Astrophys. J., 457, 922–932.

Ip, W.-H. (1997), On the neutral cloud distribution in the Saturnian magne-tosphere, Icarus, 126, 42–57.

Jia, X., K. C. Hansen, T. I. Gombosi, M. G. Kivelson, G. Toth,D. L. DeZeeuw, and A. J. Ridley (2012), Magnetospheric configurationand dynamics of Saturn’s magnetosphere: A global MHD simulation,J. Geophys. Res., 117, A05225, doi:10.1029/2011JA017367.

Johnson, R. E., M. R. Combi, J. L. Fox, W.-H. Ip, F. Leblanc, M. A.McGrath, V. I. Shematovich, D. F. Strobel, and J. H. Waite Jr. (2008),Exospheres and atmospheric escape, Space. Sci. Rev., 139, 355–397,doi:10.1007/s11214-008-9415-3.

Jurac, S., R. E. Johnson, and J. D. Richardson (2001a), Saturn’s E ring andproduction of the neutral torus, Icarus, 149, 384–396.

Jurac, S., R. E. Johnson, J. D. Richardson, and C. P. Paranicas (2001b),Satellite sputtering in Saturn’s magnetosphere, Planet. Space Sci., 49,319–326.

Jurac, S., and J. D. Richardson (2005), A self-consistent model of plasmaand neutrals at Saturn: Neutral cloud morphology, J. Geophys. Res., 110,A09220, doi:10.1029/2004JA010635.

Jurac, S., and J. D. Richardson (2007), Neutral cloud interaction withSaturn’s main rings, Geophys. Res. Lett., 34, L08102, doi:10.1029/2007GL029567.

Kane, M., D. G. Mitchell, J. F. Carbary, S. M. Krimigis, and F. J. Crary(2008), Plasma convection in Saturn’s outer magnetosphere determinedfrom ions detected by the Cassini INCA experiment, Geophys. Res. Lett.,35, L04102, doi:10.1029/2007GL032342.

Kirsch, E., S. M. Krimigis, W. H. Ip, and G. Gloeckler (1981), X-rayand energetic neutral particle emission from Saturn’s magnetosphere:Measurements by Voyager-1, Nature, 292, 718–721.

Kivelson, M. G. (2006), Does Enceladus govern magnetospheric dynamicsat Saturn?, Science, 311, 1391–1392.

Krimigis, S. M., et al. (2004), Magnetospheric Imaging Instrument on theCassini Mission to Saturn/Titan, Space Sci. Rev., 114, 233–329.

Krimigis, S. M., et al. (2005), Dynamics of Saturn’s magneto-sphere from MIMI during Cassini’s orbital insertion, Science, 307,1270–1273.

Krimigis, S. M., N. Sergis, D. G. Mitchell, D. C. Hamilton, and N. Krupp(2007), A dynamic rotating ring current around Saturn, Nature, 450,1050–1053, doi:10.1038/nature06425.

Mauk, B. H., D. G. Mitchell, S. M. Krimigis, E. C. Roelof, and C. P.Paranicas (2003), Energetic neutral atoms from a trans-Europa gas torusat Jupiter, Nature, 421, 920–922.

Mauk, B. H., et al. (2005), Energetic particle injections in Saturn’s magne-tosphere, Geophys. Res. Lett., 32, L14S05, doi:10.1029/2005GL022485.

McEntire, R. W., and D. G. Mitchell (1989), Instrumentation for globalmagnetospheric imaging via energetic neutral atoms, in Solar SystemPlasma Physics, Geophys. Monogr. Ser., vol. 54, edited by J. H. WaiteJr., J. L. Burch, and R. L. Moore, pp. 69–80, AGU, Washington, D. C.

Melin, H., D. E. Shemansky, and X. Liu (2009), The distribution of atomichydrogen and oxygen in the magnetosphere of Saturn, Planet. Space Sci.,57(14-15), 1743–1753, doi:10.1016/j.pss.2009.04.014.

Mitchell, D. G., et al. (2005), Energetic ion acceleration in Saturn’smagetosphere: Substorms on Saturn?, Geophys. Res. Lett., 32, L20S01,doi:10.1029/2005GL022647.

Mitchell, D. G., et al. (2009), Recurrent energization of plasma in themidnight-to-dawn quadrant of Saturn’s magnetosphere, and its rela-tionship to auroral UV and radio emissions, Planet. Space Sci., 57,1732–1742.

Nakamura, R., et al. (2002), Motion of the dipolarization front during a flowburst event observed by CLUSTER, Geophys. Res. Lett., 29(20), 1942,doi:10.1029/2002GL015763.

Nosé, M., H. Koshiishi, H. Matsumoto, P. C. son Brandt, K. Keika, K. Koga,T. Goka, and T. Obara (2010), Magnetic field dipolarization in the deepinner magnetosphere and its role in development of O+rich ring current,J. Geophys. Res., 115, A00J03, doi:10.1029/2010JA015321.

Paranicas, C., D. G. Mitchell, E. C. Roelof, P. C. Brandt, D. J. Williams,S. M. Krimigis, and B. H. Mauk (2005), Periodic intensity variationsin global ENA images of Saturn, Geophys. Res. Lett., 32, L21101,doi:10.1029/2005GL023656.

Paranicas, C. P., D. G. Mitchell, S. M. Krimigis, D. C. Hamilton,E. Roussos, N. Krupp, G. H. Jones, R. E. Johnson, J. F. Cooper,and T. P. Armstrong (2008), Sources and losses of energetic protonsin Saturn’s magnetosphere, Icarus, 197, 519–525, doi:10.1016/j.icarus.2008.05.011.

Paranicas, C. P., et al. (2010), Transport of energetic electronsinto Saturn’s inner magnetosphere, J. Geophys. Res., 115, A09214,doi:10.1029/2010JA015853.

Perry, M. E., B. Teolis, H. T. Smith, R. L. McNutt Jr., G. Fletcher,W. Kasprzak, B. Magee, D. G. Mitchell, and J. H. Waite Jr. (2010),Cassini INMS observations of neutral molecules in Saturn’s E ring,J. Geophys. Res., 115, A10206, doi:10.1029/2010JA015248.

Porco, C. C., et al. (2006), Cassini observes the active south pole ofEnceladus, Science, 311, 1393–1401.

Press, W. H., S. A. Teuklosky, W. T. Wetterling, and B. P. Flannery(1992), Modeling of data, in Numerical Recipes in C, 2nd ed., chap. 15,pp. 656–706, Cambridge Univ. Press, Cambridge, U. K.

Richardson, J. D., A. Eviatar, M. A. McGrath, and V. M. Vasyliunas(1998), OH in Saturn’s magnetosphere: Observations and implications,J. Geophys. Res., 103, 20,245–20,255.

Roelof, E. C., and A. J. Skinner (2000), Extraction of ion distributions ofmagnetospheric ENA and EUV images, Space Sci. Rev., 91, 437–459.

Roelof, E. C., D. G. Mitchell, and D. J. Williams (1985), Energetic neutralatoms (E 50 keV) from the ring current: IMP 7/8 and ISEE 1, J. Geophys.Res., 90(A11), 10991–11008.

Rymer, A. M., et al. (2009), Cassini evidence for rapid interchange transportat Saturn, Planet. Space Sci., 57, 1779–1784.

Saur, J., B. H. Mauk, A. Kaner, and F. M. Neubauer (2004), A model for theazimuthal plasma velocity in Saturn’s magnetosphere, J. Geophys. Res.,109, A05217, doi:10.1029/2003JA010207.

Sergis, N., S. M. Krimigis, D. G. Mitchell, D. C. Hamilton, N. Krupp,B. M. Mauk, E. C Roelof, and M. Dougherty (2007), Ring currentat Saturn: Energetic particle pressure in Saturn’s equatorial magneto-sphere measured with Cassini/MIMI, Geophys. Res. Lett., 34, A05217,doi:10.1029/2006GL029223.

Shemansky, D. E., and D. T. Hall (1992), The distribution of atomichydrogen in the magnetosphere of Saturn, J. Geophys. Res., 97,4143–4161.

Shemansky, D. E., P. Matheson, D. T. Hall, H.-Y. Hu, and T. M. Tripp(1993), Detection of the hydroxyl radical in the Saturn magnetosphere,Nature, 363, 329–331, doi:10.1038/363329a0.

Shemansky, D. E., X. Liu, and H. Melin (2009), The Saturn hydro-gen plume, Planet. Space Sci., 57, 1659–1670, doi:10.1016/j.pss.2009.05.002.

Sittler, E. C., R. E. Hartle, R. E. Johnson, J. F. Cooper, A. S. Lipatov,C. Bertucci, A. J. Coates, K. Szego, M. Shappirio, D. G. Simpson, andJ.-E. Wahlund (2010), Saturn’s magnetospheric interaction with Titan asdefined by Cassini encounters T9 and T18: New results., Planet. SpaceSci., 58, 327–350, doi:10.1016/j.pss.2009.09.017.

Smith, H. T., R. E. Johnson, M. E. Perry, D. G. Mitchell, R. L. McNutt,and D. T. Young (2010), Enceladus plume variability and the neutralgas densities in Saturn’s magnetosphere, J. Geophys. Res., 115, A10252,doi:10.1029/2009JA015184.

3040

Page 15: The extended Saturnian neutral cloud as revealed by global ENA simulations using Cassini/MIMI measurements

DIALYNAS ET AL.: SATURNIAN NEUTRAL CLOUD

Smyth, W. H., and M. L. Marcony (1993), The nature of the hydrogen toriof Titan and Triton, Icarus, 101, 18–32.

Tadokoro, H., H. Misawa, F. Tsuchiya, Y. Katoh, A. Morioka, andM. Yoneda (2012), Effect of photo-dissociation on the spreading of OHand O clouds in Saturn’s inner magnetosphere, J. Geophys. Res., 117,A09226, doi:10.1029/2011JA017492.

Thomsen, M. F., and J. A. Van Allen (1980), Motion of trapped electronsand protons in Saturn’s inner magnetosphere, J. Geophys. Res., 85,5831–5834.

Westlake, J. H., J. H. Waite Jr., K. E. Mandt, N. Carrasco, J. M. Bell,B. A. Magee, and J.-E. Wahlund (2012), Titan’s ionospheric compo-sition and structure: Photochemical modeling of Cassini INMS data,J. Geophys. Res., 117, E01003, doi:10.1029/2011JE003883.

Young, D. T., et al. (2005), Composition and dynamics of plasma in Saturn’smagnetosphere, Science, 307, 1262–1266.

3041