JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ???, XXXX, DOI:10.1029/,

*[1]

The Tailward Flow of Energetic Neutral Atoms Observed atMarsA. Galli and P. Wurz,1 E. Kallio,2 A. Ekenback, M. Holmstrom, S. Barabash, A.Grigoriev, Y. Futaana,3 and H. Gunell4

Abstract.The ASPERA-3 experiment on Mars Express provides the first measurements of en-

ergetic neutral atoms (ENAs) from Mars. These measurements are used to study the globalstructure of the interaction of the solar wind with the Martian atmosphere.

In this study we describe the tailward ENA flow observed at the nightside of Mars.After characterizing energy spectra of hydrogen ENA signals, we present composite im-ages of the ENA intensities and compare them to theoretical predictions (empirical andMHD models). We find that the tailward flow of hydrogen ENAs is mainly generatedby shocked solar wind protons. Despite intensive search, no oxygen ENAs above the in-strument threshold are detected. The results challenge existing plasma models and con-strain the hydrogen exospheric densities and atmospheric hydrogen and oxygen loss ratesat low solar activity.

1. IntroductionThe Analyzer of Space Plasmas and Energetic Atoms,

ASPERA-3, instrument consists of four different sensors[Barabash et al., 2006]. For in situ plasma measurements

1Physikalisches Institut, University of Bern, Bern,Switzerland.

2Institutet for Rymdfysik, Kiruna, Sweden.3Geospace Physics Laboratory, NASA Goddard Space

Flight Center, Greenbelt, MD, USA.4Department of Physics, West Virginia University,

Morgantown, WV, USA.

Copyright 2008 by the American Geophysical Union.0148-0227/08/$9.00

Figure 1. Maps of the median values of the electronfluxes (with energies Ee = 40–60 eV) and the proton num-ber density at the dawn and dusk sides in MSO cylindri-cal coordinates with a grid size of 0.1 RM. Measurementsare from the ELS and IMA sensors, respectively, for theperiod from 1 February 2004 and 1 February 2006. Solidcurves show the position of the bow shock, BS, and in-duced magnetospheric boundary, MB. Figure taken fromDubinin et al. [2008].

there are the Ion Mass Analyzer (IMA) and the ElectronSpectrometer (ELS). For remote sensing using energetic neu-tral atoms there are the Neutral Particle Imager (NPI) andthe Neutral Particle Detector (NPD). NPI has a 360◦×5◦ in-stantaneous field-of-view with good angular resolution, butlacks the energy measurement and the particle identifica-tion. For a recent NPI analysis see Gunell et al. [2006a].NPD has a 180◦ × 6◦ instantaneous field-of-view with mod-erate angular resolution, but with energy measurement andmass identification. The present study is focused on theNPD data recorded at the nightside of Mars.

In the last four years, the ASPERA-3 experiment on MarsExpress has revealed many aspects of the plasma environ-ment around Mars [Barabash et al., 2007a]. A comprehen-sive summary of the first years of IMA and ELS observationsof plasma boundaries, ion outflow, escape processes, influ-ence of crustal magnetic fields and photoelectrons was givenby Dubinin et al. [2006]; a statistical study of the plasmamoments derived from IMA data were presented by Franzet al. [2006] (the proton and electron densities around Marsare shown in Figure 1). Ion escape rates were studied byBarabash et al. [2007a], the chemical composition of escap-ing plasma was investigated by Carlsson et al. [2006].

The interaction of the solar wind plasma with the plan-etary environment leads to the formation of two distinctboundaries. Figure 1 shows the electron and proton densi-ties in the Mars environment with the two plasma bound-aries (thin black lines) superimposed on the data with densi-ties derived from ELS and IMA measurements, respectively[Dubinin et al., 2008]. At the bow shock, the solar windis slowed down to subsonic velocities, the magnetosheathbetween the bow shock and the inner plasma boundary isdominated by slow, heated up solar wind. The inner plasmaboundary can either be defined as the boundary where theplanetary ions start to outnumber the magnetosheath ionsor as the boundary where the interplanetary magnetic fieldBIMF piles up around the ionosphere. The two different con-cepts are equivalent [Boßwetter et al., 2004]. In this paperwe shall use the term “Induced Magnetosphere Boundary”(IMB) for the inner plasma boundary.

The motivation for the energetic neutral atom (ENA) ob-servations at Mars is to provide a global view of the interac-tion of the solar wind with the neutral atmosphere of Mars,a non-magnetized planet, to complement the local plasmameasurements. To fully interpret the ENA observations, acomparison with ENA models is needed.

1

X - 2 GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS

An ENA is the product of a charge exchange reaction be-tween an energetic ion and a neutral atom. In the presentstudy on ENA observations at Mars the energetic ions aretypically solar wind protons and the neutrals are hydrogenatoms of the planetary exosphere. For a flow of mono-energetic hydrogen ions jp(r) that are neutralized alongtheir way through a pure hydrogen atomic gas with a densitynH(r) the ENA production equation reads:

JENA = σ

Z

LOS

dr nH(r)jp(r), (1)

with JENA the ENA intensity in cm−2 sr−1 s−1 and σ thecharge exchange cross-section. For 1 keV, which is the typi-cal energy of ENAs measured with NPD, σ ≈ 2 · 10−15 cm2

for a proton−hydrogen reaction, and σ ≈ 1 · 10−15 cm2

for O+−hydrogen charge exchange [Lindsay and Stebbings,2005]. ENA measurements therefore reflect the plasma dis-tribution as well as the neutral densities along the line-of-sight (LOS) of the sensor.

Before Mars Express arrived at Mars, three different ENAsources were expected around a non-magnetized planet withan atmosphere. NPD observations of the dayside of Marsconfirmed the existence of an ENA albedo caused by solarwind ENAs that are scattered back from the dayside at-mosphere [Futaana et al., 2006a]. The NPD measurementsalso revealed a narrow ENA stream of shocked solar windright above the sub-solar point in the magnetosheath [Fu-taana et al., 2006b; 2006c; Grigoriev et al., 2006; Mura etal., 2008]. At the nightside of Mars, the expected flow oftailward (moving towards the magnetotail) ENAs was alsodetected [Galli et al., 2006b]. Unperturbed solar wind pro-tons, magnetosheath plasma, and accelerated planetary ionsare possible parent ions for this tailward ENA flow, the latterparent population being the only one that also may producea measurable amount of oxygen ENAs. Exospheric hydro-gen is the most important neutralizing agent because of itslarge scale height as we will show below.

A first study on the tailward flowing ENAs was presentedby Galli et al. [2006b]. The present work is a comprehen-sive re-evaluation of the earlier study, which became neces-sary because the final calibration of the NPD sensor becameavailable since the publication of the earlier study [Grig-oriev, 2007]. Moreover, we added data from all NPD sensormodes to the present analysis. Finally, we performed a sta-tistical comparison of the observations with ENA model pre-dictions. A similar study of the Venusian nightside ENAs,based on data obtained with the ASPERA-4/NPD sensoron Venus Express, has been presented recently by Galli etal. [2008].

We will briefly describe the NPD sensor in Section 2, be-fore presenting the database that underlies this work (Sec-tion 3). In Section 4 we will show the global intensity imagesof tailward flowing hydrogen ENAs; they will be comparedto model predictions in Section 5. Also, a brief compari-son to the H-ENA tailward flow measured with ASPERA-4/NPD at the nightside of Venus is given. The search foroxygen ENAs will be summarized in Section 6, followed bythe discussion in Section 7.

2. The NPD sensorNPD is designed to measure hydrogen and oxygen ENAs

at energies between 0.2 and 10 keV, using the time-of-flight(TOF) technique. Angular resolution is provided by havingtwo NPD sensors (NPD1 and NPD2), each with three angu-lar channels (NPD1 0, NPD1 1, NPD1 2, NPD2 0, NPD2 1,and NPD2 2). Each NPD channel has a field-of-view (FOV)of 40◦ × 6◦ (see Figure 2, top panel, for the viewing di-rections) giving a total instantaneous FOV of 180◦ × 6◦

[Barabash et al., 2006; Grigoriev, 2007]. We can distinguishbetween hydrogen and oxygen ENAs mainly because theirTOF values do not overlap for most energies that occur inthe Martian plasma environment.

NPD can be run in three different data acquisition modes.BINNING mode offers high temporal resolution (see lowerpanel in Figure 3) with a resolution of only 16 TOF bins.RAW mode and TOF mode offer 256 TOF bins between 0and 2048 ns (see middle panel in Figure 2) but feature a lowduty cycle and therefore have a lower temporal resolution.In this work we have included measurements obtained withall three data acquisition modes. We will focus on the in-tegral intensity of ENAs because this is the parameter thatis most reliably derived from all different data acquisitionmodes.

Apart from ENAs, NPD is also sensitive to UV photons.Observations with the Sun in the FOV therefore have tobe avoided. Figure 3 shows the closest possible approachof the FOV to the Sun outside the Mars shadow. On theother hand, the UV sensitivity of NPD can also be used toestimate the neutral hydrogen exospheric density with theLyman−α radiation as a proxy [Galli et al., 2006c].

;————————————————-

3. Data baseFor this study we included all NPD measurements during

which the spacecraft was inside the IMB at the nightsideof Mars. Most of these observations fall into the periodbetween 23 April and 26 May 2004. To estimate the exten-sion of the IMB we relied on simultaneous ion and electronmeasurements with IMA and ELS. NPD measurements withpoor counting statistics were excluded from further analy-sis. The measurements were then divided into intervals of1 to 10 minutes integration time, one minute being enoughonly for the most intense ENA signals. All derived ENA in-tensities apply to the energy range from 0.2 to 10 keV. Thedetection threshold is 5 ·103 cm−2 sr−1 s−1. The typical un-certainty (the combination of statistical and instrumentaluncertainty) of intensities for a single time interval amountsto 20%.

The reconstruction of differential ENA intensities fromraw count rates follows the method established in Galli etal. [2006a]. As improvement over the first report on tail-ward ENAs [Galli et al., 2006b] we have now completed thedatabase for observations inside the IMB and we have in-corporated the final laboratory calibration data for NPD[Grigoriev, 2007]. Compared to the estimates used in ourfirst publication [Galli et al., 2006b] the geometric factorhas not changed but the detection efficiency is a factor oftwo higher. All ENA intensities therefore decrease by a fac-tor of two.

The time span covered by the NPD observations at thenightside of Mars corresponds to low solar activity (solaractivity index is around 40, F10.7 is around 100). Unfortu-nately, we do not know the exact angle of the interplanetarymagnetic field BIMF for a particular orbit, lacking a mag-netometer on board Mars Express. This makes it harder tointerpret the ENA tailward flow because planetary ENAs areexpected to be organized around the BIMF draping direction[Lichtenegger et al., 2002]. About the solar wind parametersfor particular orbits we do not know much either: IMA wasnot designed as a solar wind monitor, and there are severalNPD observations for whose orbits no plasma measurementsoutside the bow shock exist at all. Franz et al. [2006] find,as an average from February 2004 to February 2006, 300 to400 km s−1 as best estimate for the median solar wind bulkspeed and 1 to 3 cm−3 for the density of the unperturbedsolar wind. This is a typical solar wind strength expectedfor 1.5 AU heliocentric distance.

GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS X - 3

Figure 2. Top: Location of the six field-of-views of NPDis shown in the ecliptic coordinate system. Also the Marslimb (solid line) and the location of the Sun (cross) areindicated. Middle: Typical H-ENA TOF spectrum (thinline) from inside the IMB at the nightside of Mars on 23April 2004 recorded with channel 2 of NPD1 (NPD1 2).The thick line is the reconstructed ENA signal with thebackground removed. Bottom: Energy spectrum calcu-lated from the TOF spectrum.

4. ResultsFigure 2, top panel, shows the viewing directions of the

2×3 and FOVs of the NPD1 and NPD2 angular channels rel-ative to the Mars limb (circle) and the Sun (cross) for a typ-ical observation of tailward ENAs. Figure 2, middle panel,shows a typical example of raw TOF data of the tailwardflowing hydrogen ENAs. The spectrum was measured withNPD1 2 channel with the viewing geometry given in panel1 and accumulated over 4 minutes, where the original signal(thin line) and the reconstructed ENA signal (bold line) areindicated. In addition, the flat background from UV pho-tons is indicated by the dashed line. The integral ENA in-tensities for the ENA tailward flow are in the range between104 and 105 cm−2 sr−1 s−1, and the intensity for the par-ticular measurement presented in Figure 2 is 4.5 · 104 cm−2

sr−1 s−1. The bottom panel in Figure 2 shows the ENA en-ergy spectrum derived from the reconstructed TOF signal(bold line in middle panel), with 1σ-error bars and an esti-mate of the integral intensity. The derivation of the ENAenergy spectrums works for energies between 0.3 keV and5 keV, below and above this energy range the spectral val-ues are obviously ill constrained by NPD measurement. Thesolid line gives a fit with a two-component power law energyspectrum with a roll-over around 1 keV. The NPD2 sectorsthat are directed toward the anti-Mars hemisphere, i.e., the

hemisphere when looking in zenith direction (see top panelof Figure 2) detect only a weak TOF signal correspondingto ENA intensities below 1 · 104 cm−2 sr−1 s−1.

Figure 3 shows an ideal spacecraft orientation and loca-tion to measure intense ENA signals: With up to 2.4 · 105

cm−2 sr−1 s−1 the signal measured on 25 April 2004, during21:14–21:44 UT, is the highest ENA intensity ever registeredin the tailward flow at Mars. After 21:20 UT, the Sun disap-pears behind the planetary limb and the ENA signal dropsto 2·104 cm−2 sr−1 s−1 within 1 minute. Unfortunately, it isnot possible to reconstruct ENA intensities for significantlyshorter time intervals because of the counting statistics. Af-ter Mars Express has entered into eclipse, NPD1 2 continuesto measure a weak ENA signal because the FOV is directedclose to the limb and a part of the ENAs originating from

Figure 3. Top panel: two observation configurationsfor 21:20 UT, when Mars Express was about to enter theshadow of Mars, and for 21:40 UT, when the eclipse pe-riod approached the end. Bottom panel: Intensity of tail-ward flowing H-ENAs as a function of time, seen in chan-nel NPD1 2 on 25 April 2004, 21:14–21:45 UT. The blackline shows the raw count rates (including background)with one-second resolution, the red numbers are integralENA intensities derived from a few minutes integrationtime.

0.2 0.5 1 2energy [keV]

1e+03

1e+04

1e+05

1e+06

inte

gral

flux

[cm

s]

#11, MHD model, integral flux#12, MHD model, integral flux#12, most intense MEX/NPD signal#11 and #12, Typical MEX/NPD spectrum

Figure 4. MHD model prediction for the energy spec-trum of the integral ENA flux for spacecraft position #11(red line) and #12 (orange line) versus typical energyspectra measured with NPD (black lines).

X - 4 GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS

the hot solar wind in the magnetosheath are emitted intothe shadow of Mars.

The intensity variation over minutes exhibited in Figure 3are clearly spatial effects as the spacecraft position changesrelative to the Mars and Sun direction. There are also tem-poral oscillations but among a dozen different time seriesno preferred oscillation period is found. This is differentfrom the subsolar ENA jet at the dayside, which was foundto oscillate with oscillation periods of 50 and 300 seconds[Grigoriev et al., 2006]. For the nightside observations thereare not enough data to conduct detailed investigations oftemporal fluctuations, and we always face the problem ofa large FOV size of at least 40◦ × 6◦ and of a fast movingspacecraft. Moreover, we usually do not have simultaneous

111213

Figure 5. Orbit plot of Mars Express on 3 May 2004 incylindrical coordinates. The time periods during whichthe ENA measurements incorporated in images #11,#12, and #13 were obtained are marked with coloredlines.

Figure 6. ENA intensity image of Mars #11, including281 intervals from 23 April to 23 May 2004. The averageFOV footprint is 40◦× 6◦, the average solar zenith angle(SZA) of the spacecraft is 147◦, the average altitude is1.3 RM.

information of the unperturbed solar wind in front of theMartian bow shock or the BIMF directions, which is a severedeficiency when discussing individual ENA observations.

ENA signals that are strong enough can be inverted intoenergy spectra [Galli et al., 2006a], which can be describedby a two-component power-law (see Figure 2 for a typi-cal example). Considering all of these energy spectra wefind a median roll-over around 1.0 keV, a median slope atlower energies of −2.2, and a slope of −2.7 at energies above1.0 keV. These results are in agreement with our previousreport [Galli et al., 2006b]. These energy spectra are consis-tent with ENAs from shocked solar wind.

Figure 7. ENA intensity image of Mars #12, including249 intervals from 23 April to 23 May 2004. The averageFOV footprint is 50◦× 10◦, the average SZA is 152◦, theaverage altitude is 0.8 RM. Color scheme is the same asin Figure 6.

Figure 8. ENA all sky intensity image of Mars #13, in-cluding 184 intervals from 27 April to 26 May 2004. Theaverage FOV footprint is 60◦ × 20◦, the average SZA is143◦, the average altitude is 0.45 RM. Color scheme isthe same as in Figure 6.

GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS X - 5

Figure 4 shows the comparison of the MHD model pre-diction and the NPD measurement where two model ENAenergy spectra are given: one ENA spectrum for unper-turbed solar wind ions if the Sun is outside the Mars limb(red line), and one spectrum for a spacecraft position in-side the Mars eclipse (orange line). The total flux rate islower for the MHD ENA spectrum inside the Mars eclipse,and the spectral shape is smeared to a two-component powerlaw with a roll-over at 1.5 keV (the spacecraft positions #11and #12 correspond to the vantage points in Figures 6 and7). The smeared energy spectrum is similar to the energyspectra actually observed with NPD, the solid and dashedblack lines shown in Figure 4. The ENA spectrum for un-perturbed solar wind ions is considerably different from theobserved spectra. However, we do not have a NPD measure-ment from the solar direction (see image shown in Figure 6)because NPD cannot observed with the Sun inside the FOV.Actually, there are very few well-constrained energy spectrafrom the near-Sun direction. Therefore, we cannot confirmthat the transition in the ENA spectra is as pronounced assuggested by the MHD model. Note that the height of theobserved spectrum is arbitrary in Figure 4 since NPD didnot cover the maximum of the ENA outflow for image #11(Figure 6).

For the quantitative comparison with ENA models wewill now focus on the global image of integral intensities.

From single measurements such as those shown in Fig-ure 3 we synthesized 3 ENA intensity images that containvirtually (714 out of 792) all measurement intervals. Themeasurement intervals were organized in three different plotcategories (#11, #12, and #13) for which the Mars centervaries within ±10◦ at most and the variations in altitudecorrespond to a variation of apparent planetary radius of≤ 10◦. The vantage point of the spacecraft for the threecategories is depicted in Figure 5. To create composite im-ages the single intensity measurements were averaged over a10◦ × 10◦ square mesh. The results are shown in Figures 6,7, and 8. These three images update and complete Figure9 in the report of Galli et al. [2006b]. The plots are shownin a cylindrical projection of the Mars Solar Orbital (MSO)reference frame, the x-axis is the MSO longitude, the y-axisis the MSO latitude in degrees. The MSO reference frame isdefined as follows: X points from Mars to the Sun, Z pointsto the North pole of the Martian orbital plane. By this def-inition, the Sun direction is always at (0◦/0◦). Bins thatare covered more than once by the NPD FOV are shownin color corresponding to the observed ENA intensity. Theidentical color scale is used for all images in this work. Pix-els that have not been covered are left white. Because of thefast spacecraft proper motion during the pericenter passageof Mars Express single measurements are associated with aFOV footprint of typically 50◦ × 10◦. This has to be takeninto account when comparing model images with a muchhigher angular resolution to the observations.

The composite ENA images show the following big pic-ture: Around the Mars limb close to the Sun direction (1to 2)·105 cm−2 sr−1 s−1 H-ENAs are seen. If the Sun isoutside the limb and the NPD FOV is directed almost to-wards it this signal may reach up to (2.4 ± 0.8) · 105 cm−2

sr−1 s−1. The ENA intensities decrease as the FOV is di-rected to the limb further away from the Sun direction, butdistinct signals of up to 4 · 104 cm−2 sr−1 s−1 are still seen(blue pixels in Figures 7 and 8). These signals cannot beexplained as a spatial smudging effect of the FOV size, theyoriginate really from the exosphere around the far-Sun limb.The nightside of the planet itself and the deep space, on theother hand, do not emit ENA signals above an intensity of1·104 cm−2 sr−1 s−1 (see the black areas in Figures 7 and 8).The boundary between the purple and black area (around−100◦ longitude in Figure 8) marks the end of the region ofdetectable ENA production. The exact transition is a little

arbitrary because the size of the NPD FOV broadens theapparent ENA outflow considerably. Nonetheless, Mars isa broad ENA source. In all three ENA images the entireSun-ward hemisphere emits ENAs, except the Mars surfaceitself.

The intensity of the maximum ENA outflow and its con-centration around the limb close to the Sun direction areconsistent with H-ENAs from shocked solar wind but as wewill show in the next Section 5 the opening angle of the out-flow (up to 60◦ around the Sun direction) is hard to reconcilewith a model of pure shocked solar wind ENAs. The open-ing angle of the ENA tailward flow, α, as observed in theENA images (Figures 6 to 8) should scale with the plasmatemperature of the parent ions around the terminator with:

α ≈ arctanp

kBT/E (2)

where kB is the Boltzmann constant, T is the ion temper-ature, and E is energy corresponding to the bulk flow ofthe plasma. If the tailward ENA flow is produced only fromshocked solar wind, i.e., magnetosheath ions, Equation 2 im-plies for α = 60◦ a temperature of the plasma between bowshock and IMB of kBT = 0.5 keV, assuming a flow velocityof 200 km s−1 [Franz et al., 2006]. This plasma temper-ature is higher than the 0.15 keV established by Franz etal. [2006] from proton measurements in the magnetosheath.This discrepancy is statistically significant, as will be elabo-rated in Section 5. The thermal energy of the unperturbedsolar wind is even two orders of magnitude smaller thanits kinetic energy, which implies a scatter angle α of a fewdegrees (from Equation 2). Therefore, the contribution ofunperturbed solar wind ENAs can safely be neglected assoon as the spacecraft is inside the Mars shadow (Figures 7and 8).

The black and purple pixels of the nightside itself indi-cate an upper limit of nightside ENA emission of 1 · 104

cm−2 sr−1 s−1. Such an emission would be produced byprotons that precipitate into the atmosphere on the night-side of the planet where they are neutralized and partiallyscattered back. This process has been observed with NPDabove the dayside of Mars [Futaana et al., 2006a]. Assum-ing a backscatter efficiency of 58% for the neutralized ions[Kallio and Janhunen, 2001] the NPD images impose an up-per limit on protons that flow inside the magnetotail towardsthe nightside of Mars:

jH+ <1

0.582π × 104 cm−2 sr−1 s−1 = 1 · 105 cm−2 s−1.(3)

The vast majority of the anti-Mars hemisphere was notcovered or, where covered, it was found to emit less than1·104 cm−2 sr−1 s−1 (black pixels in Figure 8). The singleoccasions when NPD detects a signal of 1·104 cm−2 sr−1 s−1

are still too close to the Martian limb to decide whether theweak ENA signals are due to anti-tailward flowing ENAsinside the IMB or whether they have a non-planetary ori-gin. The ENA intensity images shown in this work onlyimpose an upper limit of a few 104 cm−2 sr−1 s−1 on pos-sible non-planetary H-ENAs. This question is discussed indetail elsewhere [Galli et al., 2006a; Wurz et al., 2008].

5. Statistical comparison of the observedH-ENA tailward flow with ENA models

To predict the ENA production of a planet, two compo-nents are required according to Equation 1: a simulationplasma environment and a model of the neutral exosphere.

X - 6 GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS

For the plasma we have relied on the empirical model pre-sented by Kallio et al. [1997] and on the MHD calculation ofMa et al. [2002]. Both models only include solar wind pro-tons. To reproduce the three ENA images shown in Figures6, 7, and 8 we have then evaluated Equation 1 by combiningthe predicted proton flow with a neutral exosphere model.The latter is based on a Chamberlain model [Chamberlainand Hunten, 1987], the only input parameters required arethe exobase density and temperature of the exospheric com-ponents. The ENA model predictions used in this work havebeen compared to each other by Gunell et al. [2006b]. Wewill now compare the models with the ENA measurements.This comparison will show whether the models also repro-duce the global picture of solar wind interaction with theMartian atmosphere correctly by simulating by the globalplasma distribution, and from that the ENA emission usinga model of the exospheric gas densities. Since we can easilyvary the exospheric parameters in the two models this com-parison will also constrain some of the neutral exosphericdensities. The ENA outflow is of course only one of severalobservational constraints for the models. The input param-eters of the ENA model prediction must also be consistentwith independent measurements of ion and neutral densi-ties. Among all model parameters, the solar wind dynamicpressure and the density and temperature of the neutral hy-drogen exosphere have the biggest effect on the predictedENA outflow.

For the solar wind parameters the empirical model as-sumes the following values: nSW = 2.5 cm−3, TSW = 15 eV,and vSW = 400 km s−1. For the MHD model Ma et al.[2002] assumed nSW = 4 cm−3, and vSW = 500 km s−1. Thedifference between the two models corresponds to the plau-sible range of uncertainty about the true solar wind plasmaparameters in spring 2004. More extreme deviations fromnominal solar wind conditions need not be studied since theENA images shown in Figures 6, 7, and 8 are averages overone month of observations. Given the limited number ofENA observations available, we cannot investigate the ef-fect of the orientation of the upstream BIMF on the ENAemission, using IMF directions from MGS data as has beendone in other studies [e.g. Barabash et al., 2007a]. Regard-ing the effect of the crustal fields on the ion populations ithas been concluded that they do not significantly affect theoverall ion distribution, although the strongest anomalieshave a small effect on the local ion outflow [Nilsson et al.,2006]. Therefore, we do not consider the crustal field in ouranalysis.

The exospheric densities show a much higher degree of un-certainty in the literature. As starting point we define a fa-vorite scenario for the Martian neutral exosphere at low solaractivity, for which the parameters are listed in Table 1. Theneutral hydrogen is the most important exospheric speciesfor the ENA production. In contrast, the hydrogen ENAobservations are not sensitive to neutral helium (too tenu-ous and the cross-section is too small). Probably, H2 doesnot affect the H-ENA production either, the correspondingcharge exchange cross-section is 5 times lower than for aH+−H reaction. The hot oxygen component is also of mi-nor importance because of the presence of hot hydrogen that

Table 1. Default model parameters for the Martian exo-sphere at low solar activity.

Species Temperature [K] Exobase density [m−3]thermal H 200 1 · 1011

hot H 600 1 · 1010

H2 200 1 · 1012

thermal O 200 1 · 1013

hot O 4400 4 · 109

dominates the ENA production at high altitudes. The coldhydrogen component is constrained by the UV limb emissionmeasurements with NPD and SPICAM: for the year 2004Galli et al. [2006c] find a very low value of a few 1010 m−3

for the North pole region, and for the year 2005 Chaufrayet al. [2008] derive 1 · 1011 m−3 and 2 · 1011 m−3 above thesubsolar region and the terminator. The hot hydrogen com-ponent is found by both teams to be 1 · 1010 m−3 at theexobase, with a temperature of T > 500 K. The parametersof the hot and thermal oxygen components are motivated

Figure 9. Predicted H-ENA outflow based on the empir-ical plasma model, calculated for the observation config-uration of image #13 (Figure 8). The lower panel showsthe smeared version of the original model image in theupper panel.

GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS X - 7

by recent models of Lichtenegger et al. [2006] and Chaufrayet al. [2007].

The empirical plasma model, combined with the parame-ters listed in Table 1, yields ENA intensities that are similarto the NPD images within a factor of two for the regionsof maximum ENA outflow. Figure 9 shows the model pre-diction calculated for the same planetary distance and solarzenith angle as in Figure 8. The upper panel shows the orig-inal model prediction with a grid size of 2.5◦×2.5◦, the lower

Figure 10. Predicted H-ENA outflow based on theMHD model, calculated for the observation configura-tion of image #12. The lower panel shows the smearedversion of the original model image in the upper panel.Color scheme is the same as in Figure 6.

panel shows the same image, but convolved with the FOVfootprints that are underlying the observation in Figure 8.

The spatial distribution of the predicted outflow lookssimilar to the measurement, the maximum occurs close tothe Mars limb symmetrical around the Sun position, whilea faint halo of H-ENAs extends along the limb 90◦ away.Figure 9 is dominated by H-ENAs that result from chargeexchange with neutral hydrogen. The H-ENAs neutralizedwith neutral oxygen form a thin ring-like structure but theH-ENAs originating from charge exchange with neutral hy-

Figure 11. Predicted H-ENA outflow based on theMHD model, calculated for the observation configura-tion of image #13. The lower panel shows the smearedversion of the original model image in the upper panel.Color scheme is the same as in Figure 6.

X - 8 GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS

drogen are more intense even for a LOS intersecting theexobase. Given the coarse spatial resolution of NPD, theH-ENA images are not sensitive to the exospheric oxygen,i.e., its density cannot be estimated from NPD observations.

In spite of the similarity between Figures 9 and 8 thereare two discrepancies between the empirical model and theNPD observations. First, the ENA intensities of the most

Figure 12. H-ENA image of the nightside of Venus,adapted from Galli et al. [2008]. The upper panel showsthe ASPERA-4/NPD image obtained in August 2006.The lower panel shows the smeared version of the MHDmodel image for the same observation configuration. Theformat and the color scale are identical to the ENA im-ages of Mars.

intense pixels close to the Sun are predicted too low by afactor of two. Second, and statistically more significant, theobserved outflow is much broader (ENAs are observed allalong the Mars limb and from the exosphere outside the far-Sun limb) than predicted. This problem cannot be fixed byassuming a denser hydrogen exosphere.

The resulting ENA images from the MHD model lookidentical to the ones obtained with the empirical model, theχ2-fit probability (Equation 4) does not differ significantly.In Figures 10 and 11 we show the resulting images for twodifferent MHD simulation runs, to be compared again to theNPD images #12 and #13 shown in Figures 7 and 8. Com-parisons to image #11 are not shown because they do notadd further information.

To assess if the ENA model images (lower panels in Fig-ures 10 and 11) are a plausible fit to the NPD observationswe calculate the square sum over all image pixels of observedENA intensity jobs minus predicted ENA intensity jmod, di-vided by the uncertainty of the observed ENA intensity σj:

χ2 = Σi

µjobs,i − jmod,i

σj,i

∂2

. (4)

Then we check the probability P (χ2, f) that χ2 fromEquation 4 is likely caused by chance or if the model pre-diction is really far from reality. The degree of freedom f isthe number i of statistically reliable image pixels (approxi-mately the number of independent measurement intervals)minus the number of model parameters. For the numberof model parameters we set 12, appropriate to describe thesolar wind plasma parameters and the exospheric densitiesof the thermal and hot hydrogen. The typical uncertaintiesof single pixels σj are half as large as jobs.

For images #11 and #12 the default MHD model with theexosphere parameters described in Table 1 yields a very lowfit probability of P (χ2, f) < 10−6. Obviously, the modellacks something to correctly reproduce the observed ENAimages shown in Figures 6 and 7. For image #13 the agree-ment is better but far from satisfying (P = 1%). Roughlythe same fit probabilities emerge for the empirical plasmamodel. We conclude that the ENA models of shocked so-lar wind resemble the observations but the observed ENAoutflow is significantly broader than predicted.

A copy of NPD is operating since 2006 as part of theASPERA-4 experiment [Barabash et al., 2007b] on VenusExpress. Recently, we presented a study of the tailwardENA flow observed with the ASPERA-4/NPD sensor [Galliet al., 2008]. The ENA tailward flow at Venus is similar tothe one seen at the nightside of Mars. The only difference isthat the ENA outflow at Venus is concentrated in a narrowerangular range. The opening angle of the ENA tailward flowat Venus is only 30◦ compared to 60◦ at Mars, although thelarge FOV footprints tend to blur the features. Note thatthe reference frame and the color scale of the ENA intensi-ties are identical to the Mars images. The reason for Venusbeing a narrower ENA source is the stronger gravitationalpull of Venus, which results in lower exospheric scale heightsabove the exobase. This difference was predicted by ENAmodels prior to Venus Express [Gunell et al., 2006b]. Un-like for Mars, for Venus the measured ENA image can bereproduced with a MHD code that only takes into accountshocked solar wind ENAs. One example is shown in Fig-ure 12, where the upper panel shows the NPD image andthe lower panel shows the MHD model image (again smearedby the FOV footprints). Although a compilation of severalweeks of ENA measurements is also needed for Venus ENAimages, similar to the present Mars observations, we findgood agreement with the MHD model results. Thus, theeffect of a variable BIMF direction on the ENA image mustbe lower than the present image resolution.

We have investigated the following ideas to explain thediscrepancy between models and observations at Mars:

GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS X - 9

1. Increasing the exospheric densities of H2, helium, oxy-gen or any other heavy species does not explain the width ofthe H-ENA outflow because the corresponding scale heightsare much smaller than the ones of hydrogen. The tenuoushot oxygen component does not influence the H-ENA im-ages anywhere because of the hot hydrogen component.We have verified with the empirical model that leaving awaythe exospheric oxygen altogether does not significantly al-ter the predicted H-ENA image. The neutral oxygen con-tributes to the H-ENA production only in a thin ring ofa few 100 km around the exobase. An exobase density of1014 m−3 for the cool oxygen would explain some of theENA emissions at the Mars limb far away from the Sun.However, Chaufray et al. [2007] have deduced O+ and O-ENA loss rates (see Section 6) that are consistent with theASPERA-3 observations [Barabash et al., 2007a], assumingan oxygen exobase density of only nO = 1013 m−3.

2. The cool and the hot hydrogen exosphere cannot bemuch denser without contradicting Lyman-α measurementsmade in 2004 [Galli et al., 2006c] and 2005 [Chaufray et al.,2008]. The temperature of the Martian thermal exospherehas been shown to be 200 K by several in situ and aer-obraking measurements [Lichtenegger et al., 2006], and thevertical profile of the hot hydrogen component does not varysignificantly for arbitrarily high temperatures above 600 K[Galli et al., 2006c].In the future we want to investigate whether the Chamber-lain model for the density profile is an oversimplification, re-sulting in too low densities at high altitudes. But we doubtthat this could explain the entire discrepancy. To see if theneutral hydrogen exosphere is responsible for the shortcom-ings of the model we have just varied the predicted ENAintensity by c×jENA, with c = 1 to 10. This is equivalent tovarying the exobase densities of the thermal hydrogen be-tween 1011 and 1012 m−3 and the one of the hot componentbetween 1010 and 1011 m−3. The results are not satisfying.First, the optimized c, which minimizes χ2 of Equation 4, isnot consistent for different images; for image #12 c = 6, andfor image #13 c = 2. Second, even for optimized c the abso-lute fit probability P is still below 10−3 for images #11 and#12. Even with a hot hydrogen component whose exobasedensity is an order of magnitude higher than measured byGalli et al. [2006c] and by Chaufray et al. [2008] the modelimages do not reproduce the outflow lobe shown in Figure 8by the blue and purple pixels, or in Figure 7 by the greenpixels. It seems that other trajectories of the magnetosheathplasma or new ion populations are required to explain theENA outflow far away from the Sun direction.

3. Extreme solar wind conditions offer no plausible ex-planation: The MHD model does not fit significantly betterto the observations although the assumed solar wind fluxupstream of the bow shock nSWvSW is twice as high as inthe empirical model.

4. Maybe the plasma temperature in the magnetosheathis underestimated by the models or the MHD approximationis insufficient for the shocked solar wind around Mars. Atpresent, we are working on a hybrid plasma model to testthis possibility.

5. As observed with NPD [Futaana et al., 2006a] thedayside of Mars emits several 105 cm−2 sr−1 s−1 hydrogenENAs. A part of this signal is caused by reflected ENAs, theother part is caused by solar wind protons that precipitateinto the atmosphere before being neutralized and scatteredback to space as ENAs. For the precipitating protons Equa-tion 3 yields a flux of a few 106 cm−2 s−1 at subsolar point.According to model calculations by Kallio and Janhunen[2001] the flux of precipitating protons at the terminator re-gion still reaches 1% to 10 % of this value. The precipitatingsolar wind protons thus might explain the ENA signals of afew 104 cm−2 s−1 originating directly from the Mars limb.For the broad ENA outflow outside the limb another processhas to be responsible.

6. The broad ENA outflow of several 104 cm−2 sr−1 s−1

(blue lobe in Figure 8) far away from the Sun direction mightbe explained with planetary protons that are swept away bythe convective electric field from the exosphere around theterminator region. If the models show that we have to in-clude planetary H-ENAs the lack of accurate magnetometerdata poses a severe problem. Since we do not know theBIMF draping direction we cannot organize the ENA dataalong the convective electric field direction to check for acorrelation. The energy spectrum of planetary H-ENAs issimilar to solar wind ENAs because planetary pick-up ionsand shocked solar wind protons share comparable energies[Lichtenegger et al., 2002].

6. Upper limit of observed O-ENAs atMars

Observing oxygen ENAs at Mars would directly reveal anatmospheric erosion process. If there is an observable oxy-gen ENA signal it is truly planetary since in the solar windthe oxygen abundance is 5000 times lower than hydrogen.The most likely location to identify Martian O-ENAs is thetailward flow close to the limb. According to the laboratorycalibration already an O-ENA signal of more than 1 · 104

cm−2 sr−1 s−1 is visible even in the presence of UV back-ground and a H-ENA peak if the energy lies above 0.5 keV.

For O-ENAs in the tailward flow of Mars we find, after im-plementing the new O-ENA instrument response [Grigoriev2007] for all observations inside the IMB from February toJune 2004 an upper limit of 5 · 104 cm−2 sr−1 s−1 for 0.4to 3.5 keV energy. This is the same result we found with asmaller data sample before [Galli et al., 2006b].

Figure 13 shows the best candidate for a real O-ENA sig-nal. The measurement was obtained in BINNING mode,where the TOF bins 10 to 15 correspond to a TOF rangeof 486 ns to 1900 ns. If the ENA is a hydrogen atom thisTOF range corresponds to energies below 0.1 keV. SinceNPD is insensitive to H-ENAs below 0.1 keV, any signalregistered in the TOF bins 10 to 15 has to be caused byUV photons or by O-ENAs (the last of the 16 TOF binsin Figure 13 would correspond to O-ENA energies around0.4 keV). UV light, however, is known − from calibrationsin the laboratory and from stellar Lyman-alpha observationsduring cruise phase − to produce a flat background in theTOF spectrum (see the TOF response in Figure 2). Fig-ure 13 shows a clear bulge in the TOF spectrum at O-ENAenergies around 1 keV. But this is the only occasion wheresuch a feature is statistically significant against the flat UVbackground (estimated here by the count rate in the verylast TOF bin) by a 3-sigma confidence interval. The cor-responding O-ENA intensity amounts to only (3 ± 2) · 104

cm−2 sr−1 s−1.

original TOF spectrum

possible oxygen ENA peak

Figure 13. Original TOF spectrum measured withchannel NPD2 0 on 18 May 2004, 22:18–22:22 UT at thenightside of Mars. This is the only NPD measurementwith a significant TOF peak at O-ENA energies 0.4 to3.5 keV (the five TOF bins marked with red). The bulgeabove the dashed line (estimate of the UV background)corresponds to an ENA intensity of 3 ·104 cm−2 sr−1 s−1.

X - 10 GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS

As shown in Galli et al. [2006b] oxygen ENA intensitiesof a few 104 cm−2 sr−1 s−1 impose an upper limit on theglobal production rate of O-ENAs of the order of 1022 s−1.

7. DiscussionThe present plasma models, combined with the neutral

hydrogen and oxygen exosphere as observed during MarsExpress, reproduce the observed hydrogen ENA outflow atthe nightside within a factor of two. The tailward ENA flowseems to consist mainly of shocked solar wind ENAs. Thenon-observation of oxygen ENAs is consistent with recentexosphere models as well. However, the outflow of ENAs atthe limb and the terminator region far away from the Sundirection is much broader than predicted by current plasmamodels. The ENA images are a challenge for future modelefforts. At present, our conjecture is that planetary protonsand ENA limb emission due to proton precipitation into theatmosphere have to be included.

We have tested various hydrogen exobase densities, andour preferred value is 1 · 1011 m−3 within a factor of two.As long as new ion populations have to be included in themodels to reproduce the ENA images we cannot be more ac-curate about the neutral hydrogen densities. But we confirmthe earlier findings [Galli et al., 2006c; Chaufray et al., 2008]that the neutral hydrogen exosphere is significantly thinnerat low solar activity than theoretically predicted [Krasnopol-sky 2002]. For oxygen we only can give upper limits, andthe other chemical constituents of the Martian exosphere arenot constrained by NPD measurements.

For oxygen ENAs around Mars we find for all observa-tions inside the IMB from February to June 2004 an upperlimit of 5 · 104 cm−2 sr−1 s−1 in the energy range from 0.4to 3.5 keV energy. This imposes on the global productionrate of O-ENAs an upper limit of

QO−ENA = 1022 to 1023 s−1 (5)

This value is consistent with the model prediction byChaufray et al. [2007] who predict QO−ENA = 4 ·1022 s−1 ofescaping O-ENAs for solar minimum condition. This givesus confidence that the recent oxygen exosphere models areapproaching reality. Charge exchange reactions in generalare not important for the atmospheric loss of oxygen at lowsolar activity [Barabash et al., 2007a].

We cannot compute integral fluxes (let alone total H-ENAproduction rates) directly from the NPD images becauselarge areas, including the maximum outflow regions, werenot covered with NPD. The global H-ENA production pre-dicted by the default MHD model is QH−ENA = 5 ·1024 s−1.This number is only 40% lower if the thermal hydrogen ex-ospheric component is reduced by a factor of two and theoxygen density is reduced by an order of magnitude. This,once more, illustrates that exospheric oxygen is not impor-tant for H-ENA production. On the other hand, the trueH-ENA production PH−ENA may be larger but does not ex-ceed 1 · 1025 s−1. This upper limit is reached if a neutralhydrogen exobase density of 2 · 1011 m−3 [Chaufray et al.,2008] is assumed for all solar zenith angles. The hydrogenENA sources omitted from the MHD model do not con-tribute much to the total ENA production rate: The ENAalbedo of the Martian dayside [Futaana et al., 2006a] ac-counts for QH−ENA = 1 · 1024 s−1. The tailward ENA flowfar away from the Sun direction, which has been shown to bebeyond the capabilities of the plasma models, is even moreinsignificant. Assuming JENA = 5 · 104 cm−2 sr−1 s−1, anaperture angle of 30◦ (equal to the IMB angle at the termi-nator) and a production region of roughly A = π(RM/2)2 weestimate for the additional tailward ENA flow a productionrate of only:

Q = 2AJENA

Z 2π

0

dφ

Z 0.26

0

dθ sin θ cos θ < 1022 s−1 (6)

At present, we therefore estimate the total hydrogen ENAproduction of Mars to be

QH−ENA =

µ6

+4−2

∂× 1024 s−1 (7)

The higher production rate derived by Gunell et al. [2005] re-sulted from assuming too dense exospheric hydrogen densi-ties at Mars [Krasnopolsky, 2002]. For Venus, a recent com-parison of the ASPERA-4/NPD measurements to a Venu-sian MHD model yields a total H-ENA production rate of6 · 1024 s−1 as well [Galli et al., 2008]. We conclude thatthe total H-ENA production rates of Mars and Venus aresimilar for low solar activity.

The loss rate of planetary hydrogen due to charge ex-change reactions, QH,CX, has to be equal or smaller thanEquation 7 but the lower limit is difficult to derive fromthe NPD observations. Even if we can prove that planetaryENAs contribute to the NPD images, their contribution tothe total loss is insignificant. Exospheric, i.e., planetary, hy-drogen is ionised by charge exchange with a solar wind pro-tons, and is picked up by the magnetosheath plasma flowand carried away from the planet to eventually become partof the solar wind. We try to constrain QH,CX with NPDmeasurements nonetheless, because IMA is not able to esti-mate the loss rate of planetary H+. Most of the hydrogenENAs produced in the MHD model result from the chargeexchange between solar wind protons and planetary hydro-gen. The production rate of planetary H+ thus is not muchsmaller than the 5 ·1024 s−1 calculated from that model. Allof these ions are produced outside the exobase, and the frac-tion that is produced outside the IMB will be picked up bythe magnetosheath plasma and thus removed. This meansthat Equation 7 is still the only quantitative estimate for thetotal hydrogen loss due to charge exchange. Taking all theseuncertainties into account the ENA observations combinedwith the ENA model predictions suggest a total escape rateof planetary hydrogen, neutral and ionized, due to chargeexchange reactions of the order of

QH,CX = 1024 s−1 (8)

The ASPERA-3 measurements show the following about thetotal loss rates of hydrogen and oxygen:

The Mars exosphere is much thinner than previouslymodeled for low solar activity. Whereas the bulk, i.e., ther-mal, exosphere temperature is 200 K (and this varies onlybetween 180 K and 230 K over solar cycle according to Licht-enegger et al. [2006]) there is a population of extremely hothydrogen that is visible in the ENA tailward flow and thatneeds to be taken into account when discussing atmosphericerosion. We have also been able to estimate the total QH,CX;QO,CX = 1023 s−1 has been measured with IMA [Barabashet al., 2007a]. ASPERA-3 measurements have shown in gen-eral that charge exchange reactions are not important for theatmospheric erosion of hydrogen or oxygen at low solar ac-tivity. The situation may change for the oxygen loss at highsolar activity [Lundin et al., 1989]. Even so, the ASPERA-3observations are all consistent with a global hydrogen lossof 1026 s−1 (dominated by Jeans escape) and a global lossof oxygen of 1025 s−1 (dominated by dissociative recombi-nation) for any solar activity.

GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS X - 11

ENA production on its own is not important as an atmo-spheric loss process but it allows for a quick, global overviewon the interaction of the solar wind with planetary atmo-spheres. The ENA sensors flying on Mars Express and VenusExpress allow us to perform quantitative research within afactor of two; the ENA images derived from NPD data willforce theorists to refine the existing models of plasma andneutral exospheres of the non-magnetized planets.

Acknowledgments. We thank Y. Ma and A.F. Nagy forproviding the ion fluxes from their MHD model that we used tocompute the ENA fluxes. This work is supported by the SwissNational Science Foundation.

ReferencesBarabash, S., R. Lundin, H. Andersson, K. Brinkfeldt, A. Grig-

oriev, H. Gunell, M. Holmstrom, M. Yamauchi, K. Asamura,P. Bochsler, P. Wurz, A.J. Coates, D.R. Linder, D.O. Kataria,C.C. Curtis, K.C. Hsieh, B.R. Sandel, R.A. Frahm, J.R. Shar-ber, J.D. Winningham, M. Grande, E. Kallio, H.E.J. Koski-nen, P. Riihela, W. Schmidt, T. Sales, J. Kozyra, N. Krupp, J.Woch, S. Livi, J. Luhmann, S. McKenna-Lawlor, E. Roelof,D.J. Williams, J.-A. Sauvaud, A. Fedorov, and J.-J. Tho-caven, (2006), The analyzer of space plasmas and energeticatoms (ASPERA-3) for the Mars Express mission, Space Sci-ence Rev., 126, 113–164.

Barabash, S., A. Fedorov, R. Lundin, and J. A. Sauvaud (2007a),Martian atmospheric erosion rates, Science, 315, 501.

Barabash, S., J.-A. Sauvaud, H. Gunell, H. Andersson, A. Grig-oriev, K. Brinkfeldt, M. Holmstrom, R. Lundin, M. Yamauchi,K. Asamura, W. Baumjohann, T. Zhang, A.J. Coates, D.R.Linder, D.O. Kataria, C.C. Curtis, K.C. Hsieh, B.R. Sandel, A.Fedorov, C. Mazelle, J.-J. Thocaven, M. Grande, H.E.J. Kosk-inen, E. Kallio, T. Sales, P. Riihela, J. Kozyra, N. Krupp, J.Woch, J. Luhmann, S. McKenna-Lawlor, S. Orsini, R. Cerulli-Irelli, A. Mura, M. Milillo, M. Maggi, E. Roelof, P. Brandt,C.T. Russel, K. Szego, J.D. Winningham, R.A. Frahm, J.Scherrer, J.R. Sharber, P. Wurz, and P. Bochsler (2007c),The analyzer of space plasmas and energetic atoms (ASPERA-4) for the Venus Express mission, Planet. Space Science, 55,1772–1792.

Boßwetter, A., T. Bagdonat, U. Motschmann, and K. Sauer(2004), Plasma boundaries at Mars: a 3-D simulation study,Annales Geophysicae, 22, 4363.

Carlsson, E., A. Fedorov, S. Barabash, E. Budnik, A. Grigoriev,H. Gunnel, H. Nilsson, J.-A. Sauvaud, R. Lundin, Y. Futaana,M. Holmstrom, H. Andersson, M. Yamauchi, J.D. Winning-ham, R.A. Frahm, J.R. Sharber, J. Scherrer, A.J. Coates,D.R. Linder, D.O. Kataria, E. Kallio, H. Koskinen, T. Sles,P. Riihela, W. Schmidt, J. Kozyra, J. Luhmann, E. Roelof,D. Williams, S. Livi, C.C. Curtis, K.C. Hsieh, B.R. Sandel,M. Grande, M. Carter, J.-J. Thocaven, S. McKenna-Lawlor,S. Orsini, R. Cerulli-Irelli, M. Maggi, P. Wurz, P. Bochsler,N. Krupp, J. Woch, M. Franz, K. Asamura, and C. Dierker,(2006), Mass composition of the escaping plasma at Mars,Icarus, 182, 320.

Chamberlain, J. W. and D. M. Hunten (1987), Theory of Plane-tary Atmospheres, 2nd edition, Academic Press, Orlando.

Chaufray, J.Y., R. Modolo, F. Leblanc, G. Chanteur, R.E. John-son, and J.G. Luhmann (2007), Mars solar wind interaction:formation of the Martian corona and atmospheric loss to space,JGR, 112, E09009.

Chaufray, J.Y., J.L. Bertaux, F. Leblanc, and E. Quermerais(2008), Lyman alpha detection of a hot hydrogen exosphericcomponent by SPICAM on Mars Express, Icarus, submitted.

Dubinin, E., M. Franz, J. Woch, E. Roussos, S. Barabash,R. Lundin, J.D. Winningham, R. Frahm, and M. Acuna(2006), Plasma morphology at Mars. ASPERA-3 observations,SSR, 126, 209.

Franz, M., E. Dubinin, E. Roussos, J. Woch, J.D. Winning-ham, R. Frahm, A.J. Coates, A. Fedorov, S. Barabash, andR. Lundin (2006), Plasma moments in the environment ofMars, Space Sci. Rev., 126, 165–207.

Futaana, Y., S. Barabash, A. Grigoriev, M. Holmstrom, E. Kallio,P. C:son Brandt, H. Gunell, K. Brinkfeldt, R. Lundin, H.Andersson, M. Yamauchi, S. McKenna-Lawler, J.D. Winning-ham, R.A. Frahm, J.R. Sharber, J. Scherrer, A.J. Coates, D.R.Linder, D.O. Kataria, T. Sales, P. Riihela, W. Schmidt, H.Koskinen, J. Kozyra, J. Luhmann, E. Roelof, D. Williams, S.Livi, C.C. Curtis, K.C. Hsieh, B.R. Sandel, M. Grande, M.Carter, J.-A. Sauvaud, A. Fedorov, J.-J. Thocaven, S. Orsini,R. Cerulli-Irelli, M. Maggi, P. Wurz, P. Bochsler, A. Galli,N. Krupp, J. Woch, M. Franz, K. Asamura, and C. Dierker(2006a), First ENA observations at Mars: ENA emissions fromthe Martian upper atmosphere, Icarus, 182, 424.

Futaana, Y., . Barabash, A. Grigoriev, M. Holmstrom, E. Kallio,P. C:son Brandt, H. Gunell, K. Brinkfeldt, R. Lundin, H.Andersson, M. Yamauchi, S. McKenna-Lawler, J.D. Winning-ham, R.A. Frahm, J.R. Sharber, J.R. Scherrer, A.J. Coates,D.R. Linder, D.O. Kataria, T. Sales, P. Riihela, W. Schmidt,H. Koskinen, J. Kozyra, J. Luhmann, E. Roelof, D. Williams,S. Livi, C.C. Curtis, K.C. Hsieh, B.R. Sandel, M. Grande,M. Carter, J.-A. Sauvaud, A. Fedorov, J.-J. Thocaven, S.Orsini, R. Cerulli-Irelli, M. Maggi, P. Wurz, P. Bochsler,N. Krupp, J. Woch, M. Franz, K. Asamura, and C. Dierker(2006b), First ENA observations at Mars: Subsolar ENA jetat Mars, Icarus, 182, 413.

Futaana, Y., S. Barabash, A. Grigoriev, J.D. Winningham,R. Frahm, M. Yamauchi, and R. Lundin (2006c), Global re-sponse of Martian plasma environment to an interplanetarystructure: from ENA and plasma observations at Mars, SSR,126, 315.

Galli, A., P. Wurz, S. Barabash, A. Grigoriev, R. Lundin, Y. Fu-taana, H. Gunell, M. Holmstrom, E.C. Roelof, C.C. Curtis,K.C. Hsieh, A. Fedorov, J.D. Winningham, R.A. Frahm, R.Cerulli-Irelli, P. Bochsler, N. Krupp, J. Woch, and M. Franz(2006a), Direct measurements of energetic neutral hydrogen inthe interplanetary medium, ApJ, 644, 1317.

Galli, A., P. Wurz, S. Barabash, A. Grigoriev, H. Gunell,R. Lundin, M. Holmstrom, and A. Fedorov (2006b), Energetichydrogen and oxygen atoms observed on the nightside of Mars,SSR, 126, 267.

Galli, A., P. Wurz, H. Lammer, H. I. M. Lichtenegger, R. Lundin,S. Barabash, A. Grigoriev, M. Holmstrom, and H. Gunell(2006c), The hydrogen exospheric density profile measuredwith ASPERA-3/NPD, SSR, 126, 447.

Galli, A., M.-C. Fok, P. Wurz, S. Barabash, A. Grigoriev, Y. Fu-taana, M. Holmstrom, A. Ekenback, E. Kallio, and H. Gunell(2008), The tailward flow of Energetic Neutral Atoms observedat Venus, JGR, in press.

Grigoriev, A., Y. Futaana, S. Barabash, and A. Fedorov (2006),Observations of the Martian subsolar ENA jet oscillations,SSR, 126, 299.

Grigoriev, A. (2007), The Neutral Particle Detector on the Marsand Venus Express missions, PhD thesis, Institute of SpacePhysics, Kiruna.

Gunell, H., M. Holmstrom, H. K. Biernat, and N. V. Erkaev(2005), Planetary ENA imaging: Venus and a comparison withMars, P&SS, 53, 433.

Gunell, H., K. Brinkfeldt, M. Holmstrom, P. C:son Brandt, S.Barabash, E. Kallio, A. Ekenback, Y. Futaana, R. Lundin,H. Andersson, M. Yamauchi, A. Grigoriev, J.D. Winning-ham, R.A. Frahm, J.R. Sharber, J. Scherrer, A.J. Coates,D.R. Linder, D.O. Kataria, T. Sales, P. Riihela, W. Schmidt,H.E.J. Koskinen, J. Kozyra, J. Luhmann, E. Roelof, D.Williams, S. Livi, C.C. Curtis, K.C. Hsieh, B.R. Sandel,M. Grande, M. Carter, J.-A. Sauvaud, A. Fedorov, J.-J. Tho-caven, S. McKenna-Lawlor, S. Orsini, R. Cerulli-Irelli, M.Maggi, P. Wurz, P. Bochsler, N. Krupp, J. Woch, M. Franz,K. Asamura, and C. Dierker (2006a), First ENA observationsat Mars: Charge exchange ENAs produced in the magne-tosheath, Icarus, 182, 2, 431–438.

Gunell, H., M. Holmstrom, S. Barabash, E. Kallio, P. Janhunen,A. F. Nagy, and Y. Ma (2006b), Planetary ENA imaging: Ef-fects of different interaction models for Mars, P&SS, 54, 117.

Kallio, E., J.G. Luhmann, and S. Barabash (1997), Charge ex-change near Mars: the solar wind absorption and energeticneutral atom production, JGR, 102, 22183.

Kallio, E. and P. Janhunen (2001), Atmospheric effects of protonprecipitation in the Martian atmosphere and its connection tothe Mars-solar wind interaction, JGR, 106, 5617.

X - 12 GALLI ET AL.: THE TAILWARD FLOW OF ENAS AT MARS

Krasnopolsky, V.A. (2002), Mars’ upper atmosphere and iono-sphere at low, medium, and high solar activities: Implicationsfor evolution of water, JGR, 107, 5128.

Lichtenegger, H.I.M., H. Lammer, and W. Stumptner (2002), En-ergetic neutral atoms at Mars, 3. Flux and energy distributionsof planetary energetic H atoms, JGR, 107, 1279.

Lichtenegger, H.I.M., H. Lammer, Y.N. Kulikov, S. Kazemine-jad, G.H. Molina-Cuberos, R. Rodrigo, B. Kazeminejad, andG. Kirchengast (2006), Effects of low energetic neutral atomson Martian and Venusian dayside exospheric temperature es-timations, SSR, 126, 469.

Lindsay, B.G. and R.F. Stebbings (2005), Charge transfer crosssections for energetic neutral atom data analysis, JGR, 110,A12213.

Lundin, R., H. Borg, B. Hultqvist, A. Zakharov, and R. Pelli-nen,(1989), First measurements of the ionospheric plasma es-cape from Mars, Nature, 341, 609.

Ma, Y., A.F. Nagy, K.C. Hansen, D.L. DeZeeuw, and T.I. Gom-bosi (2002), Three-dimensional multispecies MHD studies ofthe solar wind interaction with Mars in the presence of crustalfields, JGR, 107, 1282.

Mura, A., S. Orsini, A. Milillo, E. Kallio, A. Galli,S. Barabash, P. Wurz, A. Grigoriev, Y. Futaana, M. Holm-strom, H. Andersson, R. Lundin, M. Yamauchi, M. Franz,N. Krupp, J. Woch, K. Asamura, A.J. Coates, C.C. Curtis,K.C. Hsieh, B.R. Sandel, A. Fedorov, M. Grande, H. Koskinen,

J. Kozyra, J. Luhmann, S. McKenna-Lawlor, R. Cerulli-Irelli,R. D’Amicis, M. Maggi, S. Massetti, E. Roelof, P.C. Brandt,K. Szego, D. Winningham, R. Frahm, and J. Sharber, (2007),ENA detection in the dayside of Mars: ASPERA-3 NPD sta-tistical study, Planet. Space Sci. 56, 840–845.

Nilsson, H., E. Carlsson, H. Gunell, Y. Futaana, S. Barabash,R. Lundin, A. Fedorov, Y. Soobiah, A. Coates, M. Franz, andE. Roussos, Investigation of the Influence of Magnetic Anoma-lies on Ion Distributions at Mars, Space Science Rev., (2006)126, 355–372.

Wurz, P., A. Galli, S. Barabash, and A. Grigoriev (2008), En-ergetic Neutral Atoms from the Heliosheath, Astrophys. Jou.,(2008) 683, 248–254.

A. Galli, P. Wurz, Physikalisches Institut, University of Bern,Sidlerstrasse 5, 3012 Bern, Switzerland. (galli@space.unibe.ch)

E. Kallio, Finnish Meteorological Institute, P.O. Box 503,00101 Helsinki, Finland.

A. Ekenback, M. Holmstrom, S. Barabash, A. Grigoriev, andY. Futaana, Institutet for Rymdfysik, P.O. Box 812, 98128Kiruna, Sweden.

M.-C. Fok, Geospace Physics Laboratory, NASA GoddardSpace Flight Center, Code 673 Greenbelt, Maryland 20771, USA.

H. Gunell, Department of Physics, West Virginia University,P.O. Box 6315, Morgantown, West Virginia 26506, USA.