Rosetta-Alice Observations of Exospheric Hydrogen and Oxygen on Mars
Paul D. Feldman, Andrew J. Steffl, Joel Wm. Parker, Michael F. A'Hearn, Jean-Loup Bertaux, S. Alan Stern, Harold A. Weaver, David C. Slater, Maarten Versteeg, Henry B. Throop, Nathaniel J. Cunningham, Lori M. Feaga
aa r X i v : . [ a s t r o - ph . E P ] J un Rosetta-Alice Observations of Exospheric Hydrogenand Oxygen on Mars
Paul D. Feldman a, ∗ , Andrew J. Steffl b , Joel Wm. Parker b , Michael F. A’Hearn c ,Jean-Loup Bertaux d , S. Alan Stern b , Harold A. Weaver e , David C. Slater f ,Maarten Versteeg f , Henry B. Throop b , Nathaniel J. Cunningham g , Lori M.Feaga c a Johns Hopkins University, Department of Physics and Astronomy, 3100 N. Charles Street,Baltimore, MD 21218 USA b Southwest Research Institute, Department of Space Studies, Suite 300, 1050 Walnut Street,Boulder CO 80302-5150 USA c Department of Astronomy, University of Maryland, College Park MD 20742-2421 USA d LATMOS, CNRS/UVSQ/IPSL, 11 Boulevard d’Alembert, 78280 Guyancourt, France e Johns Hopkins University Applied Physics Laboratory, Space Department, 11100 JohnsHopkins Road, Laurel, MD 20723-6099 USA f Southwest Research Institute, P. O. Drawer 28510, San Antonio TX 78228-0510 USA g Physics Department, Nebraska Wesleyan University, Lincoln, NE 68504-2794 USA
Abstract
The European Space Agency’s
Rosetta spacecraft, en route to a 2014 en-counter with comet 67P/Churyumov-Gerasimenko, made a gravity assist swing-by of Mars on 25 February 2007, closest approach being at 01:54 UT. The Aliceinstrument on board
Rosetta , a lightweight far-ultraviolet imaging spectrographoptimized for in situ cometary spectroscopy in the 750–2000 ˚A spectral band, wasused to study the daytime Mars upper atmosphere including emissions from exo-spheric hydrogen and oxygen. Offset pointing, obtained five hours before closestapproach, enabled us to detect and map the HI Lyman- a and Lyman- b emissionsfrom exospheric hydrogen out beyond 30,000 km from the planet’s center. Thesedata are fit with a Chamberlain exospheric model from which we derive the hy-drogen density at the 200 km exobase and the H escape flux. The results arecomparable to those found from the the Ultraviolet Spectrometer experiment onthe Mariner 6 and fly-bys of Mars in 1969. Atomic oxygen emission at 1304 ˚Ais detected at altitudes of 400 to 1000 km above the limb during limb scans shortlyafter closest approach. However, the derived oxygen scale height is not consistentwith recent models of oxygen escape based on the production of suprathermal1xygen atoms by the dissociative recombination of O + . Keywords:
Mars, Mars atmosphere, Atmospheres, evolution
1. Introduction
The extended atomic hydrogen corona of Mars was first detected by the ultra-violet spectrometer experiments on the
Mariner 6 and spacecraft that measuredresonantly scattered solar Lyman- a radiation (Barth et al., 1971) to a planetocen-tric distance of 24,000 km. This was followed by a similar experiment on the or-biting Mariner 9 mission (Barth et al., 1972). Anderson and Hord (1971), usingradiative transfer theory, analyzed the early data to derive a hydrogen escape rate,which they found to be compatible with the water photodissociation rate at Mars.Barth et al. (1972) recognized that the apparently constant H escape rate derivedfrom
Mariner 6, 7 , and , if extended backward over geological time scales, wouldresult in an oxygen abundance in the lower atmosphere several orders of magni-tude larger than observed. McElroy (1972) suggested that dissociative recombi-nation of O + , the dominant ion in the atmosphere, would produce oxygen atomswith sufficient energy to escape. However, since the energetic oxygen atoms aremostly produced below the exobase, determination of the escaping fraction re-quires detailed modeling, represented by the recent work of Lammer et al. (2003),Fox and Ha´c (2009), Shematovich et al. (2007), Valeille et al. (2009, 2010), andothers. Barth et al. (1971) also reported observations of O I l Mariner 9 , were interpreted byStrickland et al. (1973) in terms of a cool, optically thick oxygen exosphere. Sincethen, the only measurement of exospheric oxygen on Mars is from the SPICAMinstrument on
Mars Express (Chaufray et al., 2009), but those data extended onlyup to 400 km where radiative transfer effects in the O I l I Lyman- a emission at 1216 ˚A, but only up to 4,000 km.Again, they find that a two temperature H exosphere is possible, although the ∗ Corresponding author
Email address: [email protected] (Paul D. Feldman)
Preprint submitted to Icarus November 8, 2018 esult is inconclusive. Clarke et al. (2009) also suggest a two-component H distri-bution from monochromatic Lyman- a images of Mars showing the H corona outto 4 Mars radii ( R M ) taken by the Solar Blind Channel of the Advanced Camerafor Surveys on HST . Radiative transfer modeling is necessary to extract densitiesfrom both of these data sets. Clarke et al. also noted a variation in the Lyman- a brightness over a period of two months which they ascribe to a seasonal variationin the H O loss rate.We report here on observations of both the extended hydrogen and oxygencoronae of Mars made with the Alice far-ultraviolet imaging spectrograph (Stern et al.,2007) on
Rosetta during the spacecraft’s gravity assist swing-by of Mars on 25February 2007. Offset exposures enabled us to detect and map the H I Lyman- a and Lyman- b emissions to beyond 30,000 km from the planet’s center. Moreover,except near the planet’s limb, the Lyman- b emission is optically thin, allowing usto use a spherical Chamberlain model to determine the temperature and densityof H at the exobase without the need for radiative transfer modeling. From limbpointings at spacecraft distances closer to the planet, oxygen emission above theexobase is detected, allowing us to constrain the density of hot oxygen withoutthe need for radiative transfer modeling. These data can be used to derive atomicescape rates and address the question of stoichiometric loss of water vapor fromMars.
2. Observations
Rosetta approached Mars from the day side making its closest approach (CA)at 01:54 UT on 25 February 2007 at an altitude of 250 km. In order to satisfythe scientific goals of the various remote sensing instruments (see, e.g., Coradini,2010), the common instrument boresight was programmed for a number of fixedpointings towards both the sunlit and dark hemispheres of Mars and offset fromMars, as well as raster scans across the sunlit limb. Due to operational constraints,observations were not possible during the immediate CA period. The pointings ofinterest in this paper (denoted by the operational designation ALxx) were AL03pre-CA, centered on the sunlit disk, AL10E, offset pointings centered 2.5 ◦ and7.5 ◦ from Mars along the equator, and AL11B, scans across the illuminated cres-cent post-CA. Observation start times and geometry parameters are given in Ta-ble 1. At the time of closest approach, Mars was 1.445 AU from the Sun and theareocentric longitude, L s , was 189.9 ◦ . Solar activity was very low for an extendedtime, including when Earth faced the same solar longitude 9 days earlier, with F . ≈
72 at 1 AU. 3lice is a lightweight, low-power, imaging spectrograph optimized for in situ cometary far-ultraviolet (FUV) spectroscopy. It is designed to obtain spatially-resolved spectra in the 750-2000 ˚A spectral band with a spectral resolution be-tween 8 and and 12 ˚A for extended sources that fill its field-of-view. The slit is inthe shape of a dog bone, 5.5 ◦ long, with a width of 0.05 ◦ in the central 2.0 ◦ whilethe ends are 0.10 ◦ wide. Each spatial pixel along the slit is 0.30 ◦ . Alice employsan off-axis telescope feeding a 0.15-m normal incidence Rowland circle spectro-graph with a concave holographic reflection grating. The imaging microchannelplate detector utilizes dual solar-blind opaque photocathodes (KBr and CsI) andemploys a two-dimensional delay-line readout. Details of the instrument are givenby Stern et al. (2007).The Alice slit geometry, illustrating the shape of the multi-segment slit, isshown in Fig. 1 for the first pre-CA offset pointing of 2.5 ◦ . The second offsetmoved the slit an additional 5.0 ◦ away from Mars parallel to the Martian equator.Except near the limb, the only features seen in the offset spectra are H I Lyman- a and Lyman- b , and these data are used to extract the spatial profiles of theseemissions.A similar diagram for the post-CA limb scans, beginning 25 February 2007at UT 03:33:02, is shown in Fig. 2. At this time each spatial pixel projected to280 km in altitude but as Rosetta receded from Mars the projected size of eachpixel increased. The scan slowly shifted the boresight ∼
250 km towards Marsover a 15-minute period. This is illustrated in Fig. 3. Because of the scanningmotion, these spectra were acquired in “pixel-list” mode, that is the position ofeach photon count is recorded together with a time tag so that spectra could bereconstructed with the motion accounted for. Because the
Rosetta -Alice instru-ment has only a single data buffer, the data gaps seen in Fig. 3 result from thetime required to read out the buffer to the spacecraft. The typical time to fill thebuffer was 30 s, so in practice we accumulated individual spectra correspondingto ∼
30 s integrations. Even so, to detect O I emission at high altitudes, we needto co-add multiple spectra, as described below.
3. Data Analysis
During the gravity assist swing-bys of both Mars and the Earth, the
Rosetta instruments were powered on and operated primarily to provide flight verifica-tion of instrument performance and to acquire calibration data such as standardultraviolet star fluxes and detector flat-fields. There were also opportunities to4xercise the full range of instrument parameters that could be adjusted by remotecommand in flight in order to optimize the signal-to-noise performance of theinstrument. For the Mars swing-by, the detector high voltage level was set at–3.8 kV. Subsequent operations and analysis showed the optimum setting to be–3.9 kV and all observations beginning in the fall of 2007 were made at that volt-age. Nevertheless, by comparison of stellar standards at different voltages (and atdifferent times), we are able to transfer the current absolute flux calibration to theepoch of the Mars swing-by and this is incorporated into the current version ofthe data pipeline (version 3) with which all of the data have since been processed.The data used in this study are publicly available, both from NASA’s PlanetaryData System ( http://pdssbn.astro.umd.edu/ ) and ESA’s Planetary ScienceArchive ( ).To derive spatial profiles of the observed emissions requires an accurate flat-field calibration. This poses a rather acute problem for Lyman- a at 1216 ˚A be-cause the instrument was designed for this wavelength to fall in the gap betweenthe KBr and CsI photocathode coatings. The intent was to utilize the low detectionefficiency of the bare microchannel plate to compensate for the very high expectedLyman- a photon flux from the comet. However, because of a slight misalignmentin the coating edges, the Lyman- a sensitivity varies by a factor of two along theportion of the detector that is mapped onto the sky. In contrast, at Lyman- b thevariation across the slit is ∼ ±
20% odd-even detector roweffect that is accentuated at the lower detector high voltage of –3.8 kV.Fortuitously, following the Mars encounter,
Rosetta -Alice was pointed to-wards Jupiter and recorded many hours of spectra in support of the
New Hori-zons fly-by of Jupiter in February 2007. These observations were made with thesame instrument parameters used for the Mars observations. Since, from the or-bit of Mars the Jovian system only filled a single row of the detector, high S/Nmeasurements of detector flat-fields at Lyman- a and Lyman- b were obtained, to-gether with a measure of grating scattered Lyman- a as a function of detector row.An added bonus is that since Jupiter was only 20 ◦ away on the sky from thecoordinates of the offset pointing, these observations provided a measure of theinterplanetary Lyman- a and Lyman- b background for the Mars observations. Weused a co-added accumulation of 630,000 s of data obtained between 1 March and10 March 2007 to derive the flat-fields used in the analysis described below. We briefly discuss the dayglow spectrum of Mars, obtained under AL03 pre-CA. It consisted of four exposures, each of 1028 seconds. A composite of the5ve central rows of the sum of four exposures is shown in Fig. 4. The viewingparameters are given in Table 1. At the start of the sequence, the angular diameterof Mars was 1.62 ◦ so that the five central rows of the detector were uniformlyfilled with the illuminated Martian disk. For comparison with previous work, wealso show the Mars full disk spectrum recorded by the Hopkins Ultraviolet Tele-scope (HUT) on board the Space Shuttle in March 1995 (a full solar cycle earlier)(Feldman et al., 2000), convolved to the spectral resolution of Alice. At the time,Mars was 1.666 AU from the Sun, L s was 70.5 ◦ , and F . was ≈
75. The HUTspectrum is multiplied by a factor of 0.80 to match the brightness observed byAlice, and this is quite good agreement considering the Alice calibration uncer-tainty and the fact that the HUT spectrum measured the integrated disk brightness,not just a central stripe of the disk. The comparison also serves to validate thewavelength calibration and provides a reference spectrum with which to comparethe exospheric spectra obtained in AL10E and AL11B. As noted by Barth et al.(1972) and Leblanc et al. (2006), with the exception of the H I Lyman series andO I l I , are confined to alti-tudes below ∼
200 km.
The two offset exposures were taken immediately following the full disk spec-tra as
Rosetta approached Mars. H I Lyman- b was detected along the full lengthof the Alice slit for both offsets. Because the slit is segmented, for each row alongthe slit the observed Lyman- b signal was fit to a gaussian profile superimposedon a grating-scattered background that was fit to a second-order polynomial. Theflux was then obtained by integrating under the gaussian and then correcting forthe detector flat-field that was derived from subsequent Jupiter observations as de-scribed in Section 3.1. The result is shown as a histogram in Fig. 5. The errorsshown are statistical in the count rate. The Lyman- a profile is similarly derivedexcept that the detector background, due to dark counts, was negligible, and isshown in Fig. 6. The shape of this profile is very similar to the slant intensityprofiles derived from the Mariner 6 and fly-bys by Barth et al. (1971), althoughlower in absolute brightness as the Mariner fly-bys occurred at a time of highsolar activity. Neither Fig. 5 nor Fig. 6 have had the interplanetary backgroundsubtracted as did the plots of Barth et al. The model fits to the data are discussedbelow in Section 4.1. 6 .4. Limb scan spectra
From the data acquired in pixel list mode during the limb scans schematicallyillustrated in Fig. 3, we can extract a spectrum spanning a given time interval.However, because the line-of-sight of the spectrogram was changing its positionabove the limb with time, it is necessary to balance the motion with the needfor a sufficiently long integration time to obtain an adequate signal-to-noise ratiofor weak emission features. At the same time, we need to avoid contaminationof the spectrum by thermospheric emissions. This was done by co-adding thephoton counts from the first four exposures for row 14; the first 12 exposures forrow 15; and the last 14 exposures for rows 16 and 17. This results in an altitudeweighted average with a trapezoidal shape of ≈
320 km centered at 420, 665, 910,and 1240 km, respectively for rows 14 to 17, respectively.Examples of extracted spectra for rows 15 and 16 are shown in Fig. 7. Notethat only H I and O I emissions are detected. A possible feature at 1657 ˚A in therow 15 spectrum that could be a signature of escaping carbon atoms (Fox and Ha´c,1999; Cipriani et al., 2007), is most likely an instrumental artifact as no emissionat this wavelength appears in the row 14 spectrum. The background is due tograting scattered Lyman- a , which is variable from row to row. Also, the spectrahave not been corrected for the odd-even row variation noted in Section 3.1 whichis estimated to be ∼
25% for Lyman- b and ∼
10% for O I l
4. Discussion
For the analysis of the Lyman- b profile we follow the same procedure asAnderson and Hord (1971), using the exospheric model of Chamberlain (1963)but ignoring satellite orbits, as they can be excluded by the observed Lyman- b brightness near 30,000 km planetocentric distance. For solar minimum conditionswe take the exobase to be at 200 km and the exobase temperature, T e , to be 200 K(Krasnopolsky, 2002; Fox and Ha´c, 2009), leaving the hydrogen density at thislevel as a variable. We assume that Lyman- b is optically thin and calculate a flu-orescence efficiency (g-factor) of 3 . × − photons s − atom − at 1 AU usinga solar minimum line profile and flux from Lemaire et al. (2002). The interplane-tary Lyman- b background is fixed at 1.0 rayleigh based on the subsequent Jupiterobservations that were used to derive the detector flat-field at 1026 ˚A (see Sec-tion 3.1). All fluxes are referenced to row 15 which is the nominal Alice boresightand which is used for almost all of the stellar calibration measurements.7he result is shown by the solid curve in Fig. 5. The derived H density at200 km is 2 . × cm − and the escape flux is 7 . × cm − s − . The modelatmosphere is given in Table 2. Optical depth unity along the line-of-sight is at ∼ Mariner
Lyman- a data, we expectthe single scattering intensity to be about a factor of two higher than the radiativetransfer corrected intensity at this optical depth, and this is consistent with thedata shown in Fig. 5. Anderson and Hord, for their solar maximum observations,assuming T e =
350 K at an exobase altitude of 250 km, found an H density andescape flux of 3 . × cm − and 1 . × cm − s − , respectively. Fig. 5 alsoshows a model for 260 K (dashed line), which, normalized at 30,000 km, provideswhat superficially appears to be a better fit to the observed profile. While it isdifficult to choose between the models for planetocentric distances greater than10,000 km, the absence of a strong optical depth effect in the latter suggests thatthe lower temperature model, corresponding to a typical exospheric temperatureat solar minimum, is probably correct. For 260 K, the H density and escape fluxare 9 . × cm − and 1 . × cm − s − , respectively.The same models, applied to Lyman- a using a Lyman- a /Lyman- b ratio of 250derived from the IPM measurements, are shown in Fig. 6. The agreement is excel-lent and the deviation from the optically thin emission is consistent with an opticaldepth along the line-of-sight of 1 at 10,000 km planetocentric distance. There isno apparent need for a suprathermal H component as suggested by Chaufray et al.(2008) and Clarke et al. (2009).An interesting measurement of exospheric Lyman- a emission from Mars Ex-press has recently been reported by Galli et al. (2006). They found that the NeutralParticle Detector of the ASPERA-3 experiment was sensitive to Lyman- a photonsand measured a signal, attributed to exospheric hydrogen, out to a tangent heightof 7,250 km above the Martian limb. Considering their large measurement uncer-tainties that include calibration, statistics, pointing, and background subtraction,their measured emission profile is in general accord with the Alice data shown inFig. 6. However, they interpret this profile in terms of an optically thin resonancescattering model and derive an apparent temperature > a emission is optically thick below 10,000 km planetocentricdistance ( ∼ .2. Two-component oxygen model For oxygen we use a two-component model, again taking for the cold com-ponent, T e , to be 200 K, and an oxygen density at 200 km of 3 . × cm − (Fox and Ha´c, 2009). The curve in Fig. 8 represents an added hot component of1200 K with an oxygen density at 200 km of 1 . × cm − (see Table 2). Thedensity decrease with altitude is considerably faster than the predictions of recentexospheric models of Chaufray et al. (2009) and Valeille et al. (2010) based ona hot atomic O source due to dissociative recombination of O + , and which arenecessary to support a stoichiometric escape of water vapor from the atmosphereof Mars. However, it has been noted that such inferences from a single viewinggeometry during a period of low solar activity can be misleading as this mecha-nism is quite sensitive to solar activity and is dependent on solar zenith angle atthe observation point.Nevertheless, the present observations raise concern about some of the as-sumptions and physical parameters used in the recent modeling of oxygen escapefrom Mars. Fox and Ha´c (2009), in their comparison of exobase and Monte Carlomodels, summarize the literature on modeling efforts from the past few decadesand conclude that “efforts to balance the escape rates in the stoichiometric propor-tion of water are premature.” Fox and Ha´c focus on a comparison of escape ratesand therefore do not compute the oxygen density profiles from their models. Suchcalculation is warranted by the present data which would allow for a determinationof the line-of-sight column densities appropriate to the Alice observations.Similar questions arise in the analogous modeling of the hot oxygen environ-ment around Venus (Gr¨oller et al., 2010). Bovino et al. (2011) discuss the needfor accurate data on energy transfer collisions between hot oxygen atoms and theneutral atmosphere (they are mainly interested in helium) which is at the core ofthe escape models. Finally, we note that Simon et al. (2009) point out that an ad-ditional constraint on the models might be provided by SPICAM measurementsof the forbidden O I l S – P) line, which is produced by both photodisso-ciation of CO and by dissociative recombination of O + .
5. Conclusion
The
Rosetta swing-by of Mars on 25 February 2007, provided the first spec-troscopic observations of exospheric hydrogen and oxygen on Mars from outsideMars’ atmosphere since the
Mariner 6 and fly-bys in 1969. The spatial distribu-tion of H I Lyman- a out to beyond 30,000 km from the planet’s center is similar to9hat found from Mariner 6 and . A Chamberlain model, with current solar mini-mum model values of exospheric temperature of 200 K and an exobase altitude of200 km, provides a good fit to the observed Lyman- b profile, although the data donot exclude temperatures up to ∼
260 K. A suprathermal component, suggested byseveral authors, is not needed to match the data. The hydrogen escape flux derivedfrom the 200 K model, 7 . × cm − s − , is comparable to that derived from theearlier measurements. The distribution of atomic oxygen, derived from 1304 ˚Aemission observed to altitudes of 1000 km above the limb, is not consistent withrecent models of oxygen escape based on the production of suprathermal oxygenatoms by dissociative recombination of O + . Acknowledgments
We thank the ESA Rosetta Science Operations Centre (RSOC) and MissionOperations Center (RMOC) teams for their expert and dedicated help in planningand executing the Alice observations of Mars. We thank Darrell Strobel for help-ful discussions. The Alice team acknowledges continuing support from NASA’sJet Propulsion Laboratory through contract 1336850 to the Southwest ResearchInstitute. The work at Johns Hopkins University was supported by a sub-contractfrom Southwest Research Institute. 10 eferences
Anderson, Jr., D.E., Hord, C.W., 1971. Mariner 6 and 7 ultraviolet spectrometerexperiment: Analysis of hydrogen Lyman-alpha data. J. Geophys. Res. 76,6666–6673.Barth, C.A., Hord, C.W., Pearce, J.B., Kelly, K.K., Anderson, G.P., Stewart, A.I.,1971. Mariner 6 and 7 ultraviolet spectrometer experiment: Upper atmospheredata. J. Geophys. Res. 76, 2213–2227.Barth, C.A., Stewart, A.I., Hord, C.W., Lane, A.L., 1972. Mariner 9 Ultravi-olet Spectrometer Experiment: Mars Airglow Spectroscopy and Variations inLyman Alpha. Icarus 17, 457–468.Bovino, S., Zhang, P., Gianturco, F.A., Dalgarno, A., Kharchenko, V., 2011. En-ergy transfer in O collisions with He isotopes and Helium escape from Mars.Geophys. Res. Lett. 38, 2203.Chamberlain, J.W., 1963. Planetary coronae and atmospheric evaporation. Planet.Space Sci. 11, 901–960.Chaufray, J.Y., Bertaux, J.L., Leblanc, F., Qu´emerais, E., 2008. Observation ofthe hydrogen corona with SPICAM on Mars Express. Icarus 195, 598–613.Chaufray, J.Y., Leblanc, F., Qu´emerais, E., Bertaux, J.L., 2009. Martian oxy-gen density at the exobase deduced from O I 130.4-nm observations by Spec-troscopy for the Investigation of the Characteristics of the Atmosphere of Marson Mars Express. J. Geophys. Res. 114, 2006.Cipriani, F., Leblanc, F., Berthelier, J.J., 2007. Martian corona: Nonthermalsources of hot heavy species. J. Geophys. Res. 112, 7001.Clarke, J.T., Bertaux, J., Chaufray, J., Gladstone, R., Quemerais, E., Wilson, J.K.,2009. HST Observations Of The Extended Hydrogen Corona Of Mars. Bull.Am. Astron. Soc. 41, 49.11 (abstract).Coradini, A., et al., 2010. Martian atmosphere as observed by VIRTIS-M onRosetta spacecraft. J. Geophys. Res. 115, 4004.Feldman, P.D., Burgh, E.B., Durrance, S.T., Davidsen, A.F., 2000. Far-UltravioletSpectroscopy of Venus and Mars at 4 ˚A Resolution with the Hopkins UltravioletTelescope on Astro-2. Astrophys. J. 538, 395–400.11ox, J.L., Ha´c, A., 1999. Velocity distributions of C atoms in CO + dissocia-tive recombination: Implications for photochemical escape of C from Mars. J.Geophys. Res. 104, 24729–24738.Fox, J.L., Ha´c, A.B., 2009. Photochemical escape of oxygen from Mars: A com-parison of the exobase approximation to a Monte Carlo method. Icarus 204,527–544.Galli, A., Wurz, P., Lammer, H., Lichtenegger, H.I.M., Lundin, R., Barabash, S.,Grigoriev, A., Holmstr¨om, M., Gunell, H., 2006. The Hydrogen ExosphericDensity Profile Measured with ASPERA-3/NPD. Space Sci. Rev. 126, 447–467.Gr¨oller, H., Shematovich, V.I., Lichtenegger, H.I.M., Lammer, H., Pfleger, M.,Kulikov, Y., Macher, W., Amerstorfer, U.V., Biernat, H.K., 2010. Venus’ atomichot oxygen environment. J. Geophys. Res. 115, 12017.Krasnopolsky, V.A., 2002. Mars’ upper atmosphere and ionosphere at low,medium, and high solar activities: Implications for evolution of water. J. Geo-phys. Res. 107, 5128.Lammer, H., Lichtenegger, H.I.M., Kolb, C., Ribas, I., Guinan, E.F., Abart, R.,Bauer, S.J., 2003. Loss of water from Mars:Implications for the oxidation ofthe soil. Icarus 165, 9–25.Leblanc, F., Chaufray, J.Y., Lilensten, J., Witasse, O., Bertaux, J., 2006. Mar-tian dayglow as seen by the SPICAM UV spectrograph on Mars Express. J.Geophys. Res. 111, E09S11.Lemaire, P., Emerich, C., Vial, J.C., Curdt, W., Sch¨uhle, U., Wilhelm, K., 2002.Variation of the full Sun hydrogen Lyman- a and b profiles with the activitycycle, in: Wilson, A. (Ed.), From Solar Min to Max: Half a Solar Cycle withSOHO, Noordwijk: ESA SP-508. pp. 219–222.McElroy, M.B., 1972. Mars: An Evolving Atmosphere. Science 175, 443–445.Shematovich, V.I., Tsvetkov, G.A., Krestyanikova, M.A., Marov, M.Y., 2007.Stochastic models of hot planetary and satellite coronas: Total water loss in theMartian atmosphere. Solar System Res. 41, 103–108.12imon, C., Witasse, O., Leblanc, F., Gronoff, G., Bertaux, J., 2009. Dayglow onMars: Kinetic modelling with SPICAM UV limb data. Planet. Space Sci. 57,1008–1021.Stern, S.A., Slater, D.C., Scherrer, J., Stone, J., Versteeg, M., A’Hearn, M.F.,Bertaux, J.L., Feldman, P.D., Festou, M.C., Parker, J.W., Siegmund, O.H.W.,2007. Alice: The Rosetta Ultraviolet Imaging Spectrograph. Space Sci. Rev.128, 507–527.Strickland, D.J., Stewart, A.I., Barth, C.A., Hord, C.W., Lane, A.L., 1973.Mariner 9 ultraviolet spectrometer experiment: Mars atomic oxygen 1304- ˚Aemission. J. Geophys. Res. 78, 4547–4559.Valeille, A., Combi, M.R., Tenishev, V., Bougher, S.W., Nagy, A.F., 2010. Astudy of suprathermal oxygen atoms in Mars upper thermosphere and exosphereover the range of limiting conditions. Icarus 206, 18–27.Valeille, A., Tenishev, V., Bougher, S.W., Combi, M.R., Nagy, A.F., 2009. Three-dimensional study of Mars upper thermosphere/ionosphere and hot oxygencorona: 1. General description and results at equinox for solar low conditions.J. Geophys. Res. 114, 11005. 13 able 1: Observation Parameters. Observation ID AL03 AL10E AL11BStart time (UT 2007) 24 Feb 18:28:14 24 Feb 20:13:14 25 Feb 03:33:02Boresight pointing disk center offset 2.5 ◦ , 7.5 ◦ W E limb scanDistance to Mars a (km) 239,800 184,200 51,600Solar elongation a ◦ ◦ ◦ Longitude of tangent point at limb b ◦ ◦ ◦ Latitude of tangent point at limb b –1.8 ◦ –4.1 ◦ –26.5 ◦ Solar zenith angle at tangent point b ◦ ◦ ◦ Data mode Histogram Histogram Pixel listTotal integration time (s) 4112 483 + 1028 a At start of observation. b At sub-observer point for AL03. 14 able 2: Model Atmosphere Densities.
Altitude H O (cold) O (hot)(km) (cm − ) (cm − ) (cm − )200 2.50 × × ×
400 1.69 × × ×
600 1.19 × ×
800 8.61 × × × × − × × × × IGURE CAPTIONS
Fig. 1. Projection of the Alice slit on the sky with the common boresight offset2.5 ◦ from the center of Mars along the equator. Orange grid lines outline theilluminated region of the disk. The shape of the multi-segment slit is shown.Detector row numbers increase from left to right with the boresight (+) in row 15.Fig. 2. Projection of the Alice slit on Mars at the beginning of the limb scansfollowing closest approach. Orange grid lines outline the illuminated crescent ofthe disk. Detector row numbers increase from right to left with the boresight (+)in row 15.Fig. 3. Projection of individual rows of the Alice slit above the Mars limb dur-ing the slow limb scan following closest approach. Initially, each spatial pixelprojected to 280 km in altitude but as Rosetta receded from Mars the projectedsize of each pixel increased. The scan also slowly shifted the boresight ∼
250 kmtowards Mars over a 15-minute period. The cross-hatched areas indicate the timeduring which photon events were accumulated while the horizontal lines show thetimes for which photon events were co-added for each row.Fig. 4. The dayglow spectrum of Mars is a composite of the five central rowsof four exposures, each of 1028 seconds. For comparison with previous work,we also show the Mars full disk spectrum recorded by the Hopkins UltravioletTelescope (HUT) at 4 ˚A spectral resolution in March 1995 (a full solar cycleearlier) (Feldman et al., 2000), convolved to the spectral resolution of Alice andmultiplied by a factor of 0.8.Fig. 5. H I Lyman- b was detected along the full length of the Alice slit for bothoffsets. The data are shown as a histogram: black, 2.5 ◦ offset, red: 7.5 ◦ offset.The errors shown are statistical in the count rate. Optically thin Chamberlainmodels without satellite orbits for T=200 K (solid line) and 260 K (dashed line)are shown, superimposed on a 1 rayleigh interplanetary background. The radiusof Mars is indicated.Fig. 6. Same as Fig. 5 for H I Lyman- a . The shape of this profile is very similar tothe slant intensity profiles derived from the Mariner 6 and fly-bys by Barth et al.(1971), although lower in absolute brightness as the Mariner fly-bys occurred ata time of high solar activity. The same models are shown, superimposed on an16nterplanetary background of 250 rayleighs.Fig. 7. Extracted limb scan spectra. For row 15 (top), the photon counts from thefirst 12 exposures (376 s, see Fig. 3) were co-added. For row 16 (bottom), the last14 exposures (476 s) were co-added. Only H I and O I emissions are detected.Fig. 8. Extracted O I l igure 1: Projection of the Alice slit on the sky with the common boresight offset 2.5 ◦ from thecenter of Mars along the equator. Orange grid lines outline the illuminated region of the disk. Theshape of the multi-segment slit is shown. Detector row numbers increase from left to right withthe boresight (+) in row 15. igure 2: Projection of the Alice slit on Mars at the beginning of the limb scans following closestapproach. Orange grid lines outline the illuminated crescent of the disk. Detector row numbersincrease from right to left with the boresight (+) in row 15. D i s t an c e abo v e M a r s li m b ( k m ) Row 14Row 13Row 15Row 16
Figure 3: Projection of individual rows of the Alice slit above the Mars limb during the slow limbscan following closest approach. Initially, each spatial pixel projected to 280 km in altitude but asRosetta receded from Mars the projected size of each pixel increased. The scan also slowly shiftedthe boresight ∼
250 km towards Mars over a 15-minute period. The cross-hatched areas indicatethe time during which photon events were accumulated while the horizontal lines show the timesfor which photon events were co-added for each row.
00 1000 1200 1400 1600 1800Wavelength (Å)0246810 B r i gh t ne ss ( R a y l e i gh s Å − ) OII Ly g Ly b +OIOI OIArI CI CICO CO OI OI CO A−X Figure 4: The dayglow spectrum of Mars is a composite of the five central rows of four exposures,each of 1028 seconds. For comparison with previous work, we also show the Mars full diskspectrum recorded by the Hopkins Ultraviolet Telescope (HUT) at 4 ˚A spectral resolution in March1995 (a full solar cycle earlier) (Feldman et al., 2000), convolved to the spectral resolution of Aliceand multiplied by a factor of 0.8. km)110 B r i gh t ne ss (r a y l e i gh s ) R Mars
Lyman− b Figure 5: H I Lyman- b was detected along the full length of the Alice slit for both offsets. Thedata are shown as a histogram: black, 2.5 ◦ offset, red: 7.5 ◦ offset. The errors shown are statisticalin the count rate. Optically thin Chamberlain models without satellite orbits for T=200 K (solidline) and 260 K (dashed line) are shown, superimposed on a 1 rayleigh interplanetary background.The radius of Mars is indicated. km)100100010000 B r i gh t ne ss (r a y l e i gh s ) R Mars
Lyman− a Figure 6: Same as Fig. 5 for H I Lyman- a . The shape of this profile is very similar to the slantintensity profiles derived from the Mariner 6 and fly-bys by Barth et al. (1971), although lowerin absolute brightness as the Mariner fly-bys occurred at a time of high solar activity. The samemodels are shown, superimposed on an interplanetary background of 250 rayleighs.
000 1200 1400 1600Wavelength (Å)012345 B r i gh t ne ss ( R a y l e i gh s Å − ) row 15 Ly g Ly b OI B r i gh t ne ss ( R a y l e i gh s Å − ) row 16 Ly g Ly b OI Figure 7: Extracted limb scan spectra. For row 15 (top), the photon counts from the first 12exposures (376 s, see Fig. 3) were co-added. For row 16 (bottom), the last 14 exposures (476 s)were co-added. Only H I and O I emissions are detected.
10 100 1000OI 1304 Brightness (Rayleighs)200400600800100012001400 A l t i t ude ( k m ) Figure 8: Extracted O I l1304 brightness from rows 14–17 as a function of altitude above theMartian limb. The vertical bars illustrate the extent of the trapezoidal altitude weighting functionfor each row while the horizontal bars are the statistical uncertainty in the count rate. The two-component oxygen model shown is described in the text.