A&A 580, L10 (2015) Astronomy

DOI: 10.1051/0004-6361/201526377 &

c ESO 2015
Astrophysics
L   E

First detection of the 63 µm atomic oxygen line

in the thermosphere of Mars with GREAT/SOFIA?
L. Rezac1 , P. Hartogh1 , R. Güsten2 , H. Wiesemeyer2 , H.-W. Hübers3,4 , C. Jarchow1 , H. Richter3 ,
B. Klein2,6 , and N. Honingh5

1
Max-Planck-Institut für Sonnensystemforschung, Justus-von-Liebig-Weg 3, 37077 Göttingen, Germany
e-mail: rezac@mps.mpg.de
2
Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
3
Deutsches Zentrum für Luft- und Raumfahrt e.V., Institute of Optical Sensor Systems, Rutherfordstr. 2, 12489 Berlin, Germany
4
Humboldt-Universität zu Berlin, Department of Physics, Newtonstr. 15, 12489 Berlin, Germany
5
Physikalisches Institute der Universität zu Köln, Zülpicher Str. 77, 50937 Köln, Germany
6
University of Applied Sciences Bonn-Rhein-Sieg, Grantham-Allee 20, 53757, Sankt Augustin, Germany

Received 21 April 2015 / Accepted 27 July 2015

ABSTRACT

Context. The Stratospheric Observatory for Infrared Astronomy (SOFIA) with its 2.5 m telescope provides new science opportunities
for spectroscopic observations of planetary atmospheres in the far-infrared wavelength range.
Aims. This paper presents first results from the 14 May, 2014 observing campaign of the Martian atmosphere at 4.7 THz using the
German REceiver for Astronomy at Terahertz frequencies (GREAT) instrument.
Methods. The atomic oxygen 63 µm transition, OI, was detected in absorption against the Mars continuum, with a high signal-to-noise
ratio (∼35). A beam-averaged atomic oxygen from a global circulation model was used as input to the radiative transfer simulations
of the observed line area and to obtain a new estimate on the column density using a grid-search method.
Results. Minimizing differences between the calculated and observed line intensities in the least-square sense yields an atomic oxygen
column density of (1.1 ± 0.2) × 1017 cm−2 . This value is about twice as low as predicted by a modern photochemical model of Mars.
The radiative transfer simulations indicate that the line forms in the upper atmospheric region over a rather extended altitude region
of 70–120 km.
Conclusions. For the first time, a far-infrared transition of the atomic oxygen line was detected in the atmosphere of Mars. The
absorption depth provides an estimate on the column density, and this measurement provides additional means to constrain the
photochemical models in global circulation models and airglow studies. The lack of other means for monitoring the atomic oxygen in
the Martian upper atmosphere makes future observations with the SOFIA observatory highly desirable.
Key words. planets and satellites: atmospheres – planets and satellites: individual: Mars – planets and satellites: detection –
line: profiles

1. Introduction the considerable local time variation in atomic O. These mea-

surements were consistent with afternoon atomic O density be-
Atomic oxygen is the key element in several processes govern- ing higher than in the morning observations, a behavior not yet
ing the energy and mass flow in the mesosphere-thermosphere- reproduced by the models (Valeille et al. 2010). Since atomic
ionosphere of Mars. It controls the radiative cooling from the O has a multitude of electronic transitions, different spectral re-
CO2 15 µm bands in the Martian thermosphere (Bougher & gions may be used in estimating the O abundances, for example,
Roble 1991). Atomic oxygen also plays a significant role in es- 557 and 663 nm (Krasnopolsky & Krysko 1976; Krasnopolsky
tablishing the local exobase height (varying from 155–190 km & Feldman 2002). Oxygen abundances were also estimated from
between night and day), thereby affecting the ion production, the CO+2 number density (Stewart et al. 1992) and from ion data
plasma heating, and ultimately the atmospheric erosion through (Hanson et al. 1977). The more recent Mars Express mission
escape processes (Leblanc & Johnson 2002; Valeille et al. 2010). (2003) carries a UV spectrometer (SPICAM) that also measures
Despite the recognized need for accurate knowledge of the O(130.4 nm) electronic transition (Bertaux et al. 2006), but
atomic O abundances in the upper atmosphere, there have been to our knowledge, the O density has not yet been derived below
only a few dedicated measurements. The analysis of UV airglow the exobase (Chaufray et al. 2009). A powerful imaging spec-
measurements at 130.4 nm from the Mariner 6, 7, and 9 fly- trograph (UVIS) onboard the MAVEN spacecraft, which arrived
bys (1969, 1971) provided the first information on the relative at Mars in September 2014, also aims at obtaining the atomic
abundance of atomic O (Barth et al. 1972). Stewart et al. (1992) O density from the UV airglow at 130.4 nm (McClintock et al.
reanalyzed the Mariner 9 measurements and compared the re- 2014). The mass spectrometer NGIMS on the same satellite may
sults with a thermospheric general circulation model to explain obtain O density (Mahaffy et al. 2014) down to ≈150 km. While
?
Appendix A is available in electronic form at sensing the UV emissions from neutral and/or ionic species is
http://www.aanda.org very suitable for probing the upper atmosphere state, extracting

Article published by EDP Sciences L10, page 1 of 5

 A&A 580, L10 (2015)

the reliable information on the neutral atmosphere (especially Table 1. Mean observing conditions of Mars, May 14, 2014.
the atomic oxygen) is a challenging task. Complex modeling is
needed because of the various energy sources and chemical path- Condition Value
ways that generate and/or contaminate (scattering, dust opacity) Heliocentric distance 1.59 AU
the observed UV emissions (Chaufray et al. 2009; Huestis et al. Geocentric distance 0.70 AU
2010). To date, the only in situ (although not direct) determi- Apparent diameter 13.4500
nation of the atomic O density has been made from the data of Geocentric velocity 8.31 km s−1
the neutral mass spectrometer taken during the descent of the Phase angle 26◦
Viking 1 and 2 landers in the Martian atmosphere. These data Ls 130◦
revealed that CO2 is the main component of the Martian atmo- ∆RA beam 1.200
sphere below ∼180 km (with the homopause at ∼125 km) and ∆Dec beam 0.200
that atomic O is the most abundant among the odd-oxygen com- Mean T surf (269 ± 15) K
Mean psurf (6.64 ± 0.84) mbar
pounds (McElroy et al. 1976; Nier & McElroy 1977).
The electronic ground state of atomic oxygen, O(3 P), is Notes. Mean surface pressure and temperature are from
a triplet state with transition frequencies falling into the far- the v5.2 Mars climate database averaged over the field of view
infrared spectrum. The 63 µm, (3 P1 –3 P2 ), transition has been (http://www-mars.lmd.jussieu.fr). Standard deviation is pro-
observed in astrophysical contexts, such as molecular clouds and vided to indicate the modeled variability within the sampled beam.
star-forming regions. However, this is only possible from satel- Geocentric velocity from SOFIA’s observing coordinates (42N, 123W).
lite platforms, aircraft, or balloon-mounted instruments since the The continuum emissions at 63 µm are understood to originate up
water vapor opacity in the Earth’s troposphere is completely to several wavelengths below the surface, therefore, the brightness
opaque to the far-IR photons. This transition has also been pe- temperatures are expected to be lower than the surface temperatures).
The ∆RA and ∆Dec represent the beam center pointing offset with
riodically observed in Earth’s thermosphere (Mlynczak et al.
respect to the disk center.
2004). In this paper we report for the first time on the obser-
vation of this transition in another planet and on the potential for
offset from f0 (MHz)
systematically monitoring the thermospheric atomic O on Mars
-60 -30 0 30 60
using the modern SOFIA/GREAT instrument. 130 1.05
The paper is structured as follows: Sect. 2 provides the de- 120
scription of the instrumental setup, details on the observation,
110
Main beam RJ T(K)

and data calibration. Then an analysis of the observation is pre- 1

Line/Continuum
sented to derive the atomic O column density. In the last sec- 100
tion we summarize the science results and implication for future 90
0.95
observations.
80

70
0.9
2. Observations 60 A) B)
While far-infrared observations of O2 in the Martian atmosphere 50
-15 -10 -5 0 5 10 15 -4 -2 0 2 4
were performed for the first time (Hartogh et al. 2010a) with -1 -1
Velocity (km s ) Velocity (km s )
Herschel/HIFI (de Graauw et al. 2010) as part of the HssO key
program (Hartogh et al. 2009), the atomic oxygen line at 63 µm, Fig. 1. Panel A): measured spectra centered at the telluric line demon-
[OI], observed in absorption against the continuum of Mars, strating a clear detection of the atomic O in both atmospheres. The scale
was for the first time detected using the GREAT1 instrument is in Raleigh-Jeans main beam temperature [K], single side-band cali-
(Heyminck et al. 2012) onboard the SOFIA aircraft. The obser- bration. Panel B): only the Martian line already corrected for Doppler
vations were performed on a high-altitude flight (13 km) out of velocity and plotted as line/continuum ratio. A third-degree baseline
Palmdale, California, on May 14, 2014 during good observing correction was applied to this spectrum.
conditions (PWV 11 µm, and T sys = 3684 K). The observing
conditions for the planet are summarized in Table 1.
The high-resolution spectrometer GREAT for the first time prominent, narrow terrestrial OI line measured simultaneously in
operated its H-channel, tuned to the ground-state transition of our spectrum (only shifted 132 MHz from the martian OI line)
[OI] at 4744.77749 GHz. Major technological breakthroughs en- to correct for these frequency drifts (see next section).
abled these observations: a superconducting NbN hot-electron The boresight and focus position of the GREAT instrument
bolometer (HEB) waveguide mixer (Büchel et al. 2015), pumped was determined on Mars just before the OI spectroscopy. The
by a novel quantum-cascade laser (QCL; Richter et al. 2015) as source coupling efficiency (ηs = 0.67) and beamwidth (6.600 )
local oscillator provides receiver temperatures as low as 1000 K. of the H-channel were derived from scans across the planet.
The data were processed in a new-generation fast Fourier trans- The SOFIA optical guide cameras established pointing to an ac-
form spectrometer (Klein et al. 2012), providing a 44 kHz spec- curacy of a few arcsecs. The OI spectroscopy was performed
tral resolution over a 2.5 GHz bandwidth. in double-beam chopped mode for a total of 45 min, with a
During this first technology demonstration, the QCL was not chop rate of 1 Hz and a throw of ±4000 . The Doppler correction
frequency locked, drifting by several 10 MHz with time and re- changed by only +0.005 km s−1 within the integration time.
vealing an “intrinsic” width of about 10 MHz (on the timescale The measured spectra are shown in Fig. 1 panel A, cov-
of our data dump cycle of 0.5 to 1 s). We took advantage of the ering both transitions. In panel B a zoom-in of the Martian
spectrum is shown, with an apparent linewidth of 26 MHz
1
GREAT was developed by the MPI für Radioastronomie and (symetrized FWHM). Both lines are broader than expected from
the KOSMA/Universität zu Köln, in cooperation with the MPI für pure Doppler broadening (11 MHz FWHM at 180 K), which is
Sonnensystemforschung and the DLR Institut für Planetenforschung. the dominant broadening mechanism of this transition in both

L10, page 2 of 5
 L. Rezac et al.: Atomic oxygen on Mars

atmospheres. This most likely reflects the intrinsic line width 103
(10–20 MHz) of the reference signal from the QCL that dur-
ing this experiment was not frequency stabilized to the required
precision. Slow frequency drifts were corrected against the fre-
quency of the telluric OI line seen in the normalized sky spec-
trum, but jitter on timescales shorter than our integration time
of 0.5–1 s per phase remained. The convolution of this intrinsic 102
OI and instrumental QCL line shapes was incorporated into our

Misfit χ2
modeling and is described below.

3. Data analysis 101

The residual Doppler shift of the measured lined was detected,

but we were unable to obtain a reliable estimate on the mean
winds due to the uncertainty in the estimated accuracy of the
frequency scale (±90 m/s). We also studied the asymmetric blue
wing of the line, (see Fig. 1), with synthetic forward calcula- 100
15.5 16.0 16.5 17.0
tions using realistic wind vertical profiles (which can peak in Log10([O] Column density cm−2 )
the range 100–250 m/s at 90–120 km) on Mars as provided
by the Laboratoire de Météorologie Dynamique (LMD) Mars Fig. 2. Difference between measured and calculated radiance in terms
General Circulation Model data, (Forget et al. 1999), avail- of χ2 vs. [O] column density. The gray circles stacking in vertical are
able through the v5.2 Mars Climate Database (MCD). It was for different ranges of surface temperature and Gauss convolution to
simulate the LO broadening (see text for details).
found that an asymmetric line profile can develop for an opti-
cal depth >1. However, this happens only if the maximum wind
speed is present on the line of sight. This situation is unlikely
to occur, given the nearly disk-centered beam of our observing The best-fit atomic oxygen column density estimate was ac-
geometry, including the 200 uncertainty in the absolute point- complished through a brute-force grid search minimizing the
ing. Additionally, we were unable to easily explain the overall χ2 differences between measured and calculated radiances for a
redshift of the line with the strong blue-wing asymmetry in our number of atmospheric scenarios. The atmospheric inputs were
modeling. Therefore, we consider the possibility that the asym- generated by scaling the beam-averaged atomic O VMR pro-
metric line shape is at least partially due to yet unknown instru- file derived from v5.2 MCD, as discussed previously. In total,
mental effects (most likely a feature of the QCL), and the further 60 scaling values were used, providing 30 points-per-decade in
analysis relies on the symmetrized lines shape. column density space, and then for each O VMR profile, an addi-
Modeling input parameters of the Martian atmosphere were tional 14 variants were produced by varying the surface temper-
taken from the v5.2 MCD, and the beam averaged tempera- ature between 200–265 K at 5 K steps. For each of the 840 syn-
ture, pressure and atomic O volume mixing ratio (VMR) pro- thetic radiance profiles a convolution with a Gaussian of varying
files for a given observing geometry, season, and local time width (3–21 MHz at 2 MHz step) was performed to simulate the
were calculated (see Fig. A.1). A forward line-by-line (LBL) ra- additional LO broadening. The global view of the χ2 fitting for
diative transfer model described in Jarchow & Hartogh (1995) all the different profiles described above is shown in Fig. 2. The
that was previously applied to Mars (Hartogh et al. 2010b) best-fit profiles within the χ2 ≤ 3 are shown in Fig. 3 as the
and Venus (Rengel et al. 2008) was used to calculate beam- gray envelope. The measured line with the 1σ noise error bars
averaged synthetic spectra in spherical geometry. The broaden- shows the measured spectrum, and the spectrum for the best fit
ing due to planet rotation is negligible for this transition given is plotted in red.
the subobserver latitude, rotational axis tilt, and beam center off- Figures 2 and 3 illustrate that the mean column density
set (100 from the central meridian), even though it was taken is rather well constrained by the measured line and the col-
into account in our simulations. The line shape was modeled umn densities cluster in the confined space around the (mean)
with a Voigt profile. The line center rest frequency was mea- value 1.1 × 1017 cm−2 for the best-fitted spectra. Surprisingly,
sured by Zink & Evenson (1991), and other spectroscopic line only a very narrow range of widths for the simulated Gaussian
parameters were taken from HITRAN 2012 (Rothman et al. LO broadening is obtained (11, 13, and 12) MHz for the mini-
2013). The default (no-knowledge) value for the half-width of mum, maximum, and average over the simulated spectra within
the Lorenzian component,(0.05 cm−1 /atm) was provided by the the specified 3 χ2 limit. This agrees well with a reasonable value
HITRAN database. Because there is no estimate on the actual of the LO jitter estimate. We took the spread of the retrieved
value of the collisional broadening, we performed a sensitivity column densities because the subsurface temperatures are varied
test by increasing the coefficient ten times. Even at this value, the to represent an estimate of uncertainty in the line contrast due
collisional broadening is negligible, and the Doppler broadening to T atm –T surf uncertainties. We assumed that the line strength is
regime dominates above ≈5 km. The electronic partition func- known with an accuracy better than 3% (the JPL and HITRAN
tion and stimulated emission term for scaling line intensity to at- line strengths for this transition differ by about 1%), which leads
mospheric temperature is significantly different from unity and to an additional uncertainty of ≈3% in the derived column den-
was explicitly calculated using fine-structure level energies and sity. The beam-averaged surface pressure is taken to be known
prescribed degeneracies from HITRAN 2012. Over the range within 5%, which results in 5% errors in the column density es-
of pressures where the line forms, we can assume that the lo- timate. The root-sum-square uncertainty resulting from the at-
cal thermodynamic equilibrium (LTE) applies for the transition mospheric and spectroscopic input inaccuracies is estimated to
(Sharma et al. 1994). be 18%.

L10, page 3 of 5
 A&A 580, L10 (2015)
1.02
difficult to obtain by any other method. The factor of 2 difference
Best fit
Measured
suggests that the LMD model, while not perfect, is a reasonably
1.00 accurate representation of the mean oxygen in the upper Martian
atmosphere, although a more robust validation is needed, requir-
ing additional observations for different latitudes and local times.
0.98 Future and long-term observations of this transition for differ-
ent hemispheres and local solar times are highly desirable, and
Line/Continuum

0.96
we will follow-up our results at the next Mars visibility, within
scheduling constraints of the observatory, hopefully with a better
control of the QCL.
0.94
Acknowledgements. Based on observations made with the NASA/DLR
Stratospheric Observatory for Infrared Astronomy (SOFIA). SOFIA science
0.92 mission operations are conducted jointly by the Universities Space Research
Association, Inc, under NASA contract NAS2-97001, and the Deutsches SOFIA
Institut, Universität Stuttgart under DLR Space Administration contract 50 OK
0901. SOFIA is carried out by the DLR Space Administration with funding by
0.90
the Federal Ministry of Economics and Technology based on a resolution by the
German Bundestag. The development of the QCL LO was supported under DLR
Space Administration contract 50OK1104. We gratefully acknowledge the sup-
0.88 port by the observatory staff. Part of this work was supported by the German
−45 −30 −15 0 15 30 45
Center freqeuency offset MHz DFG project number HA3261/7-1. We thank E. Lellouch for his careful reading
and helpful comments that improved the manuscript.
Fig. 3. Best-fit profile (red), all acceptable profiles within 3 χ2 (gray),
and the measured (symmetrized and centered) line (black; see text for
discussion).
References
Barth, C. A., Stewart, A. I., & Hord, C. W. 1972, Icarus, 17, 457
Bertaux, J.-L., Korablev, O., Perrier, S., et al. 2006, J. Geophys. Res., 111,
4. Summary and discussion E10S90
Bougher, S. W., & Roble, R. G. 1991, J. Geophys. Res., 96, 11045
We have presented a measurement of the atomic oxygen fine- Büchel, D., Pütz, P., Jacobs, K., et al. 2015, IEEE Trans. Terahertz Sci. Technol.,
structure transition at 63 µm in the mesosphere of Mars by the 5, 207
GREAT instrument onboard the SOFIA observatory. This tran- Chaufray, J. Y., Leblanc, F., Quémerais, E., & Bertaux, J. L. 2009, J. Geophys.
sition was observed for the first time in another terrestrial planet, Res., 114, 2006
allowing us to derive the [O] column density in the Martian up- de Graauw, T., Helmich, F. P., Phillips, T. G., et al. 2010, A&A, 518, L6
Forget, F., Hourdin, F., Fournier, R., et al. 1999, J. Geophys. Res., 104, 24155
per atmosphere. The measured line profile is asymmetric with Hanson, W. B., Sanatani, S., & Zuccaro, D. R. 1977, J. Geophys. Res., 82, 4351
an extended blue wing that is assumed to be due to instrumental Hartogh, P., Lellouch, E., Crovisier, J., et al. 2009, Planet. Space Sci., 57, 1596
effects. In addition, the frequency scale calibration uncertainty Hartogh, P., Jarchow, C., Lellouch, E., et al. 2010a, A&A, 521, L49
did not allow us to reliably obtain a mean wind speeds at this Hartogh, P., Blecka, M. I., Jarchow, C., et al. 2010b, A&A, 521, L48
Heyminck, S., Graf, U. U., Güsten, R., et al. 2012, A&A, 542, L1
time. Huestis, D. L., Slanger, T. G., Sharpee, B. D., & Fox, J. L. 2010, Faraday
The line is detected in absorption against the Mars contin- Discuss., 147, 307
uum, and our simulations indicate that it forms over a range of al- Jarchow, C., & Hartogh, P. 1995, in Global Process Monitoring and Remote
titudes, 70–120 km. This observation was performed on May 14, Sensing of Ocean and Sea Ice, Proc. SPIE, 2586, 196
2014 corresponding to the Martian Ls = 130, with the beam Klein, B., Hochgürtel, S., Krämer, I., et al. 2012, A&A, 542, L3
Krasnopolsky, V. A., & Feldman, P. D. 2002, Icarus, 160, 86
center pointing at around 25 N. Despite an instrumental line Krasnopolsky, V. A., & Krysko, A. A. 1976, Space Res., 16, 1005
broadening of the reference signal, we were able to constrain the Leblanc, F., & Johnson, R. E. 2002, J. Geophys. Res., 107, 5010
Martian oxygen column density in a brute-force inversion. The Mahaffy, P. R., Richard Hodges, R., Benna, M., et al. 2014, Space Sci. Rev., 185,
best-fit column density, (1.1 ± 0.2) × 1017 cm−2 , was found to 27
McClintock, W. E., Schneider, N. M., Holsclaw, G. M., et al. 2014, Space Sci.
be a factor of 2 below the predictions of the LMD model. Since Rev., DOI: 10.1007/s11214-014-0098-7
the atomic oxygen is the main carrier of the chemical potential McElroy, M. B., Kong, T. Y., Yung, Y. L., & Nier, A. O. 1976, Science, 194,
energy, it plays a prime role in the energy budget and chemistry 1295
of the upper atmosphere of Mars. The chemical modeling of the Medvedev, A. S., González-Galindo, F., Yigit, E., et al. 2015, J. Geophys. Res.,
formation of ozone relies on the atomic oxygen and its transport 120, 913
Mlynczak, M. G., Martin-Torres, F. J., Johnson, D. G., et al. 2004, J. Geophys.
across the nighttime terminator. In addition, the interpretation of Res., 109, 12306
the airglow measurements of electronically excited O2 relies on Nier, A. O., & McElroy, M. B. 1977, J. Geophys. Res., 82, 4341
the atomic O density because they efficiently quench the excited Rengel, M., Hartogh, P., & Jarchow, C. 2008, Planet. Space Sci., 56, 1368
states. The atomic oxygen is known to have a significant long- Richter, H., Greiner-Bär, M., Rösner, K., et al. 2015, IEEE Trans. Terahertz Sci.
Technol., submitted
and short-term variability (Hanson et al. 1977; Stewart et al. Rothman, L., Gordon, I., Babikov, Y., et al. 2013, J. Quant. Spectr. Rad. Transf.,
1992; Valeille et al. 2010; Medvedev et al. 2015), and the re- 130, 4, HITRAN2012 special issue
spective modeling can be substantially improved based on mea- Sharma, R., Zygelman, B., von Esse, F., & Dalgano, A. 1994, Geophys. Res.
surements such as are now available from the SOFIA/GREAT Lett., 21, 1731
instrument. Stewart, A. I. F., Alexander, M. J., Meier, R. R., et al. 1992, J. Geophys. Res.,
97, 91
The detection of the OI line in the Martian atmosphere Valeille, A., Bougher, S. W., Tenishev, V., Combi, M. R., & Nagy, A. F. 2010,
is an important accomplishment that will allow us to gain Icarus, 206, 28
novel scientific knowledge on this atomic species, which is very Zink, L. R., & Evenson, K. M. 1991, ApJ, 371, L85

Page 5 is available in the electronic edition of the journal at http://www.aanda.org

L10, page 4 of 5
 L. Rezac et al.: Atomic oxygen on Mars
180
Appendix A: Supporting material and figures
0 MHz
160 4 MHz
12 MHz
O [VMR]
16 MHz
1e-08 1e-07 1e-06 1e-05 0.0001 0.001 0.01 0.1 26 MHz
180 140

Altitude (km)
120
150

100
120
Altitude [km]

80
90
60
60
40
0 1e-05 2e-05 3e-05 4e-05
Jacobians d( ln(radiance) ) / d( ln([O] VMR))
30
Fig. A.2. Calculated beam-averaged radiance sensitivity to changes in
0 the O VMR (Jacobians) plotted for five frequency values shown as
100 150 200 250 300 offsets. This calculation was performed for a column density of 8 ×
Temperature [K]
1016 cm−2 , and the simulated LO smoothing with 10 MHz sigma was
Fig. A.1. Vertical profile of temperature (solid line) and atomic oxygen applied. For higher abundance (several times 1017 cm−2 ), the line center
VMR (dashed line) derived from the LMD v5.2 database that were used peak shifts up to (100–110 km), with similar relative altitude shifts at
in the radiative transfer calculations. other frequencies.

O contribution functions. The OI line is not completely with altitude, while the total density exponentially decreases,
optically thin (τ > 1) for column densities higher than resulting in the [O] number density to have a peak just near
about 9 × 1016 cm−2 , and the observed line intensity forms at the 90–100 km altitude and then smoothly decaying above. To a
altitudes below 130 km, but over a rather extended altitude re- lesser degree, the “LO” broadening of the line, which was taken
gion, as shown in Fig. A.2. The sensitivity functions peak nearly into account in these calculations, also plays a role in the de-
at the same altitude for the different frequencies. This behavior graded altitude resolution. In general, the wings of a Doppler
is mainly due to the profile of the oxygen VMR profile, which line only depend on temperature and do not provide strong alti-
extends from an insignificant value at 50 km to sharply increase tude information.

L10, page 5 of 5