Investigating trace gases in the Martian atmosphere using the ExoMars Trace
Gas Orbiter Part 2: TauREM and PSG retrievals of ESA PSA NOMAD SO
Channel Data
George H. Cann , University College London, London, UK (george.cann.15@ucl.ac.uk). Ingo Waldmann,
Dave Walton, Jan-Peter Muller, University College London, London, UK.
Introduction:
We present TauREx for Mars (TauREM) and
Planetary Spectrum Generator (PSG) Villanueva et
al. (2018) [1], Villanueva et al. (2022) [2], retrievals
using the European Space Agency's (ESA) Trace Gas
Orbiter (TGO) Nadir and Occultation for MArs Discovery (NOMAD) instrument, Solar Occultation
(SO) channel Vandaele et al. (2015) [3], as a means
of searching for H2O, HDO, O3, and 12CH4 and
13
CH4 in the Martian atmosphere.
TauREM retrieval experiments are performed,
including varying the number of live points used,
performing wavenumber shifts to the model spectrum, and targeting spectra at varying tangent
heights. Retrieval experiments, with and without applying AOTF, Blaze, and ILS functions are performed. TauREM and PSG retrieval results are interpreted and discussed.

observed and an important reality check on the solution.
Within TauREM NOMAD SO channel blaze
function, acousto-optic tunable filter (AOTF) modules have been incorporated into the forward model.
A set of geometry modules have been included in
order to calculate the atmospheric layer line-of-sight
(LOS) intersection points and corresponding path
lengths. Martian absorption cross-sections have been
generated using HITRAN 2016, Gordon et al. (2017)
[11] molecular line-lists with ExoCross, Yurchenko
et al. (2018) [12], accounting for expected Martian
pressures and temperatures. Overall, this enables
TauREM to produce simulated NOMAD SO channel
transmittance spectra. All code and data associated
with this research will be made open source to the
community.
Results: TauREM Nested Sampling Retrievals

Method:
TauREM is the Mars version of TauREx3[2], the
3rd generation of TauREx, Al-Refaie et al. (2019) [4]
(Tau Retrieval for Exoplanets) extending from earlier development work, Cann et al. (2020) [5].
TauREx3 is a fully Bayesian atmospheric retrieval
framework that uses Nested Sampling, Skilling
(2006) [6] and Markov chain Monte Carlo Haario et
al. (2006) [7] methods to sample the full likelihood
space of possible solutions. This allows TauREM to
produce marginalised and conditional posterior distributions of forward model parameters, which can
be used to map correlations between parameters.
Nested Sampling applies a general Monte Carlo
analysis to periodically constrain ellipsoids that encompass spaces of highest likelihood. The method is
used in calculating the Bayesian evidence, Z, a byproduct of which is the calculation of posterior distributions. Nested Sampling algorithms, Skilling
(2006) [6], are frequently used in exoplanetary science for performing atmospheric retrievals, through
Nestle, Barbary (2015) [8] and MultiNest,
Feroz et al. (2009) [9], implementations however,
their adoption by solar-system atmospheric science
has yet to be exploited. Nested Sampling retrievals,
unlike Optimal Estimation methods, Rodgers (2000)
[10], capture the full likelihood space of forward
model parameters (within their defined ranges) however, Optimal Estimation methods, used alongside
Bayesian studies, in a comparative manner allows for
the transition of different degrees of degeneracy to be

A)

Index: 174

B)

Index: 142

C)

Index: 86

D)

Index: 50

Figure 1: TauREM Nested Sampling retrieval of
NOMAD SO channel observation nmd-cal-sc-so20180922T091011-20180922T092445-a-i-134 indexes D) 50, C) 86, B) 142 and A) 174, showing the
evolution of posterior distributions of log(HDO),
log(H2O), log(O3), log(13CH4), and log(12CH4) over a

range of tangent heights.
This work presents the first comparison of
TauREx against NASA’s Planetary Spectrum Generator (PSG) and the first use of TauREM and Nested
Sampling retrievals of Martian trace gas species using NOMAD SO channel observations.
Retrieval development experiments are presented
including varying the number of live points used,
performing wavenumber shifts to the model spectrum, and targeting spectra at varying tangent
heights. Moreover, retrieval experiments, with and
without, applying AOTF, Blaze, and ILS functions
are performed, using models described in the literature, Neefs et al. (2015) [13], Thomas et al. (2018)
[14], Liuzzi et al. (2019) [15], Liuzzi and Villanueva
(2021) [16]. These development experiments are
important in understanding how TauREM performs
and are useful in refining the model.
The results presented in this work should be
deemed preliminary. While the results are novel, no
claims are made of superiority of accuracy to the
results derived from using other NOMAD SO forward models. Possible causes of the discrepancies
between TauREM and the PSG are proposed to include:

residuals (TR) of diffraction order 134, for morning,
evening and combined observations, extending from
Investigating trace gases in the Martian atmosphere
using the ExoMars Trace Gas Orbiter Part 1: Analysis of ESA PSA NOMAD SO Channel Data.

Figure 2: PSG forward models (red), gain (green),
and transmittance data (blue) for all NOMAD SO
channel bin 3 observations for nmd-cal-sc-so20180922T091011-20180922T092445-a-i-134,
showing PSG forward models fitting remarkably well
to the data.

1. The use of N2 instead of CO2 broadened absorption cross-sections.
2. Discrepancies in the calculated path lengths, especially at the impact layer.
3. Discrepancies in baseline and wavenumber shift
corrections.
4. The influence of NOMAD SO AOTF, Blaze and
ILS functions.
All of the aforementioned points will be further investigated as part of a future development of
TauREM.
Results: PSG Optimal Estimation Retrievals
In the absence of a validated forward model the
PSG enables an established method of retrieving
atmospheric parameters from transmittance spectra,
comparing real data to PSG synthetically generated
spectra, through employing Optimal Estimation, as
described in Rodgers (2000) [10]. We present an
example NOMAD SO channel diffraction order 134
spectrum is used with the PSG to retrieve the abundance of H2O and the results evaluated. The
NOMAD SO forward model is compared against the
observation, followed by a comparison against the
derived Jacobians and an evaluation of the retrieved
values.
Progressing from the example spectrum vertical
profiles are derived from a set of solar occultations,
Figure 2, before finally performing experimental
retrievals for H2O and CH4. These experimental retrievals are performed on the average transmittance

Figure 3: Diffraction Order 134. Plot of the average
transmittance residual, for 2418 solar occultation
observations. No clear evidence of CH4 is observed,
although evidence for the magnetic dipole of CO2,
between 3016-3021 cm−1, Trokhimovskiy et al.
(2020) [17] is.

Figure 4: Experimental PSG retrieval of NOMAD
SO average transmittance residuals for diffraction
order 134, shown for “All H2O” in Figure 3, for H2O
for morning, evening and combined terminator ob-

servations from -5-70 km, showing abundance levels
similar to those reported in the literature.

Figure 4: Experimental PSG retrieval of NOMAD
SO average TR for diffraction order 134 for CH4 for
morning, evening and combined terminator observations from -5-20 km.

Science Programme, STFC ST/P002528/1. We
would also like to thank Geronimo Villanueva and
Giuliano Liuzzi at NASA Goddard Space Flight
Center for their support in understanding and utilising the Planetary Spectrum Generator. We would
also like to acknowledge the achievements of the
NOMAD experiment. The NOMAD experiment is
led by BIRA-IASB and assisted by Co-PI teams from
Spain (IAA-CSIC), Italy (INAF-IAPS), and the
United Kingdom (Open University).
References:
[1] Villanueva et al. (2018), Planetary Spectrum
Generator: An accurate online radiative transfer suite
for atmospheres, comets, small bodies and
exoplanets, Volume 217, 2018, Pages 86-104.
[2] Villanueva et al. (2022), Fundamentals of the
Planetary Spectrum Generator 2022 Edition, NASA
Goddard Space Flight Center, January 2022.

Conclusions:
Analysis of the results indicates retrieved H2O
profiles are consistent with the levels in published
sources, Aoki et al. (2019) [18]. H2O vertical profiles
are known to significantly vary with local time, season, location and the occurrence of dust storms, Aoki
et al. (2019) [18], Alday et al. (2021) [19], Villanueva et al. (2015) [20], Villanueva et al. (2021) [21].
As such, the retrieved abundances from individual
solar occultations may significantly differ from those
shown in Figure 4.
No clear evidence for CH4 is observed to the levels reported by Formisano et al. (2004) [22],
Krasnopolsky et al. (2004) [23], Mumma et al.
(2009) [24], Webster et al. (2015) [25], Webster et
al. (2018) [26] is observed. However, the low abundance levels of CH4 retrieved do support the results
of Korablev et al. (2019) [27], Knutsen et al. (2021)
[28] and Montmessin et al. (2021) [29]. The results
in this work should be deemed preliminary and more
work should be performed in terms of error analysis.
TauREM requires further development and validation, despite this TauREM and PSG retrievals,
Nested Sampling and Optimal Estimation respectively, have been run on a set of NOMAD SO spectra.
Within this work TauREM retrieval development
experiments have been performed. These development experiments are important in understanding
how TauREM performs and are useful in refining the
model. All the aforementioned points on refining the
forward model will be further investigated as part of
a future development of TauREM and the application
of Nested Sampling to understanding the Martian
atmosphere.

[3] Vandaele et al. (2015), Science objectives and
performances of NOMAD, a spectrometer suite for
the ExoMars TGO mission, Planetary and Space
Science, Volume 119, 2015, Pages 233-249.

Acknowledgements:
We would like to thank the UK Space Agency for
their support of this studentship through the Aurora

[11] Gordon et al. (2017), The HITRAN2016 molecular spectroscopic database, Journal of Quantitative Spectroscopy and Radiative Transfer, Volume

[4] Al-Refaie, A.F., Changeat, Q., Waldmann, I.P.,
Tinetti, G., 2019. TauREx III: A fast, dynamic and
extendable framework for retrievals. Submitted to
The Astrophysical Journal.
[5] Cann, G., Al-Rafaie, A., Waldmann, I., Walton,
D., and Muller, J.-P.: Ares: A retrieval framework
Mars model for ExoMars Trace Gas Orbiter
NOMAD
solar
occultation
measurements,
Europlanet Science Congress 2020, online, 21 September–9 Oct 2020, EPSC2020-433, 2020.
[6] Skilling, J., 2006. Nested sampling for general
Bayesian computation. Bayesian Anal. 1, no. 4, 833859.
[7] Haario et al. (2006), DRAM: Efficient adaptive
MCMC, Statistics and Computing, 16, 339, 2006.
[8] Barbary, K. 2015. Nestle,
https://github.com/kbarbary/nestle.

GitHub.

[9] Feroz et al. (2009) MultiNest: an efficient and
robust Bayesian inference tool for cosmology and
particle physics, Monthly Notices of the Royal Astronomical Society, Vol 398, 1601-1614, 2009.
[10] Rodgers, C.D., 2000. Inverse Methods for Atmospheric Sounding, Vol. 2, World Scientific.

203, 2017, Pages 3-69.
[12] Yurchenko et al. (2018), EXOCROSS: a general
program for generating spectra from molecular line
lists. Astron- omy and Astrophysics, Vol. 614, A131.
[13] Neefs et al. (2015), Optical and radiometric
models of the NOMAD instrument part II: the infrared channels - SO and LNO. Opt. Express 24, 3790.
[14] Thomas et al. (2018), NOMAD Experiment to
Archive Interface Control Document, EXM-NOICD-AER-00001, Is- sue 1, Revision 11. BIRAIASB.
[15] Liuzzi et al. (2019), Methane on Mars: New
insights into the sensitivity of CH4 with the
NOMAD/ExoMars spectrometer through its first inflight calibration. Icarus 321, 671-690, 2019.
[16] Liuzzi G., Villanueva G., Report on the ILS
Characterization of the SO channel, 2021.
https://psg.gsfc.nasa.gov/apps/exomars/data/2021ILS.pdf.
[17] Trokhimovskiy et al. (2020), First observation
of the magnetic dipole CO2 absorption band at 3.3
m in the atmosphere of Mars by the ExoMars trace
gas orbiter ACS instrument. Astron. Astrophys. 639,
A142.
[18] Aoki et al. (2019). Water vapor vertical profiles
on Mars in dust storms observed by TGO/NOMAD.
Journal of Geophysical Research: Planets, 124,
3482– 3497.
[19] Alday et al. (2021), Isotopic fractionation of
water and its photolytic products in the atmosphere
of Mars. Nat Astron 5, 943–950 (2021).
[20] Villanueva et al. (2015), Strong water isotopic
anomalies in the martian atmosphere: Probing current and ancient reservoirs, Science, Volume 348,
Number 6231, pp 218-221, 2015.
[21] Water heavily fractionated as it ascends on Mars
as revealed by ExoMars/NOMAD, Science Advances, Volume 7, Number 7, pp eabc8843, 2021.
[22] Formisano et al. (2004). Detection of methane
in the atmosphere of Mars. Science 306, 1758-1761.
[23] Krasnopolsky et al. (2004). Detection of methane in the Martian atmosphere: evidence for life.
Icarus 172, 537-547.
[24] Mumma et al. (2009), Strong release of Methane on Mars in northern summer 2003. Science
323, 1041-1045.

[25] Webster et al. (2015), Mars methane detection
and variability at Gale crater. Science 347, 415-417.
[26] Webster et al. (2018), Background levels of
methane in Mars atmosphere show strong seasonal
variations. Science 360, 1093-1096.
[27] Korablev et al. (2019), No detection of methane
on Mars from early ExoMars Trace Gas Orbiter observations. Nature 569, E2 (2019).
[28] Knutsen et al. (2021), Comprehensive investigation of Mars methane and organics with
ExoMars/NOMAD, Icarus, Volume 357, 2021.
[29] Montmessin et al. (2021), A stringent upper
limit of 20 pptv for methane on Mars and constraints
on its dispersion outside Gale crater, AA, 650
(2021).