COMPOSITION AND SIZE OF MARTIAN AEROSOLS AS SEEN BY NOMADSO DURING MY34 & 35.
Aurélien Stolzenbach, Instituo de Astrofı́sica de Andalucia, Granada, Spain (astolzenba@iaa.es), Miguel-Angel
López Valverde, Adrian Brines, Ashimananda Modak, Bern Funke, Francisco González-Galindo, Instituo de
Astrofı́sica de Andalucia, Granada, Spain, Ian Thomas, Belgian Royal Institute for Space Aeronomy, Brussels,
Belgium, Giuliano Liuzzi, Geronimo Villanueva, NASA Goddard Space Flight Center, USA, Mikhail Luginin,
Space Research Institute (IKI), Moscow, Russia, Shohei Aoki, University of Tokyo, Kashiwa, Japan.

Introduction
Climates on all planets are heavily influenced by aerosols
in their atmospheres. Their nature, size’s distribution
and content affect the energy budget, hence the atmospheric dynamic of the planet. On Mars, we know that
three types of aerosol exist. Dust, water ice and carbon
dioxide ice. Dust mainly affects the thermal structure
of the martian atmosphere through radiative heating [1].
Water ice has a broader effect since it impacts the thermal structure, as for dust, through modifying the radiative equilibrium [2] but also the martian water cycle [3].
Carbon dioxide ice condensation occurs mainly close to
the poles during winter or at high altitudes [4]. Water
and carbon dioxide ice also influence dust content and
size distribution through their condensation onto smaller
dust particles acting as condensation nuclei [5].
ExoMars TGO mission (ESA/Roscosmos) was primarily designed to study trace gases, thermal structure
and aerosol content in Mars atmosphere with unprecedented vertical resolution [6]. Previous study by [7]
retrieved dust and water ice particle size and abundances from an inversion method using NOMAD-SO
data. Their work used a fixed effective variance but
provided meaningful insight on the challenges about simultaneously retrieving information about dust and water ice from NOMAD-SO observations. Moreover, [7]
demonstrated the sharp gradient of the water ice effective
radius in the troposphere and its diurnal variability.

NOMAD-SO Data processing
NOMAD (Nadir and Occultation for MArs Discovery)
is suite of two infrared spectrometers onboard the ExoMars 2016 Trace Gas Orbiter (TGO) orbiter, covering
the spectral range of 0.2 to 4.3 µm [8]. The instrument has
three spectral channels in order to observe the Martian atmosphere in nadir, limb or solar occultation geometries
at different wavelengths. Two IR spectrometers composed by the Solar Occultation channel (SO), operating
in the range between 2.3 to 4.3 µm (2320-4350 cm−1 ) and
the Limb Nadir Occultation channel (LNO) and a third
one operating in the UV-visible (UVIS), capable of both
nadir and occultation observations. The SO channel has
been designed similar to the Solar Occultation in the In-

frared (SOIR) instrument onboard Venus Express. It uses
an echelle grating with a density of ≈ 4 lines/mm in a
Litrow configuration. An Acousto-Optical Tunable Filter (AOTF) is used to select different spectral windows
(with a width that varies from 20 to 35 cm−1 ). Each
window corresponds to the desired diffraction order to
be used during the atmospheric scan. The spectral resolution of the SO channel is λ/∆λ = 20000. The sampling
of this channel is approximately of 1 second, allowing
a vertical sampling about 1km. The AOTF also permits
a quick change between diffraction orders. As a result,
the SO channel is able to observe the atmosphere at a
given altitude with 6 different diffraction every second.
For this study, we selected a configuration of 5
diffraction orders (121,134,149,168,190) spanning the
overall spectral range of NOMAD. These orders are selected because they are regularly used together in a large
fraction of NOMAD solar occultation observations during the first year of the mission. They also cover a broad
spectral range, essential for a good characterization of
the aerosol composition. Our dataset covers a full martian year starting at MY 34 LS 180◦ and ends at the same
LS of MY 35. It represents 563 vertical scans with the 5
order selected present in each one.
We are using the level 1 transmittances spectrum
provided by the PI team at the Belgian Royal Institute
of Spatial Aeronomy (BIRA). These calibrated transmittance needs to be furthermore processed to correct
residual continuum effects and frequency shifts. The
four prominent features to be taken into account in what
we refer to "Pre-Processing phase" are the Instrument
Line Shape (ILS), the AOTF response, a signal spectral
shift and continuum bending. At the IAA, we developped an in-house preprocessing program that evaluates
the latter two features, the spectral shift and continuum
bending. To do so, we are using a radiative forward
model named KOPRA (Karlsruhe Optimized Radiative
transfer Algorithm described in [9]). The reference atmosphere is taken from the LMDZ Mars GCM [10] for
an exact location and season as the NOMAD data. The
radiative transfer forward model KOPRA uses HITRAN
2016 spectral lines absorption data for relevant gases in
each order (CO2 , CO and H2 O mainly).
BIRA, Goddard and IAA teams worked at a common
solution for the ILS and AOTF. The AOTF updated calibration presented in [11] describe the AOTF response

Composition and size of Martian aerosols as seen by NOMAD-SO during MY34 & 35
with an asymmetric sinc function added to a Gaussian,
operating as an offset. The parameters that describe the
AOTF are wavenumber dependent. The ILS can be described by an addition of two Gaussian functions with a
separation and a scaling ratio between them that varies
across the diffraction orders.
In order to evluate the local extinction due to aerosols,
we use an inversion program called RCP. it stands for
Retrieval Control Program and is a multi-parameter nonlinear least squares fitting of measured and modeled
spectra [12]. The forward model, KOPRA, was recently
adapted to limb emissions on Mars [13] and for this
work it has been adapted to solar occultation data on
Mars for the first time, and in particular to the NOMAD SO channel with implementation of the asymmetric AOTF transfer function and the double Gaussian
ILS. Regarding the aerosols, the first guess and a priori are always a null vertical profile of the extinction.
RCP solves iteratively the inverse problem [14] and is
described in details in [15]. The calculation of the model
spectra and Jacobians is performed with KOPRA, being
called in each step of the iteration. The measurement covariance matrix is calculated from the noise-equivalent
spectral radiance provided with the NOMAD level 1a
data. The regularization matrix is build from Tikhonovtype terms of different orders which can be combined to
obtain a custom-tailored regularization for any particular retrieval problem. RCP uses a Levenberg-Marquardt
damping which is forced to be zero in the last iteration. Convergence is reached when the change of the
retrieval parameters with respect to the previous iteration is smaller than a certain fraction of the expected
noise retrieval error. For this study on martian aerosol we
decided to select a regularization profile that deals with
the trade-off between very oscillating extinction vertical
profile with higher vertical resolution and a very smooth
extinction vertical profile with lower vertical resolution.
A example of the extinction profile is shown in fig. 1.
The retrieved extinctions differs from the usual evaluation using the Onion-peeling or Abel’s transform method
since with this global fit we have other key parameters
as the averaging kernel and the vertical resolution. This
is an important difference with the previous work on
aerosols using ACS data [16, 17], who used an onion
peeling inversion.

Figure 1:

Retrieved extinction vertical profiles from orbit
20180625 050514 1p0a SO A E. the absolute error of the extinction
is plotted horizontally and the vertical resolution vertically at each altitude point.

The log-normal distribution is a function of two parameters (rg , σg ). In optics, we usually change those
parameters to more suitable ones. The effective radius,
ref f , which represents the ability of the size distribution
to remove a part of the incoming light beam, and its
corresponding effective variance νef f .
For any aerosol size distribution, the extinction k at
a wavelength λ is defined as follow:
k(λ) = N σext (λ, ref f , νef f )

Where k is the extinction in km−1 , N the aerosol
number density and σext (λ) is the mean average extinction cross-section at a wavelength λ and for a specific aerosol distribution defined by the two parameters
(ref f , νef f ). In order to get rid of the number density N,
we can construct a new variable χ defined as follow:
χ(λ) =

Mean extinction cross-section ratio modelling
In order to model the optical behavior of the Martian aerosol we need to model first its size distribution.
We chose the log-normal distribution, as shown below,
which is widely used in atmospheric sciences.


(ln(r) − ln(rg ))2
N
nN (r) = √
· exp −
2ln(σg )2
2π r ln(σg )

(1)

k(λ)
k(λ0 )

(2)

On Mars, the aerosol distribution exhibits a bimodal
behavior as shown by [18]. In order to model this characteristic of martian aerosol, we define an extinction
ratio χm (eq. (3)) for a mixture of dust and water ice as
done in [16]. In eq. (3), the subscript d is for dust and
wi is for water ice. The term γ is the ratio of the number
densities of water ice over dust, Nwi/Nd
χm (λ) =

kd (λ) + γ kwi (λ)
kd (λ0 ) + γ kwi (λ0 )

(3)

Composition and size of Martian aerosols as seen by NOMAD-SO during MY34 & 35
Using evaluated refractive indexes for Dust, using
[19], and water ice,using [20, 21], we run a Lorenz-Mie
code for polydisperse spherical particle from [22] for a
set of (ref f , νef f ) in order to evaluate the mean average
extinction cross-section at the selected NOMAD order’s
wavelengths.
Aerosol composition and size distribution evaluation
We are using a mix of non-linear least square and brute
force to evaluate the best set of parameters (ref f , νef f
,γ ), for pure dust, pure H2 O ice and a mixture of the two
represented by γ . The non-linear least square algorithm
is provided by the SciPy Python package [23]. We have
selected the Trust Region Reflective (TRF) algorithm
as solver [24, 25] instead of the traditional LevenbergMarquandt since the TRF solver handles better boundary
conditions. we filter for low averaging kernel, low effective variance, bad merit function and high vertical
resolution.
The errors associated with every parameters are selected at the appropriate step by the diagonal component
of the covariance matrix provided by the NLSQ algorithm. Since the retrieved extinction provided by the
global fit of RCP have very low errors, we are confident
that results with a merit function below 10 are still relevant. To assess the robustness of our fitting procedure,
we will present results against synthetic extinction ratios
signal

[16]. Still in the northern hemisphere around LS =
240◦ , we observe an increase of the size of water ice
particles above 45 to 60 km. This increase is due to
mid to low latitudes measurements at similar LS , also
observed by [7]. The result on the size of the water ice
particles in fig. 2 shows water ice rg goes from small
sub-micron values (< 0.5 µm) to sizes in the range of a
few µm. A similar result was already observed by [7] but
with an effective radius greater than 4 µm for altitudes
above ≈ 40 km. This vertical behavior of the water
ice particle size was attributed to strong convective dust
storms, called "rocket dust storms", characteristic of this
season and location [26].
An inter-annual C-type dust storm [27, 28] occurs
every martian year around LS = 320◦ . Top level altitudes
for water ice and dust aerosol are close to 95 km. Both
their sizes stay in their global ranges, with sub-micron
particles for water ice and a decreasing size, from micron
to sub-micron, with altitude for dust. This C-type storm
event in MY 34 ends before LS = 330◦ .
We will present in much details about other parameters such as ref f , νef f ,and the mass loading. We will
also talk about the inherent bias of our retrieval scheme
and its impact on the results.

Short overview of the results
The MY 34 GDS reached a planetary scale at LS ≈ 195◦ .
Our detection of dust and water ice aerosols reached
altitudes up to 90 km in both hemisphere. The lower
deck of aerosol, with water ice particle size of a few
microns and dust particle size close to 1 µm, rise up to
35 km in both hemisphere.
In both hemisphere, starting at LS = 210◦ , we can
see the decay of the GDS starting from the poles. The
amount of aerosol does not rise up more than 30 km.
[7] showed a similar lack of water ice and dust content
for the same location and period. Unfortunately, we do
not have results from ACS for this period and location

Figure 2: Median radius rg results of the aerosol distribution properties analysis scheme in the southern hemisphere. The filtered aerosol
fitting points due to bad merit function or low/high effective variance
are shown in light grey.

REFERENCES

References

[12]

T. von Clarmann et al. “Retrieval of temperature and
tangent altitude pointing from limb emission spectra
recorded from space by the Michelson Interferometer
for Passive Atmospheric Sounding (MIPAS)”. en. In:
Journal of Geophysical Research: Atmospheres 108.D23
(2003).

[1]

Donald W. Davies. “Effects of dust on the heating of
Mars’ Surface and atmosphere”. en. In: Journal of Geophysical Research: Solid Earth 84.B14 (1979), pp. 8289–
8293.

[2]

J.-B. Madeleine et al. “Aphelion water-ice cloud mapping and property retrieval using the OMEGA imaging
spectrometer onboard Mars Express”. en. In: Journal
of Geophysical Research: Planets 117.E11 (2012).

[13]

Sergio Jiménez-Monferrer et al. “CO2 retrievals in the
Mars daylight thermosphere from its 4.3µm limb emission measured by OMEGA/MEx”. In: Icarus 353 (2021),
p. 113830.

[3]

F. Montmessin et al. “Origin and role of water ice clouds
in the Martian water cycle as inferred from a general
circulation model”. en. In: Journal of Geophysical Research: Planets 109.E10 (2004).

[14]

Clive D Rodgers. Inverse Methods for Atmospheric Sounding. WORLD SCIENTIFIC, 2000.

[15]

Jurado Navarro and Angel Aythami. Retrieval of CO2
and collisional parameters from the MIPAS spectra in
the earth atmosphere. eng. Universidad de Granada,
2016.

[16]

M. Luginin et al. “Properties of Water Ice and Dust
Particles in the Atmosphere of Mars During the 2018
Global Dust Storm as Inferred From the Atmospheric
Chemistry Suite”. en. In: Journal of Geophysical Research: Planets 125.11 (2020), e2020JE006419.

[17]

A. Stcherbinine et al. “Martian Water Ice Clouds During the 2018 Global Dust Storm as Observed by the
ACS-MIR Channel Onboard the Trace Gas Orbiter”.
en. In: Journal of Geophysical Research: Planets 125.3
(2020), e2019JE006300.

[18]

A. A. Fedorova et al. “Evidence for a bimodal size distribution for the suspended aerosol particles on Mars”.
en. In: Icarus 231 (Mar. 2014), pp. 239–260.

[19]

M. J. Wolff et al. “Wavelength dependence of dust aerosol
single scattering albedo as observed by the Compact
Reconnaissance Imaging Spectrometer”. en. In: Journal of Geophysical Research: Planets 114.E2 (2009).

[20]

Stephen G. Warren and Richard E. Brandt. “Optical
constants of ice from the ultraviolet to the microwave:
A revised compilation”. en. In: Journal of Geophysical
Research: Atmospheres 113.D14 (2008).

[21]

M. L. Clapp, D. R. Worsnop, and R. E. Miller. “Frequencydependent optical constants of water ice obtained directly from aerosol extinction spectra”. In: The Journal of Physical Chemistry 99.17 (Apr. 1995), pp. 6317–
6326.

[22]

Michael I Mishchenko, Larry D Travis, and Andrew A
Lacis. Scattering, absorption, and emission of light by
small particles. Cambridge university press, 2002.

[23]

Pauli Virtanen et al. “SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python”. In: Nature
Methods 17 (2020), pp. 261–272.

[24]

D. C. Sorensen. “Newton’s Method with a Model Trust
Region Modification”. In: SIAM Journal on Numerical
Analysis 19.2 (Apr. 1982), pp. 409–426.

[4]

[5]

R. Todd Clancy et al. “Mars Clouds”. In: The Atmosphere and Climate of Mars. Ed. by François Forget
et al. Cambridge Planetary Science. Cambridge: Cambridge University Press, 2017, pp. 76–105.
T. Navarro, F. Forget, E. Millour, and S. J. Greybush.
“Detection of detached dust layers in the Martian atmosphere from their thermal signature using assimilation”. en. In: Geophysical Research Letters 41.19 (2014),
pp. 6620–6626.

[6]

J. Vago et al. “ESA ExoMars program: The next step
in exploring Mars”. en. In: Solar System Research 49.7
(Dec. 2015), pp. 518–528.

[7]

Giuliano Liuzzi et al. “Strong Variability of Martian
Water Ice Clouds During Dust Storms Revealed From
ExoMars Trace Gas Orbiter/NOMAD”. en. In: Journal
of Geophysical Research: Planets 125.4 (2020).

[8]

A. C. Vandaele et al. “NOMAD, an Integrated Suite of
Three Spectrometers for the ExoMars Trace Gas Mission: Technical Description, Science Objectives and Expected Performance”. en. In: Space Science Reviews
214.5 (June 2018), p. 80.

[9]

Gabriele P. Stiller et al. “Karlsruhe optimized and precise radiative transfer algorithm: I. Requirements, justification, and model error estimation”. In: Optical Remote Sensing of the Atmosphere and Clouds. Ed. by
Jinxue Wang, Beiying Wu, Toshihiro Ogawa, and Zhenghua Guan. Vol. 3501. International Society for Optics
and Photonics. SPIE, 1998, pp. 257–268.

[10]

François Forget et al. “Improved general circulation models of the Martian atmosphere from the surface to above
80 km”. In: Journal of Geophysical Research: Planets
104.E10 (1999), pp. 24155–24175.

[11]

Geronimo L Villanueva et al. “The deuterium isotopic
ratio of water released from the Martian caps as measured with TGO/NOMAD”. In: GRL (2022).

REFERENCES
[25]

Mary Ann Branch, Thomas F. Coleman, and Yuying Li.
“A Subspace, Interior, and Conjugate Gradient Method
for Large-Scale Bound-Constrained Minimization Problems”. In: SIAM Journal on Scientific Computing 21.1
(Jan. 1999), pp. 1–23.

[26]

Aymeric Spiga et al. “Rocket dust storms and detached
dust layers in the Martian atmosphere”. en. In: Journal
of Geophysical Research: Planets 118.4 (2013), pp. 746–
767.

[27]

D. M. Kass et al. “Interannual similarity in the Martian atmosphere during the dust storm season”. en. In:
Geophysical Research Letters 43.12 (2016), pp. 6111–
6118.

[28]

Michael D. Smith. “THEMIS observations of Mars aerosol
optical depth from 2002–2008”. en. In: Icarus 202.2
(Aug. 2009), pp. 444–452.