ArticlePDF Available

Planet‐Wide Ozone Destruction in the Middle Atmosphere on Mars During Global Dust Storm


Abstract and Figures

Plain Language Summary Global dust storms are rare but impactful events on Mars, occurring about once in a decade. Previous investigations found how water vapor is redistributed throughout the entire atmosphere in a dust storm. Photolysis of water vapor by sunlight produces highly reactive species that destroy ozone (O3). Here we present O3 measurements taken by the NOMAD/UVIS instrument on the ExoMars Trace Gas Orbiter in the 2018 global dust storm. After advanced data filtering, they demonstrate how O3 in the middle atmosphere was much reduced compared to one Mars year later when no dust storm occurred. 3D atmospheric model simulations of atmospheric chemistry in the global dust storm confirm this planet‐wide O3 destruction, and help to understand the involved processes. The simulations also predict a strong increase in production of atomic hydrogen in the middle atmosphere, that can explain the observed increased hydrogen atmospheric escape during global dust storms.
This content is subject to copyright. Terms and conditions apply.
1. Introduction
Water vapor (H2O) is only a trace species on Mars, with abundances of at most a few parts per thousand in an
otherwise carbon dioxide (CO2) dominated atmosphere. However, it plays a key role in Mars' atmospheric chem-
istry. Indeed, CO2 is subject to photolysis and its current abundances cannot be explained without assuming that
its photolysis product carbon monoxide (CO) is recycled into CO2 through a reaction with OH, a radical resulting
from photolysis of water vapor (McElroy & Donahue,1972; Parkinson & Hunten,1972). Another relatively
abundant trace gas on Mars is ozone (O3), but it is over a thousand times less abundant than water vapor on Mars,
and over a hundred times less abundant than O3 on Earth. Ozone is mainly produced by the three-body reaction
of atomic (O) and molecular (O2) oxygen in the presence of CO2. Yet fast reactions involving odd hydrogen
species (H, OH and HO2, or HOx) resulting from water photochemistry will readily destroy O3 or suppress its
formation in catalytic cycles, making water vapor the key species in controlling the abundance of O3. In the
lower atmosphere, there is sufficient CO2 (i.e., pressure) to maintain large O3 concentrations, but its abundance
Abstract The Nadir and Occultation for MArs Discovery (NOMAD)/UV-visible (UVIS) spectrometer on
the ExoMars Trace Gas Orbiter provided observations of ozone (O3) and water vapor in the global dust storm
of 2018. Here we show in detail, using advanced data filtering and chemical modeling, how Martian O3 in the
middle atmosphere was destroyed during the dust storm. In data taken exactly 1year later when no dust storm
occurred, the normal situation had been reestablished. The model simulates how water vapor is transported to
high altitudes and latitudes in the storm, where it photolyzes to form odd hydrogen species that catalyze O3.
O3 destruction is simulated at all latitudes and up to 100km, except near the surface where it increases. The
simulations also predict a strong increase in the photochemical production of atomic hydrogen in the middle
atmosphere, consistent with the enhanced hydrogen escape observed in the upper atmosphere during global dust
Plain Language Summary Global dust storms are rare but impactful events on Mars, occurring
about once in a decade. Previous investigations found how water vapor is redistributed throughout the entire
atmosphere in a dust storm. Photolysis of water vapor by sunlight produces highly reactive species that destroy
ozone (O3). Here we present O3 measurements taken by the NOMAD/UVIS instrument on the ExoMars Trace
Gas Orbiter in the 2018 global dust storm. After advanced data filtering, they demonstrate how O3 in the middle
atmosphere was much reduced compared to one Mars year later when no dust storm occurred. 3D atmospheric
model simulations of atmospheric chemistry in the global dust storm confirm this planet-wide O3 destruction,
and help to understand the involved processes. The simulations also predict a strong increase in production
of atomic hydrogen in the middle atmosphere, that can explain the observed increased hydrogen atmospheric
escape during global dust storms.
© 2022 The Authors.
This is an open access article under
the terms of the Creative Commons
Attribution-NonCommercial License,
which permits use, distribution and
reproduction in any medium, provided the
original work is properly cited and is not
used for commercial purposes.
Planet-Wide Ozone Destruction in the Middle Atmosphere on
Mars During Global Dust Storm
F. Daerden1 , L. Neary1 , M. J. Wolff2 , R. T. Clancy2 , F. Lefèvre3 ,
J. A. Whiteway4 , S. Viscardy1 , A. Piccialli1 , Y. Willame1, C. Depiesse1 , S. Aoki5 ,
I. R. Thomas1 , B. Ristic1 , J. Erwin1 , J.-C. Gérard6, B. J. Sandor2 , A. Khayat7 ,
M. D. Smith7, J. P. Mason8, M. R. Patel8 , G. L. Villanueva7 , G. Liuzzi7,9 ,
G. Bellucci10, J.-J. Lopez-Moreno11, and A. C. Vandaele1
1Royal Belgian Institute for Space Aeronomy BIRA-IASB, Brussels, Belgium, 2Space Science Institute, Boulder, CO, USA,
3LATMOS, Sorbonne Université, UVSQ Université Paris-Saclay, CNRS, Paris, France, 4Centre for Research in Earth and
Space Science, York University, Toronto, ON, Canada, 5Institute of Space and Astronautical Science ISAS, Japan Aerospace
Exploration Agency JAXA, Sagamihara, Japan, 6LPAP, STAR Institute, Université de Liège, Liège, Belgium, 7NASA
Goddard Space Flight Center, Greenbelt, MD, USA, 8School of Physical Sciences, The Open University, Milton Keynes, UK,
9Department of Physics, School of Arts and Sciences, American University, Washington, DC, USA, 10Istituto di Astrofisica e
Planetologia Spaziali, IAPS-INAF, Rome, Italy, 11Instituto de Astrofisica de Andalucia, IAA-CSIC, Granada, Spain
Key Points:
NOMAD ozone (O3) data filtering
during the 2018 global dust storm
shows strong O3 destruction compared
to one year later with no dust storm
3D simulations of atmospheric
chemistry in the 2018 global dust
storm are presented to understand
impact on odd hydrogen and odd
The model confirms middle-
atmospheric O3 destruction in the
dust storm and predicts increased
photochemical production of hydrogen
Supporting Information:
Supporting Information may be found in
the online version of this article.
Correspondence to:
F. Daerden,
Daerden, F., Neary, L., Wolff, M. J.,
Clancy, R. T., Lefèvre, F., Whiteway,
J. A., etal. (2022). Planet-wide ozone
destruction in the middle atmosphere
on Mars during global dust storm.
Geophysical Research Letters,
49, e2022GL098821. https://doi.
Received 22 MAR 2022
Accepted 20 MAY 2022
Author Contributions:
Conceptualization: F. Daerden
Data curation: F. Daerden, L. Neary, M.
J. Wolff, I. R. Thomas, B. Ristic
Formal analysis: F. Daerden, M. J. Wolff
Investigation: F. Daerden, M. J. Wolff,
R. T. Clancy, F. Lefèvre, J. A. Whiteway
Methodology: F. Daerden, M. J. Wolff
Special Section:
ExoMars Trace Gas Orbiter -
One Martian Year of Science
1 of 11
Geophysical Research Letters
2 of 11
decreases with height. This effect is strengthened by an increase in the photolysis rate coefficient of water vapor
with height. At the same time, water vapor frequently deposits into ice clouds below ∼40km (height changing
with latitude and season) and becomes only sparsely available above the cloud zone. In that region, the O3 forma-
tion is no longer suppressed, and its abundances will increase with height, although limited by the decreasing CO2
density (Daerden etal.,2019; Khayat etal.,2021; Lefèvre etal.,2004; Patel etal.,2021; Clancy and Nair,1996).
Until recently, O3 was observed mostly for its total column abundances (Lefèvre et al., 2021; Montmessin
etal.,2017; Perrier etal.,2006; Clancy etal.,2016; Willame etal.,2017) and in some cases for its vertical profile
(Lebonnois etal.,2006; Montmessin & Lefèvre,2013). The start of science operations of the ExoMars Trace Gas
Orbiter (TGO) in 2018 formed a landmark change in the knowledge of O3 on Mars. TGO carries two sensitive
spectrometer suites, Nadir and Occultation for MArs Discovery (NOMAD, Vandaele etal., 2018, 2019) and
Atmospheric Chemistry Suite (ACS, Korablev etal.,2018,2019). The UV-visible (UVIS) channel of NOMAD
monitors the vertical profile of O3 on a daily basis with the solar occultation technique (Khayat etal., 2021;
Patel et al., 2021). Water vapor profiles are measured simulateanously with the NOMAD IR channel (Aoki
etal.,2019). In addition, the IR channel of ACS could retrieve O3 in some cases (Olsen etal.,2020).
Dramatic changes in the vertical distribution of water vapor during global dust storms (GDS) were first reported
by the SPICAM instrument on Mars Express (Fedorova etal.,2018) for the 2007 GDS, and more recently by the
TGO (Aoki etal.,2019; Fedorova etal.,2020; Vandaele etal.,2019) and SPICAM (Fedorova etal.,2021) for
the 2018 GDS. As the planet's atmosphere contains much more dust, solar radiative heating by the dust changes
the temperatures considerably (Smith,2019) compared to the same season without a GDS. This has a dramatic
impact on water vapor, as the increased temperatures prevent cloud formation, and water vapor is transported
to much larger heights and, following the global circulation (which is also enhanced in dust storms), to higher
latitudes, up to the polar regions. This phenomenon was simulated in a General Circulation Model (GCM),
producing a good agreement with the water observations of NOMAD during the 2018 GDS (Neary etal.,2020).
The strong photochemical relationship between water vapor and O3 implies that the redistribution of water vapor
during the GDS will also have an impact on O3. A first analysis of NOMAD/UVIS O3 profiles taken during
the GDS (Patel etal.,2021) found that mid- and high-latitude O3 abundances above 20km were reduced in the
GDS compared to 1year later, when no GDS occurred. Here we perform new retrievals of the O3 profiles that
allow us to increase the signal-to-noise through spectral and spatial binning, and to develop an improved data
filtering. Both of these facets are particularly helpful for the lower transmittances found in dusty conditions, and
so to provide a clearer picture in the 2018 GDS. We will also investigate the behavior of odd hydrogen and odd
oxygen (including O3) using a GCM that is operated for the conditions of Martian years (MY) 34 and 35 (Daerden
etal.,2022; Neary etal.,2020) and that includes detailed atmospheric chemistry routines (Daerden etal.,2019).
Combining the filtered observations with the model simulations then allows to obtain a detailed picture of the
changes in atmospheric chemistry during the 2018 GDS.
2. Ozone Retrieval and Data Filtering
The NOMAD/UVIS spectrometer operates in the UVIS domain between wavelengths 200–650nm. It has two
observation modes with different sensitivity: the solar occultation mode and nadir/limb viewing mode. In this
study only observations made with the solar occultation channel are considered. Details and characteristics of the
NOMAD/UVIS instrument, its solar occultation channel, and calibration were presented in various publications
(Gérard etal.,2020; Khayat etal.,2021; Patel etal.,2017,2021; Vandaele etal.,2018).
We perform the retrievals of O3 profiles in a manner that is equivalent to previous work (Khayat etal.,2021; Patel
etal.,2021), but different in detail. First, we derive extinction profiles by applying the so-called “onion-peeling”
method to the NOMAD/UVIS transmittance spectra, as was described for TGO/ACS observations (Stcherbinine
etal.,2020). In essence, an upper triangular matrix is formed (at each wavelength) using a spherically symmetric
atmospheric shell model, which is solved analytically to produce vertical profiles of opacity with error propaga-
tion given by equation 7 of Stcherbinine etal.(2020). The column density of O3 in each layer is then determined
by fitting a model of a linear continuum plus O3 absorption (Sander etal., 2011) to the data between 240 and
320nm. The best-fit solution is found using a Marquardt-Levenberg algorithm, as implemented in the package
MPFIT (Markwardt,2009). The uncertainty associated with the derived O3 densities (and linear fit parame-
ters) is provided by MPFIT from the diagonal terms of the covariance matrix. Extensive comparisons of our
Resources: F. Daerden, L. Neary, M. J.
Wolff, I. R. Thomas, B. Ristic
Software: F. Daerden, L. Neary, M. J.
Supervision: F. Daerden
Validation: F. Daerden, M. J. Wolff
Visualization: F. Daerden, L. Neary, R.
T. Clancy, F. Lefèvre
Writing – original draft: F. Daerden
Writing – review & editing: F. Daerden,
L. Neary, M. J. Wolff, R. T. Clancy, F.
Lefèvre, J. A. Whiteway, J. Erwin, J.-C.
Gérard, B. J. Sandor, M. R. Patel
Geophysical Research Letters
3 of 11
retrievals outside of dusty conditions with the published full NOMAD/UVIS O3 data set (Khayat etal.,2021;
Patel etal.,2021) showed good agreement within the uncertainties of each retrieval. The data set presented in this
paper, with full retrieval and filtering (see below) details, is publicly available (Daerden & Wolff,2022).
The motivation to develop our own retrieval process was the ability to perform sensitivity analyses and develop
additional filters to remove spurious O3 detection. Such spurious detections (i.e., O3 detection without an O3
signature present in the transmission spectra) were found to occur in regimes where the opacity spectra can be
quite noisy. In other words, relying on the covariance matrix alone for our retrievals was not sufficient, since it can
produce unphysical results. In this direction, we have taken two approaches. In addition to the formal uncertainty
returned by MPFIT, we calculate an empirical equivalent to the information content approach (Rodgers,2000) by
comparing the χ
2 of the O3 retrieval for each altitude to that of a linear-only model such that values less than one
indicate that allowing O3 provides a better fit, that is, m=χ
2(linear)<1. In other words, if a linear
model fits as well (or better) than the O3 model, the retrieved value is not statistically significant. We also allow
for the binning of transmittance data in both wavelength and height, where the latter dimension is particularly
effective in reducing the noise in the O3 retrievals when the observations employ a large number of positions per
occultation. We generally employed bins of 10nm and 4km. A somewhat similar analysis to ours presenting an
information content approach to the full NOMAD/UVIS O3 data set is in preparation by A. Piccialli etal.
Through numerical experimentation, we developed the following data filtering criteria. Data are removed from a
profile if (a) the average transmittance between 240 and 320nm was below 0.02, and (b) when m=χ
2(linear)>0.7, except during the period of April–November 2018 (which suffered from reduced signal-to-
noise ratio) where we use m>0.85. Outside of this period, the value of 0.7 is preferred since it provides a more
consistent removal of spurious data points. Both m values are the result of substantial trial-and-error processes
which include manual inspection of many (e.g., hundreds) individual profile fits per experiment. (c) A final step
in the filtering process is done when the retrieved O3 number density was less than the uncertainty returned by
3. Ozone in the Global Dust Storm
The observations are shown in Figure1 for latitude bands >45°N and <45°S respectively, for the period before
and during the 2018 GDS (in Mars year, MY, 34) and in the same season exactly one Martian year later (MY35),
when no GDS occurred. The latitude of the observations changes over time, and their geospatial distribution in
both years is shown in Figure S1 inSupporting InformationS1. The choice to show the latitude bands poleward
of 45° resulted from a compromise to minimize the mixing of data taken at different latitudes while maintain-
ing a sufficiently dense temporal coverage of the profiles. Figure2 shows the observations for all latitudes, but
averaged over wider Ls intervals. The data are plotted using the height above the MOLA Mars reference level
(Smith etal.,1999) as vertical coordinate, in order to combine observations taken over a variety of latitudes and
longitudes (and hence for different surface heights). The observation errors for the data shown in both figures are
shown in Figures S2 and S3 in Supporting InformationS1.
In the period before the 2018 GDS (Ls=160°–187°, Figure1 and Figure2a,b), a distinct O3 minimum between
30 and 50km (height varying with time and latitude) can be seen in each hemisphere at high latitudes in both
years, with O3 remaining abundant above this minimum up to 50–60km (height again varying with time and
latitude, this is the high altitude O3 layer that was discussed before, Khayat etal.,2021). This pattern is present in
both Mars years, but in MY34, at southern latitudes, it is less apparent because of the sparse observations in this
case (Figure S1 in Supporting InformationS1). The minimum in O3 is associated with an observed and simulated
maximum in water vapor in this region that is caused by transport of wet air along the ascending branches of the
Hadley circulation cells (Aoki etal.,2019; Daerden etal.,2019; Neary etal.,2020). At these heights, water is
more easily photolyzed and produces odd hydrogen species that both destroy and suppress O3 formation (Daerden
Observational constraints imposed by the orbital geometry of the TGO spacecraft caused gaps in the data of up to
∼15° wide in Ls (at different times of the year in MY34 and MY35, see Figure1), complicating the comparison of
the 2years over certain Ls ranges. Nevertheless, it is seen that in MY35 the ozone peak abundances above 30km
continue through to Ls=230° after which the observable ozone above 30km gradually disappears (Figures1b
and1d). Figure2d confirms this seasonal behavior in MY35, with a similar ozone distribution as before Ls=187°
Geophysical Research Letters
4 of 11
Figure 1. Time series of the ozone (O3) number density profiles. Panels (a and b) for latitudes north of 45°N, for the same season in Mars year 34 (a, with GDS) and in
Mars year 35 (b). Panels (c and d) show the same for latitudes south of 45°S. Left plots for the O3 observed by NOMAD/UVIS and right ones from simulations. Gaps in
the data are due to observational restrictions imposed by orbital geometry of the Trace Gas Orbiter spacecraft. Gray shading: for the observations, this represents values
that were filtered out (see Section2) or are below 10
−3; for the simulations, values are below 10
−3. Height is taken with respect to the MOLA reference level
(black shading at the bottom of the panels shows the actual surface). Model results were interpolated to the time and location of the observations. The thick vertical
dashed lines indicate the onset of the global dust storm in MY34 and the same time one Mars year later.
Geophysical Research Letters
5 of 11
Figure 2. Latitude-height distribution of the ozone (O3) number density profiles. Panels (a and b) show the average over Ls=160°–187° and (c and d) over Ls=187°–
230°, in MY34 (a and c) and MY35 (b and d). The first period is before the global dust storm (GDS) in MY34, the second period covers the GDS in MY34. Left plots
for the O3 observed by NOMAD/UVIS and right ones from simulations. Gray shading: for the observations, this represents values that were filtered out (see Section2);
for the simulations, values are below 2×10
−3. Height is with respect to the MOLA reference level (black shading shows zonally averaged MOLA topography). A
grid of 1.5° in latitude and 1.5km in height was used to compute the averages. Model results were interpolated to the time and location of the observations.
Geophysical Research Letters
6 of 11
(Figure2b) but with slightly reduced abundances. In MY34 however, ozone is strongly reduced immediately after
the onset of the GDS for all heights above 20km in the north and 30km in the south (Figures1a and1c), consist-
ent with the previous analyses (Khayat etal.,2021; Patel etal.,2021). This dramatic ozone reduction takes place
at all latitudes, as can be seen in Figure2c.
4. Atmospheric Chemistry Simulations
The differences in O3 behavior between MY34 and MY35 have been qualitatively attributed to the observed
increase of water vapor during the GDS at heights and latitudes that, in normal (non-GDS) conditions exhibit
very low water vapor abundances (Patel etal.,2021). The enhanced high-altitude water vapor abundances lead
to increased water vapor photolysis, producing odd hydrogen species that rapidly catalyze O3 loss. However, the
enhanced odd hydrogen species that are key to this process were not observable at the time. Without observational
access to these radical species that drive the chemical pathways between water vapor and O3, we turn to a model
for atmospheric chemistry (Daerden etal.,2019; Neary & Daerden,2018) to understand the enhanced destruction
of O3 during the GDS. Previous work has demonstrated how this model reproduced the redistribution of water
vapor observed by NOMAD during the 2018 GDS (Neary etal.,2020). Here the model was run consecutively
for MY34 and MY35. Both years were simulated using their respective daily dust optical depth climatologies
(Montabone etal., 2015, 2020). For a direct comparison of model and data, the time-evolved 3D simulation
results were interpolated in space and time to the NOMAD/UVIS observed O3 profiles and added to Figures1
and2 in the same format as the data, but without filtering.
The resemblance of the model results to the O3 observations is strong in morphology and often also in absolute
values, providing evidence that the model captures well the underlying chemical processes. To understand these
processes, Figure3 shows the time-evolved simulation results of a range of key species (H2O, OH, H, HO2, O and
O3) at latitudes 60°N and 60°S in both Mars years. Figure4 shows the global latitude-height cross-sections of
the same species averaged over Ls=210°–220° (peak of the GDS in MY34) in both MY34 and MY35, and their
relative change in this time window. (As the changes for odd hydrogen species between the 2years are very large
(ratio >1,000), we show them as a ratio. For O3, the differences are smaller (ratio ∼< 20), and we show them as a
percentage difference on a linear scale.) Both figures assist to the interpretation of Figures1 and 2.
In the middle atmosphere, up to 100km, and as low as 20km at at low- and midlatitudes, down to 10km north
of of 45°N latitude, and down to the surface south 60°S, the O3 abundances are severely reduced, up to 100%
(Figure 4f, right panel, and Figure S8 in Supporting Information S1, right panels). This is a planet-wide O3
destruction, taking place throughout almost the entire atmosphere, and demonstrates the massive impact of the
GDS on the atmospheric composition.
Because water vapor is redistributed in a GDS from the lower atmosphere at low latitudes to the rest of the
atmosphere (Neary etal.,2020), a small decrease in water vapor columns could be seen in measurements during
previous GDS (M. D. Smith,2004; M. Smith etal.,2018; Trokhimovskiy etal.,2015). As a consequence, we here
simulate that O3 abundances at low altitudes, low- and midlatitudes are increased (Figure4f).
In terms of total O3 columns, these changes are small in absolute terms, because (a) the air densities in the lowest
atmosphere dominate the total column contribution, and (b) because total O3 abundances in this season are
small. But the changes are not small in relative terms. Figure S1 in Supporting InformationS1 shows how the
O3 column at low latitudes is predicted to increase from below 1 to almost 2 μm-atm, this is a 100% increase. At
high latitudes, the reduction of the O3 column is up to 2 μm-atm, representing a 50% decrease in the north and a
100% decrease in the south.
Five reactions (see below) are dominating the change of the O3 in the GDS. This change in O3 follows the change
in the relative strength of these reactions during the GDS (Figures S4 through S7 in Supporting Information S1
for reaction numbering and rates). Combined, they define the main pathway from enhanced water vapor to O3
depletion, which can be described as follows:
1. The increase of water vapor in the middle and upper atmosphere and its subsequent photolysis (H2O+hν →
H+OH, J11-13) results in the increased formation of atomic hydrogen (H) and OH radicals.
2. The OH radicals react with CO (CO+OH → CO2+H, R1), and to a lesser extent with O (O+OH → O2+H,
R10), to cause an additional increase in atomic hydrogen.
Geophysical Research Letters
7 of 11
3. Atomic hydrogen reacts with molecular oxygen (H+O2+CO2 → HO2+CO2, R26) leading to an increase
in HO2.
4. HO2 then reacts with atomic oxygen, O (HO2+O → OH+O2, R11) and so reduces the abundance of atomic
oxygen. OH is returned to the atmosphere to resume step 2, and acts as a catalyzer.
5. Finally, the decrease in O suppresses the formation of O3 in O+O2+CO2 → O3+CO2 (R24)
The direct destruction of O3 by HO2 and OH is also enhanced in the GDS (Figure S7 in Supporting Informa-
tionS1, R18 and R21), but only at high altitudes/latitudes, and not in the region of highest O3 abundances, and so
contributes less to the O3 changes in the GDS.
The simulated time evolution of O3 throughout the GDS as a result of these reactions, in comparison with the
simulated evolution in MY35 and with the relative differences between both years, is shown Figure S8 in Support-
ing InformationS1. An animation showing the model simulation of the O3 vertical distribution in the GDS and
one year later is also included in the Supporting InformationS1.
A question to consider is to what extent the specific spatio-temporal distribution of the NOMAD observations
(Figure S1 in Supporting InformationS1) could have an impact on our results. In some cases, there are differ-
ences in coverage between MY34 and MY35, such as in the south before the GDS, or in the northern high-lati-
tudes after Ls∼200°. However between Ls∼215°–230°, the coverage is similar in both years, so that the results
shown in Figure2 are robust. When considering only the profiles that are close in both Ls and latitude in both
Figure 3. Simulated time series of H2O, OH, H, HO2, H and O3 number densities for latitudes 60°N (top rows) and 60°S (bottom rows) around equinox, for Martian
years 34 (with global dust storm,GDS) and 35 (no GDS). The color shading shows the number density averaged over all longitudes and local times. The dashed white
lines indicate the onset and end of the GDS in MY34 and the same times in MY35.
Geophysical Research Letters
8 of 11
Figure 4. Latitude-height cross-sections of the simulated number densities of H2O, OH, H, HO2, O and O3 for MY34 (left
column) and MY35 (center column), averaged over Ls=210°–220° (period which falls in the global dust storm (GDS) in
MY34). The right column show the ratio of the averaged number densities in MY34 and MY35, except for O3 for which
the relative difference is shown. The densities from both years were first interpolated to a common altitude grid as the
atmospheric height scale changed during the GDS.
Geophysical Research Letters
9 of 11
years (Figures S9 and S10 in Supporting InformationS1), there are more gaps in the figures, but they confirm our
conclusions about how NOMAD witnessed O3 loss in the 2018 GDS.
An interesting result from the simulations is the increased production of atomic hydrogen in the middle and
upper atmosphere (Figures3c and4c), with increased number densities by factors of 10–100 above 60km. This
result (already preliminary shown in Neary etal.,2020) seems consistent with the observed enhanced abundance
of atomic hydrogen in the upper atmosphere as observed by MAVEN during the 2018 global and regional dust
storms (Chaffin etal.,2021; Stone etal.,2020).
5. Conclusions
The two endpoints of the above chain of reactions, water vapor and O3, were observed by NOMAD before and
during (MY34) and out of (MY35) the 2018 GDS (Aoki etal.,2019; Khayat etal.,2021; Patel etal.,2021). The
model demonstrates the detailed chain of reactions occurring between these end points, involving unobserved
odd hydrogen and oxygen species. The successful reproduction of the water vapor (Neary etal.,2020) and O3
observations (this paper)—both end points in the chemistry chain—in very different conditions (in and out of a
GDS), by the model provides support for the simulation of these unobserved species and the involved reactions.
The simulated O3 destruction in the GDS, supported by available observations, occurs throughout the middle
atmosphere, that is, on a planet-wide scale. This shows the massive impact of the GDS and the redistribution of
water vapor on atmospheric chemistry and composition, as the odd hydrogen radicals resulting from water vapor
photolysis are highly reactive.
Interestingly, the simulated increase in the production of atomic hydrogen in the middle atmosphere during the
GDS (Figures 3c and 4c) is important to understand the observed enhanced atmospheric escape during dust
storms (Chaffin etal.,2021; Stone etal.,2020). It has been generally assumed that the redistribution of water
vapor during GDS is causing the observed enhanced escape (Chaffin etal.,2021; Heavens etal.,2018; Shaposh-
nikov etal.,2021; Stone etal.,2020). While a dynamical perspective to understand the transport of water vapor
to the upper atmosphere during a GDS was given in terms of wave activity (Yiğit,2021), we provide here the
photochemical processes that are also involved.
Data Availability Statement
ExoMars Trace Gas Orbiter data are publicly available through the European Space Agency's Planetary Science
Archive ( with additional access to NOMAD data through the PI institute (http:// The NOMAD/UVIS Solar Occultation O3 data set presented in this work, as well as the
results from the General Circulation Model simulations are available on the BIRA-IASB data repository (https://, Daerden & Wolff, 2022). The GEM-Mars General Circu-
lation Model is based on the Global Environmental Multiscale model 4.2.0 version of the community weather
forecasting model for Earth, which is one of the more recent versions available to the community, under the GNU
Lesser General Public Licence v2.1. The adaptation for Mars is developed and maintained at the Royal Belgian
Institute for Space Aeronomy.
Aoki, S., Vandaele, A. C., Daerden, F., Villanueva, G. L., Liuzzi, G., Thomas, I. R., etal. (2019). Water vapor vertical profiles on Mars in dust
storms observed by TGO/NOMAD. Journal of Geophysical Research: Planets, 124(12), 3482–3497.
Chaffin, M. S., Kass, D. M., Aoki, S., Fedorova, A. A., Deighan, J., Connour, K., etal. (2021). Martian water loss to space enhanced by regional
dust storms. Nature Astronomy, 5(10), 1036–1042.
Clancy, R. T., & Nair, H. (1996). Annual (perihelion–aphelion) cycles in the photochemical behavior of the global Mars atmosphere. Journal of
Geophysical Research, 101(E5), 12785–12790.
Clancy, R. T., Wolff, M. J., Lefevre, F., Cantor, B. A., Malin, M. C., & Smith, M. D. (2016). Daily global mapping of Mars ozone column abun-
dances with MARCI UV band imaging. Icarus, 266, 112–133.
Daerden, F., Neary, L., Villanueva, G., Liuzzi, G., Aoki, S., Clancy, R. T., etal. (2022). Explaining NOMAD D/H observations by cloud-induced frac-
tionation of water vapor on Mars. Journal of Geophysical Research: Planets, 127(2), e2021JE007079.
Daerden, F., Neary, L., Viscardy, S., García Muñoz, A., Clancy, R. T., Smith, M. D., etal. (2019). Mars atmospheric chemistry simulations with
the GEM-Mars general circulation model. Icarus, 326, 197–224.
Daerden, F., & Wolff, M. J. (2022). NOMAD/UVIS ozone retrievals and GEM-Mars GCM simulations of ozone on Mars during the
2018 global dust storm and one year later, for Daerden etal., [Data set]. Royal Belgian Institute for Space Aeronomy (BIRA-IASB).
This work was made possible thanks
to the reconstructed gridded maps of
column dust optical depth from Mars
Climate Sounder observations provided
by L. Montabone. The dust maps were
prepared using MCS v5.3 provided by A.
Kleinböhl and D. Kass. Dust clima-
tologies can be found at the following
mars/dust_climatology/. ExoMars is a
space mission of the European Space
Agency (ESA) and Roscosmos. The
NOMAD experiment is led by the Royal
Belgian Institute for Space Aeronomy
(IASB-BIRA), assisted by Co-PI teams
from Spain (IAA-CSIC), Italy (INAF-
IAPS), and the United Kingdom (Open
University). This project acknowledges
funding by the Belgian Science Policy
Office (BELSPO), with the financial
and contractual coordination by the
ESA Prodex Office (PEA 4000103401,
4000121493), by the UK Space Agency
(grants ST/V002295/1, ST/P001262/1,
ST/V005332/1 and ST/S00145X/1), by
the Spanish Ministry of Science and
Innovation (MCIU), and by European
funds (grants PGC2018-101836-B-I00
and ESP2017-87143-R, MINECO/
FEDER), as well as by the Italian Space
Agency (Grant 2018-2-HH.0). This work
was supported by the Belgian Fonds de
la Recherche Scientifique – FNRS (Grant
Nos. 30442502, ET_HOME). This work
has received funding from the European
Union's Horizon 2020 research and
innovation programme (grant agreement
No 101004052, RoadMap project). The
IAA/CSIC team acknowledges financial
support from the State Agency for
Research of the Spanish MCIU through
the “Center of Excellence Severo Ochoa”
award for the Instituto de Astrofísica de
Andalucía (SEV-2017-0709). US inves-
tigators were supported by the National
Aeronautics and Space Administration,
by NASA's Mars Program Office (under
WBS 604796, “Participation in the TGO/
NOMAD Investigation of Trace Gases on
Mars.”), and by NASA (award number
80GSFC21M0002). Canadian investi-
gators were supported by the Canadian
Space Agency. We thank Manuel López
Puertas and two anonymous reviewers for
useful comments on the manuscript.
Geophysical Research Letters
10 of 11
Fedorova, A., Bertaux, J.-L., Betsis, D., Montmessin, F., Korablev, O., Maltagliati, L., & Clarke, J. (2018). Water vapor in the middle atmosphere
of Mars during the 2007 global dust storm. Icarus, 300, 440–457.
Fedorova, A., Montmessin, F., Korablev, O., Lefevre, F., Trokhimovskiy, A., & Bertaux, J. L. (2021). Multi-annual monitoring of the water vapor
vertical distribution on Mars by SPICAM on Mars Express. Journal of Geophysical Research: Planets, 126(1), e2020JE006616. https://doi.
Fedorova, A. A., Montmessin, F., Korablev, O., Luginin, M., Trokhimovskiy, A., Belyaev, D. A., etal. (2020). Stormy water on Mars: The distri-
bution and saturation of atmospheric water during the dusty season. Science, 367(6475), 297–300.
Gérard, J.-C., Aoki, S., Willame, Y., Gkouvelis, L., Depiesse, C., Thomas, I. R., etal. (2020). Detection of green line emission in the dayside
atmosphere of Mars from NOMAD-TGO observations. Nature Astronomy, 4(11), 1049–1052.
Heavens, N. G., Kleinböhl, A., Chaffin, M. S., Halekas, J. S., Kass, D. M., Hayne, P.O., etal. (2018). Hydrogen escape from Mars enhanced by
deep convection in dust storms. Nature Astronomy, 2, 126–132.
Khayat, A. S. J., Smith, M. D., Wolff, M., Daerden, F., Neary, L., Patel, M. R., etal. (2021). ExoMars TGO/NOMAD-UVIS vertical profiles
of ozone: 2. The high-altitude layers of atmospheric ozone. Journal of Geophysical Research: Planets, 126(11), e2021JE006834. https://doi.
Korablev, O., Vandaele, A. C., Montmessin, F., Fedorova, A. A., Trokhimovskiy, A., Forget, F., etal. (2019). No detection of methane on Mars
from early ExoMars Trace Gas Orbiter observations. Nature, 568(7753), 517–520.
Korablev, O. I., Montmessin, F., Trokhimovskiy, A., Fedorova, A. A., Shakun, A. V., Grigoriev, A. V., et al. (2018). The atmospheric chem-
istry suite (ACS) of three spectrometers for the ExoMars 2016 trace gas orbiter. Space Science Reviews, 214(1), 7.
Lebonnois, S., Quemerais, E., Montmessin, F., Lefevre, F., Perrier, S., Bertaux, J. L., & Forget, F. (2006). Ver tical distribution of ozone on
Mars as measured by SPICAM/Mars Express using stellar occultations. Journal of Geophysical Research, 111(E9), E09S05. https://doi.
Lefèvre, F., Lebonnois, S., Montmessin, F., & Forget, F. (2004). Three-dimensional modeling of ozone on Mars. Journal of Geophysical
Research, 109(E7), E07004.
Lefèvre, F., Trokhimovskiy, A., Fedorova, A., Baggio, L., Lacombe, G., Maattanen, A., etal. (2021). Relationship between the ozone and water
vapour columns on Mars as observed by SPICAM and calculated by a global climate model. Journal of Geophysical Research: Planets, 126(4),
Markwardt, C. (2009). Non-linear least squares fitting in IDL with MPFIT. In Astronomical Data Analysis Software and Systems XVIII ASP
Conference Series, In D. A. Bohlender, D. Durand, & P. Dowler (Eds.) (Vol. 411) 251. Astronomical Society of the Pacific. Retrieved from
McElroy, M. B., & Donahue, T. M. (1972). Stability of the Martian atmosphere. Science, 177(4053), 986–988.
Montabone, L., Forget, F., Millour, E., Wilson, R. J., Lewis, S. R., Cantor, B., etal. (2015). Eight-year climatology of dust optical depth on Mars.
Icarus, 251, 65–95.
Montabone, L., Spiga, A., Kass, D. M., Kleinböhl, A., Forget, F., & Millour, E. (2020). Martian year 34 column dust climatology from Mars
climate sounder observations: Reconstructed maps and model simulations. Journal of Geophysical Research: Planets, 125(8). https://doi.
Montmessin, F., Korablev, O., Lefevre, F., Bertaux, J. L., Fedorova, A., Trokhimovskiy, A., et al. (2017). SPICAM on Mars express: A 10 year
in-depth survey of the Martian atmosphere. Icarus, 297, 195–216.
Montmessin, F., & Lefèvre, F. (2013). Transport-driven formation of a polar ozone layer on Mars. Nature Geoscience, 6(11), 930–933. https://
Neary, L., & Daerden, F. (2018). The GEM-Mars general circulation model for Mars: Description and evaluation. Icarus, 300, 458–476. https://
Neary, L., Daerden, F., Aoki, S., Whiteway, J., Clancy, R. T., Smith, M., etal. (2020). Explanation for the increase in high-altitude water
on Mars observed by NOMAD during the 2018 global dust storm. Geophysical Research Letters, 47(7), e2019GL084354. https://doi.
Olsen, K. S., Lefèvre, F., Montmessin, F., Trokhimovskiy, A., Baggio, L., Fedorova, A., etal. (2020). First detection of ozone in the mid-infrared
at Mars: Implications for methane detection. Astronomy & Astrophysics, 639, A141.
Parkinson, T. D., & Hunten, D. M. (1972). Spectroscopy and aeronomy of O2 on Mars. Journal of the Atmospheric Sciences, 29(7), 1380–1390.<1380:saaooo>;2
Patel, M. R., Antoine, P., Mason, J., Leese, M., Hathi, B., Stevens, A. H., etal. (2017). NOMAD spectrometer on the ExoMars trace gas orbiter
mission: Part 2—Design, manufacturing, and testing of the ultraviolet and visible channel. Applied Optics, 56(10), 2771–2782. https://doi.
Patel, M. R., Sellers, G., Mason, J. P., Holmes, J. A., Brown, M. A. J., Lewis, S. R., etal. (2021). ExoMars TGO/NOMAD-UVIS vertical profiles
of ozone: 1. Seasonal variation and comparison to water. Journal of Geophysical Research: Planets, 126(11), e2021JE006837. https://doi.
Perrier, S., Bertaux, J. L., Lefevre, F., Lebonnois, S., Korablev, O., Fedorova, A., & Montmessin, F. (2006). Global distribution of total ozone on
Mars from SPICAM/MEX UV measurements. Journal of Geophysical Research, 111(E9), E09S06.
Rodgers, C. D. (2000). Inverse methods for atmospheric sounding. World Science.238
Sander, S. P., Abbatt, J., Barker, J. R., Burkholder, J. B., Friedl, R. R., Golden, D. M., etal. (2011). Chemical kinetics and photochemical data for
use in atmospheric studies. Technical Report Evaluation No 17, Publication 10-6. Jet Propulsion Laboratory. Retrieved from http://jpldataeval.
Shaposhnikov, D. S., Medvedev, A. S., & Rodin, A. V. (2021). Simulation of water vapor photodissociation during dust storm season on Mars.
Solar System Research, 56(1), 23–31.
Smith, D. E., Zuber, M. T., Solomon, S. C., Phillips, R. J., Head, J. W., Gar vin, J. B., etal. (1999). The global topog raphy of Mars and implications
for surface evolution. Science, 284(5419), 1495–1503.
Smith, M. D. (2004). Interannual variability in TES atmospheric observations of Mars during 1999–2003. Icarus, 167(1), 148–165. https://doi.
Smith, M. D. (2019). THEMIS observations of the 2018 Mars global dust storm. Journal of Geophysical Research: Planets, 124(11), 2929–2944.
Smith, M., Daerden, F., Neary, L., & Khayat, S. (2018). The climatology of carbon monoxide and water vapour on Mars as observed by CRISM
and modeled by the GEM-Mars general circulation model. Icarus, 301, 117–131.
Geophysical Research Letters
11 of 11
Stcherbinine, A., Vincendon, M., Montmessin, F., Wolff, M. J., Korablev, O., Fedorova, A., etal. (2020). Martian water ice clouds during the 2018
global dust storm as observed by the ACS-MIR channel onboard the Trace Gas Orbiter. Journal of Geophysical Research: Planets, 125(3),
Stone, S. W., Yelle, R. V., Benna, M., Lo, D. Y., Elrod, M. K., & Mahaffy, P.R. (2020). Hydrogen escape from Mars is driven by seasonal and
dust storm transport of water. Science, 370(6518), 824–831.
Trokhimovskiy, A., Fedorova, A., Korablev, O., Montmessin, F., Bertaux, J. L., Rodin, A., & Smith, M. D. (2015). Mars’ water vapour mapping
by the SPICAM IR spectrometer: Five Martian years of observations. Icarus, 251, 50–64.
Vandaele, A. C., Korablev, O., Daerden, F., Aoki, S., Thomas, I. R., Altieri, F., etal. (2019). Martian dust storm impact on atmospheric H2O and
D/H observed by ExoMars Trace Gas Orbiter. Nature, 568(7753), 521–525.
Vandaele, A. C., Lopez-Moreno, J. J., Patel, M. R., Bellucci, G., Daerden, F., Ristic, B., et al. (2018). NOMAD, an integrated suite of three
spectrometers for the ExoMars trace gas mission: Technical description, science objectives and expected performance. Space Science Reviews,
214(5), 80.
Willame, Y., Vandaele, A., Depiesse, C., Lefevre, F., Letocart, V., Gillotay, D., & Montmessin, F. (2017). Retrieving cloud, dust and ozone
abundances in the Martian atmosphere using SPICAM/UV nadir spectra. Planetary and Space Science, 142, 9–25.
Yiğit, E. (2021). Martian water escape and internal waves. Science, 374(6573), 1323–1324.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
The vertical profiles of water vapor and its semi‐heavy hydrogen isotope HDO provided by instruments on ExoMars Trace Gas Orbiter constitute a unique new data set to understand the Martian water cycle including its isotopic composition. As water vapor undergoes hydrogen isotopic fractionation upon deposition (but not sublimation), the D/H isotopic ratio in water is a tracer of phase transitions, and a key quantity to understand the long‐term history of water on Mars. Here, we present 3D global simulations of D/H in water vapor and compare them to the vertically resolved observations of D/H and water ice clouds taken by NOMAD during the second half of Mars year 34. D/H is predicted to be constant with height up to the main cloud level, above which it drops because of strong fractionation, explaining the upper cut‐off in the NOMAD observations when HDO drops below detectability. During the global and regional dust storms of 2018/2019, we find that HDO ascends with H2O, and that the D/H ratio is constant and detectable up to larger heights. The simulations are within the provided observational uncertainties over wide ranges in season, latitude and height. Our work provides evidence that the variability of the D/H ratio in the lower and middle atmosphere of Mars is controlled by fractionation on water ice clouds, and thus modulated by diurnally and seasonally varying cloud formation. We find no evidence of other processes or reservoirs that would have a significant impact on the D/H ratio in water vapor.
Full-text available
We present ∼1.5 Mars Years (MY) of ozone vertical profiles, covering LS = 163° in MY34 to LS = 320° in MY35, a period which includes the 2018 global dust storm. Since April 2018, the Ultraviolet and Visible Spectrometer channel of the Nadir and Occultation for Mars Discovery (NOMAD) instrument aboard the ExoMars Trace Gas Orbiter has observed the vertical, latitudinal and seasonal distributions of ozone. Around perihelion, the relative abundance of both ozone and water (from coincident NOMAD measurements) increases with decreasing altitude below ∼40 km. Around aphelion, localized decreases in ozone abundance exist between 25 and 35 km coincident with the location of modeled peak water abundances. High‐latitude (>±55°), high altitude (40–55 km) equinoctial ozone enhancements are observed in both hemispheres (LS ∼350°–40°) and discussed in the companion article to this work (Khayat et al., 2021). The descending branch of the main Hadley cell shapes the observed ozone distribution at LS = 40°–60°, with the possible signature of a northern hemisphere thermally indirect cell identifiable from LS = 40°–80°. Morning terminator observations show elevated ozone abundances with respect to evening observations, with average ozone abundances between 20 and 40 km an order of magnitude higher at sunrise compared to sunset, attributed to diurnal photochemical partitioning along the line of sight between ozone and O or fluctuations in water abundance. The ozone retrievals presented here provide the most complete global description of Mars ozone vertical distributions to date as a function of season and latitude.
Full-text available
Solar occultations performed by the Nadir and Occultation for MArs Discovery (NOMAD) ultraviolet and visible spectrometer (UVIS) onboard the ExoMars Trace Gas Orbiter (TGO) have provided a comprehensive mapping of atmospheric ozone density. The observations here extend over a full Mars year (MY) between April 21, 2018 at the beginning of the TGO science operations during late northern summer on Mars (MY 34, Ls = 163°) and March 9, 2020 (MY 35). UVIS provided transmittance spectra of the Martian atmosphere allowing measurements of the vertical distribution of ozone density using its Hartley absorption band (200–300 nm). The overall comparison to water vapor is found in the companion paper to this work (Patel et al., 2021, Our findings indicate the presence of (a) a high‐altitude peak of ozone between 40 and 60 km in altitude over the north polar latitudes for at least 45% of the Martian year during midnorthern spring, late northern summer‐early southern spring, and late southern summer, and (b) a second, but more prominent, high‐altitude ozone peak in the south polar latitudes, lasting for at least 60% of the year including the southern autumn and winter seasons. When present, both high‐altitude peaks are observed in the sunrise and sunset occultations, suggesting that the layers could persist during the day. Results from the Mars general circulation models predict the general behavior of these peaks of ozone and are used in an attempt to further our understanding of the chemical processes controlling high‐altitude ozone on Mars.
Full-text available
Mars has lost most of its initial water to space as atomic hydrogen and oxygen. Spacecraft measurements have determined that present-day hydrogen escape undergoes large variations with season that are inconsistent with long-standing explanations. The cause is incompletely understood, with likely contributions from seasonal changes in atmospheric circulation, dust activity and solar extreme ultraviolet input. Although some modelling and indirect observational evidence suggest that dust activity can explain the seasonal trend, no previous study has been able to unambiguously distinguish seasonal from dust-driven forcing. Here we present synoptic measurements of dust, temperature, ice, water and hydrogen on Mars during a regional dust event, demonstrating that individual dust events can boost planetary H loss by a factor of five to ten. This regional storm occurred in the declining phase of the known seasonal trend, establishing that dust forcing can override this trend to drive enhanced escape. Because similar regional storms occur in most Mars years, these storms may be responsible for a large fraction of Martian water loss and represent an important driver of Mars atmospheric evolution.
Full-text available
Ozone (O3) in the atmosphere of Mars is produced following the photolysis of CO2 and is readily destroyed by the hydrogen radicals (HOx) released by the photolysis and oxidation of water vapor. As a result, an anti‐correlation between ozone and water vapor is expected. We describe here the O3‐H2O relationship derived from 4 Martian years of simultaneous observations by the SPICAM spectrometer onboard the Mars Express spacecraft. A distinct anti‐correlation is found at high latitudes, where the O3 column varies roughly with the −0.6 power of the H2O column. The O3 and H2O columns are uncorrelated at low latitudes. To evaluate our quantitative understanding of the Martian photochemistry, the observed O3‐H2O relationship is then compared to that predicted by a global climate model with photochemistry. For identical model and observed abundances of H2O, the model underpredicts observed ozone by about a factor of 2 relative to SPICAM when using the currently recommended gas‐phase chemistry. Sensitivity studies employing low‐temperature CO2 absorption cross sections, or adjusted kinetics rates, do not solve this bias. Taking into account potential heterogeneous processes of HOx loss on clouds leads to a significant improvement, but only at high northern latitudes. More broadly, the modeled ozone deficits suggest that the HOx‐catalyzed photochemistry is too efficient in our simulations. This problem is consistent with the long‐standing underestimation of CO in Mars photochemical models, and may be related to similar difficulties in modeling O3 and HOx in the Earth's upper stratosphere and mesosphere.
Full-text available
The distribution of water vapor with altitude has long remained a missing piece of the observational data set of water vapor on Mars. In this study, we present the first multi‐annual survey of water vapor profile covering the altitude range from 0 to 100 km based on the SPICAM/Mars Express occultation measurements. During the aphelion season, water remains confined below 40–60 km for all Martian years observed. The highest altitude where water vapor can be spotted is between 70 and 90 km during the southern summer (Ls = 240–300°; perihelion season), approaching the transition between the middle and upper atmosphere. In this season, years without a global dust storm (GDS) show a significant moistening of the upper atmosphere (∼100 ppmv) in the southern hemisphere, confirming a seasonal impact on the hydrogen escape rate. The two observed GDS, in MY28 and MY34, show a substantial disparity in water vapor response. The storm in MY28, which coincides with the southern summer solstice, creates the largest excess of water in both hemispheres at >80 km. This climatology of water vapor will supply a robust statistical basis to address the long‐term escape processes of water from Mars.
Evidence has accumulated in recent years that suggests that Mars used to have more habitable conditions (1). Determining the processes that led to its current cold and dry state are important for understanding habitability more generally. Observations and modeling efforts show that atmospheric escape has adversely affected Mars’ habitability primarily by irreversibly diminishing the atmospheric water reservoir through loss of atomic oxygen and hydrogen into space (2). Understanding this requires connecting dust storms, atmospheric waves, and atmospheric water.
Mars has lost most of its once-abundant water to space, leaving the planet cold and dry. In standard models, molecular hydrogen produced from water in the lower atmosphere diffuses into the upper atmosphere where it is dissociated, producing atomic hydrogen, which is lost. Using observations from the Neutral Gas and Ion Mass Spectrometer on the Mars Atmosphere and Volatile Evolution spacecraft, we demonstrate that water is instead transported directly to the upper atmosphere, then dissociated by ions to produce atomic hydrogen. The water abundance in the upper atmosphere varied seasonally, peaking in southern summer, and surged during dust storms, including the 2018 global dust storm. We calculate that this transport of water dominates the present-day loss of atomic hydrogen to space and influenced the evolution of Mars’ climate.