REAPPRAISING THE PRODUCTION AND TRANSFER OF HYDROGEN
TO THE UPPER ATMOSPHERE AT TIMES OF ELEVATED WATER
VAPOR
F. Montmessin, LATMOS/IPSL, UVSQ Université Paris-Saclay, Sorbonne Université, CNRS, Guyancourt, France, D. A.
Belyaev, Space Research Institute of the Russian Academy of Sciences (IKI RAS), Moscow, Russia, F. Lefèvre, LATMOS,
Sorbonne univ. Paris France, J. Alday, Open University, Milton Keynes, U.K., M. Vals, LATMOS, Guyancourt, A. A. Fedorova, O. I. Korablev, A. V. Trokhimovskiy, IKI Moscow, M. S. Chaffin, N. M. Schneider, LASP, Boulder, Colorado,
United States

Introduction: Water escape on Mars has recently undergone a paradigm shift with the discovery of
unexpected seasonal variations in the population of
hydrogen atoms in the exosphere where thermal escape occurs and results in water lost to space. This
discovery led to the hypothesis that, contradicting
the accepted pathway, atomic hydrogen in the exosphere was not only produced by molecular hydrogen but mostly by high altitude water vapor. Enhanced presence of water at high altitude during
southern spring and summer, due to atmospheric
warming and intensified transport, favors production
of H through photon-induced ion chemistry of water
molecules and thus appears to be the main cause of
the observed seasonal variability in escaping hydrogen (Chaffin et al., 2014; 2017; Clarke et al., 2014,
Heavens et al., 2018). This hypothesis is supported
by the large concentrations of water vapor observed
between 50 km and 150 km during the southern
summer solstice and global dust events (Fedorova et
al., 2020; 2021, Belyaev et al., 2021; Aoki et al.,
2019). Using a simplified air parcel transport model,
we have explored the formation of H from water
photolysis from 40 to 80 km and its vertical advection to the upper atmosphere. Comparing the injection modes of a variety of events (global dust storm,
perihelion periods, regional storm), we conclude that
southern spring/summer controls H production and
further ascent into the upper atmosphere on the long
term with direct implication for water escape.
Model Description: The assumption that H production above 80 km dominates from an escape
standpoint should be revisited by considering the
nature of transport (advection, not diffusion) and the
contribution of the atmosphere below. In that context, a different perspective would look at the fate of
a wet air parcel lifted from near the ground and progressively carried to an altitude range where H atoms
are produced and then diffuse up the exobase. This
supposes to adopt a Lagrangian standpoint, where
one tracks over time (i.e., altitude) the changing gaseous composition inside the parcel.
To this end, we have employed a hybrid approach combining observational results, photochemistry and transport diagnostics from a 3D Mars climate model to represent the processes affecting the
composition of an air parcel over time. The rationale
for this simplified approach is that the entire representation of H production, evolution, and transport

cannot be based solely on observations or the 3D
model. This approach of mixing observations and
modeling was introduced and used by Alday et al.
(2021) to establish the prevalence of the southern
spring/summer in the annual H production of the
middle atmosphere.
Observations can only indirectly constrain the
velocity of ascent in the middle atmosphere (Heavens et al., 2019), but they can constrain directly water vapor, pressure and temperature, which are the
main parameters to model the relevant photochemistry. On the other hand, given the current maturity of
Mars climate models (Navarro et al., 2014; Haberle
et al. 2019; Neary et al., 2020), using their transport
diagnostics is a relevant option for specifying the
range of vertical wind speeds needed for our model.

Figure 1: CO2 number density, H2O relative abundance and temperature derived from ACS data by
Belyaev et al. (2021) for the four cases investigated
in this study: MY34 GDS, Perihelion, MY35 Perihelion and C-storm.
Our hybrid model tracks the fate of a wet atmospheric parcel undergoing ascent and chemistry, using
the observed water vapor profiles reported in
Belyaev et al. (2021) and the chemical model of
Lefèvre et al. (2004) whose latest version has been
presented in Lefèvre et al. (2021). We aim at tracking the variation in H number density (NH) inside the
parcel during its ascent to the upper atmosphere at
times of intensified upward transport, i.e., during
perihelion and during the MY34 GDS. The parcel
ascends at a velocity ω of 10 cm/s as derived from
the Mars Climate database (Millour et al., 2017).
Dataset: One can distinguish between two
modes of high-altitude water vapor migration: periodic and stochastic. The periodic mode relates to
perihelion conditions as it occurs every year at the
same period of time and is only driven by the reproducible evolution of the southern spring/summer

climatic conditions, except when affected by the
occurrence of a GDS such as in MY28. The stochastic mode corresponds to the unpredictable occurrences of dust storms, whose impact on climate is
large-scale. This concerns global dust storms, and
large-scale A-, B-, C- storms. Both periodic and stochastic modes occur only during the second half of
the year, and it is now widely accepted that only
southern spring and summer contribute to water vapor migration to the upper atmosphere (Alday et al.,
2021).

and the corresponding upward flux (equal to NH ×
ω). NH shows a pronounced peak of 1.9 × 109 atoms.cm-3 at 69 km that is reminiscent of the net production peak found at 65 km. However, NH then decreases after rising above the peak as subsequent H
production diminishes and cannot compensate dilution subsequent to the parcel volume expansion. At
80 km, we find that the upward hydrogen flux is of
1.1 × 1010 cm-2s-1, a value that is nearly twice the net
photolysis product of H integrated from 80 to
100 km. The H flux at the bottom of the upper atmospheric domain is not only significant, it actually
accounts for a dominant portion of the H budget in
the upper atmosphere of the southern summer, when
H is known to escape massively.

Figure 2: Model results for the H number density in
the parcel (NH) and the resulting upward flux for the
MY34 perihelion case (black solid line).
Ls periods tested. We selected the same periods
of time as Belyaev et al. (2021) to which we added
the MY35 C-storm period from the same dataset (see
Figure 1). We thus based our simulations on data
collected during the climax of the MY34 GDS
(195°-220°), near perihelion (Ls 270°) in MY34 and
MY35, as well as during the MY35 C-storm (Ls
320°). We discarded the option to present a simulation for the MY34 C-storm (Ls 330°) because this
event was not sufficiently well covered to allow a
meaningful evaluation of the H atom transfer at that
time. Therefore, we conducted four simulations. Results were obtained for a particular latitude range
that theoretically corresponds to the main zone of H
ascent (i.e., 60°S for Perihelion and C-Storm, 0° for
the GDS).
Results:
MY34 Perihelion. Here we consider the particular
situation occurring during the transition between
southern spring and summer and that is henceforth
referred to as the Perihelion season and that encompasses the period from Ls 240° to 270°. Alday et al.
(2021) demonstrated the prevalence of the Perihelion
season for the formation of H and D atoms at 60 km
of altitude. This conclusion was based on applying
theoretical photolysis rates upon HDO and H2O profiles observed by ACS. Figure 2 shows the altitude
(or equivalently time, on the right axis) evolution of
the hydrogen number density (NH) in the air parcel

Figure 3: Predictions for a set of configurations

where production only occurs in a 5 km-thick layer
whose top altitude vary from 45 to 80 km. Plotted
quantity is the parcel H number density.
Hydrogen origin in the lower atmosphere. Our
air parcel model can be used to track the origin of the
H atoms advected to 80 km (Figure 3). To do so, we
restrict production inside 5-km atmospheric layers
located between 40 and 80 km (Figure 3). Our model confirms that no H atoms produced below 60 km
can access the upper atmosphere as they are bound to
recombine before reaching 80 km, which implies
that almost no atoms produced in the peak production region contribute to escape. After being advected from their production zone below 60 km, H atoms are completely lost after 1 or 2 km of ascent.
The production zone that matters for escape comprises altitudes from 65 to 80 km, with a dominant
contribution from the 70 to 75 km region.
Result Comparison. The MY34 configuration described above has revealed the main characteristics
of the H migration phenomenon, showing the significance of the H flux at 80 km and the partitioning of
the H origin among the atmospheric layers located
below and in particular between 60 and 80 km. We
then subsequently applied our model to the three
other configurations of interest: MY35 Perihelion,
MY34 GDS and MY35 C-storm (see Figure 1 and
Figure 4). Due to TGO’s orbit constraints, a large

part of the MY34 C-storm was not monitored by
ACS. The data gap actually covered the peak activity
of this regional storm, which implies our results
would have largely underestimated the C-storm upward flux, explaining our choice to solely focus on
the C-storm event of MY35. Yet, this event produced an intense brightening of the hydrogen corona
(Chaffin et al., 2021). In addition, the MY34
GDS/Perihelion, as well as the MY35 Perihelion
cases also suffer from a 5 to 10° sampling gap that
happened around their climax. However, because
these gaps are relatively short when compared to the
duration of these events and affected all of the main
events (GDS and Perihelion) in the same way, their
significance is likely to be small and their effect cannot bias the interpretation of one over the other.

Figure 4: A synthesis of the four configurations explored with the air parcel model: MY34
GDS/Perihelion, and MY35 Perihelion/C-storm. (A)
H Upward flux and concentration in the parcel during its rise from 20 to 100 km. (B) Same as (A) except NH is converted into H vmr (ppmv).
MY34 and MY35 Perihelion cases are in remarkable agreement, both exhibiting the same upward
flux profile (Figure 4) with a pronounced peak
around 70 km only slightly shifted upward in MY35.
This altitude offset explains why the MY35 80 km H
flux is ~30% larger than MY34, as a higher proportion of H atoms are produced closer to the 80 km
boundary. Indeed, water vapor is 15 ppmv more
abundant in MY35 at 80 km, which then directly
impacts H production and upward flux. These results
support the idea that the perihelion configuration is
highly reproductible from year to year, which provides confidence for extrapolating its contribution in
the escape budget on a longer term.
The GDS case exhibits a H flux profile that is
significantly flared compared to the perihelion case,
reflecting the distinct H2O vmr profile of the GDS
that shows a nearly uniform mixing ratio of ~70
ppmv below 70 km. The GDS H flux has a maximum value located at 65 km that is twice smaller
than for perihelion cases. At 80 km, GDS flux remains distinctly smaller than the two perihelion seasons considered here. However, at 85 km, the GDS
flux catches up with MY34 Perihelion flux. The
broadened peak of the GDS implies that a larger
proportion (~50%) of the H population origin is located close to the 80 km boundary. In addition, the

total density of the MY34 GDS above 80 km is
higher than for perihelion cases (Figure 4), which
mitigates the volume expansion (dilution) effect on
the air parcel in this altitude range.
The MY35 C-storm case produces a H upward
flux profile resembling that produced for perihelion,
likely in response to a circulation pattern that is a
remnant of the solstitial configuration, with a single
cell whose strength was recharged by the momentary
surge of dust. Its peak flux is located 5 km below the
flux peak of perihelion, yet is twice greater than that
of the GDS. However, both events end up generating
the same flux at 80 km.
Upward flux vs. neutral photolysis. We compared
the upward H flux in terms of hydrogen input into
the region above 80 km, with the direct deposition of
hydrogen from water photolysis above 80 km. On
the case of MY34 perihelion, the upward flux is
about twice the net column production of H atoms
out of photolysis between 80 and 100 km. For other
cases, we found the following: (i) for MY35 perihelion, the upward flux was 55% greater than local
photolysis production, (ii) for the MY34 GDS, the
upward flux was 30% greater, and (iii) for the MY35
C-storm, the upward flux was more than twice the
local production. These results imply that the upward
influx of H at the bottom of the upper atmospheric
domain is the dominant supplier in the context of the
neutral photochemistry.

Table 1: a synthesis of the results obtained in this
study and compared with other works (k19: Krasnopolsky, 2019; s20: Stone et al., 2020).
Discussion: Assuming that the H atoms crossing
the 80 km boundary can be then carried up to escape
altitudes, our results shed light on the potential of
each event for hydrogen escape. Table 1 summarizes
the production rates computed for most of the events
discussed here and for the various H production
modes that have been studied so far. Of all the processes involved in the production and injection of H
atoms into the upper atmosphere, the upward transfer
is found to be systematically greater than production
from water photolysis or ion chemistry. This statement is only valid outside the cold aphelion period,
where H2 molecules are the main precursor of H
atoms.
This raises the question of the fate of H atoms
crossing the 80 km boundary. Shaposhnikov et al.
(2022) explore the dynamical mechanisms that carry

volatiles into the upper atmosphere at GDS and perihelion times. Using a Mars’ GCM to address the
gravity wave breaking effect on global circulation
and the transport of water at high altitude, they show
that atmospheric updrafts are the main carrier of volatiles up to 100 km, above which molecular diffusion
combines with advection and then controls above
120 km the ascent of gases to the exobase. It is therefore legitimate to apply our advective transport model to hydrogen atoms produced below 80 km. Since
the lifetime of hydrogen atoms increases steadily
with altitude, once they have entered the upper atmospheric domain above 80 km, they are likely to
reside there long enough for a significant fraction of
them to reach the exobase. In fact, the way circulation is organized at times of high water events implies a massive upwelling either at the equator
(GDS) or at high southern latitudes (perihelion)
compensated by a massive downwelling at the
pole(s). During perihelion, the MCD indicates that
the downwelling velocity is twice larger than
upwelling (1.7, not shown, vs. 0.7 m/s maximum
values) yet is confined into a narrow band at the
north pole. Our air parcel concept only captures the
ascent part of the hydrogen journey into the upper
atmosphere. At 100 km, that is at the top of our
model, molecular diffusion will then take over the
rest of the ascent that will lead released H atoms to
escaping altitude. Yet a fraction of these atoms will
eventually return to the middle atmosphere via the
downwelling region.
Conclusion: We have used a 1D hybrid model to
represent the ascent of a wet air parcel at times of
intense dust and transport activity. This model combines observations of the ACS instrument that measured, for the first time, water vapor abundance from
20 to 120 km. These observations enable the indepth study of how the water vapor penetration to
high altitude contributes to hydrogen production
above 80 km. In contrast with other 1D models that
have been used to explore Mars’ photochemistry, our
model represents the vertical transport through advection with a constant velocity of 10 cm/s up to 100
km. Our results imply that, contrary to a common
assumption made in models used to study Mars’
photochemistry and escape processes, the region
between 60 and 80 km cannot be neglected in the
production and migration of hydrogen to the upper
atmosphere. In particular, these results imply that
upper atmosphere photochemistry models intending
to capture Southern Summer conditions need to carefully consider the flux boundary condition for H at
the lower boundary if it is higher than 80 km. Testing a variety of configurations, from the MY34 GDS
to the recent MY35 perihelion period, we have been
able to assess how the hydrogen upward flux from
above 60 km varies with events. Stochastic events
(GDS and A, B, C- storms) have a strong imprint on
the escape budget, but our results suggest perihelion

remains the dominant escape component on the long
term.
References:
Alday, et al. (2021). Nature Ast..
https://doi.org/10.1038/s41550-021-01389-x
Aoki, S., et al. (2019). J. of Geophys. Res.: Planets.
https://doi.org/10.1029/2019JE006109
Belyaev, D. A., et al. (2021). Geophys. Res. Let..
https://doi.org/10.1029/2021GL093411
Chaffin, M. S., et al. (2014). Geophys. Res. Let..
https://doi.org/10.1002/2013GL058578
Chaffin, M. S., et al. (2017). Nature Geos..
https://doi.org/10.1038/ngeo2887
Clarke, J. T., et al. (2014). Geophys. Res. Let..
https://doi.org/10.1002/2014GL061803
Fedorova, A. A., et al. (2020). Science.
https://doi.org/10.1126/science.aay9522
Fedorova, A. A., et al. (2021). J. of Geophys. Res.: Planets. https://doi.org/10.1029/2019JE006109
Haberle, R. M., et al. (2019). Icarus.
https://doi.org/10.1016/j.icarus.2019.03.026
Heavens, N. G., et al. (2018). Nature Ast..
https://doi.org/10.1038/s41550-017-0353-4
Krasnopolsky, V. A. (2019). Icarus,
https://doi.org/10.1016/j.icarus.2018.10.033
Lefèvre, F., et al. (2004). J. of Geophys. Res.: Planets.
https://doi.org/10.1029/2004JE002268
Lefèvre, F., et al. (2021). J. of Geophys. Res.: Planets.
https://doi.org/10.1029/2021JE006838
Millour, E., et al. (2017). The Mars Climate Database.
https://doi.org/10.1016/j.icarus.2014.12.034
Navarro, T., et al.. (2014). J. of Geophys. Res.: Planets.
https://doi.org/10.1002/2013JE004550
Neary, L., et al. (2020). Geophys. Res. Let..
https://doi.org/10.1029/2019GL084354
Shaposhnikov, D.S., et al. (2022). J. of Geophys. Res.:
Planets. https://doi.org/10.1029/2021JE007102
Stone, S. W., et al. (2020). Science.
https://doi.org/10.1126/science.aba5229