GLOBAL 4-D INVESTIGATION OF WATER DURING THE MARS YEAR
34 DUSTY SEASON FROM A MULTI-SPACECRAFT ASSIMILATION
J. A. Holmes, S. R. Lewis, M. R. Patel, J. Alday, School of Physical Sciences, The Open University, Milton
Keynes, UK (james.holmes@open.ac.uk), S. Aoki, Graduate School of Frontier Sciences, The University of Tokyo, Kashiwa, Japan, G. Liuzzi, G. L. Villanueva, NASA Goddard Space Flight Center, Greenbelt, MD, USA,
M. M. J. Crismani, California State University, San Bernardino, Department of Physics, CA USA, A. A. Fedorova, O. Korablev, Space Research Institute of the Russian Academy of Sciences (IKI RAS), Russia, K. S. Olsen, Department of Physics, University of Oxford, Oxford, UK, D. M. Kass, Jet Propulsion Laboratory, California Institute of Technology, USA, A. C. Vandaele, Royal Belgian Institute for Space Aeronomy, Belgium.

Introduction:
To understand the evolving martian water cycle
and variations in transport of water throughout the
atmosphere on several timescales (diurnal to seasonal), a global perspective of the combined vertical and
horizontal distribution of water is needed. The global
4-D vertical water vapour distribution is investigated
through a reanalysis that unifies water, temperature
and dust retrievals from several instruments on multiple spacecraft throughout Mars Year (MY) 34 with
a global circulation model. The global dust storm
and southern summer regional dust storm events
pushed water vapour higher in altitude across all
latitudes, and supersaturated water vapour is found
to penetrate the northern winter polar vortex. This
analysis provides new insights into water loss from
the atmosphere throughout time and helps to further
constrain estimates of the past water inventory on
Mars
The past inventory of water on Mars is largely
unconstrained as a result of the currently incomplete
understanding of how much water has escaped from
the atmosphere over time and the processes through
which this phenomenon occurs [1,2]. More complete
knowledge of how the water inventory has been altered over time requires a better understanding of the
past and present atmospheric water cycle and vertical
mobility of water vapour. Over the past 50 years
orbital and, more recently, surface observations
gathered by several missions have helped to build up
a more complete description of the seasonal trend
and interannual variations in the water cycle through
observations of the water vapour column by several
different spacecraft [3-9]. One of the outstanding
questions regarding the water cycle is a complete
understanding of the seasonal evolution of the vertical distribution of water vapour that has a large impact on the transport of water in the atmosphere and
surface-regolith interactions [10].
Until recently, the seasonal evolution of the water vapour profile had been observed primarily from
spatially sparse SPICAM profiles [7,11] that discovered the presence of supersaturation in the atmosphere of Mars. Supersaturation allows water vapour
to propagate through the hygropause, impacting the
escape of water from the martian atmosphere. Sys-

tematic mapping of the presence of supersaturation,
in particular across a complete diurnal cycle, has not
yet been conducted but is important to understand
how prevalent supersaturation is on Mars. Recent
increasingly expansive water vapour profile datasets
have been retrieved from instruments on the ExoMars Trace Gas Orbiter (TGO) spacecraft [12,13]
illustrating the impact of a global dust storm event
on the water vapour vertical structure and further
evidence of supersaturation. However, restrictions in
space and time mean they cannot provide a truly
global perspective.
A global circulation model (GCM) can be utilised for interpretation and understanding of the observed seasonal and latitudinal variations in a global
sense. They do however also contain errors of representation associated with computational constraints,
meaning it is impossible to completely represent all
the physical processes occurring across the globe on
local scales (10s of kilometres and below); GCMs
can never fully replicate the evolving atmosphere of
Mars. Obtaining the best possible analysis of the
global evolution of the water cycle requires combining observational data with a Mars GCM. This technique, called data assimilation, is widely used in
conjunction with Earth GCMs and three different
Mars reanalysis datasets are now publicly available
[14,15,16].
In this study we investigate the global distribution of water vapour and the saturation state of the
atmosphere throughout LS = 159-358° in MY 34
utilising a global reanalysis that combines spacecraft
observations from several instruments with a Mars
GCM.
Spacecraft observations:
This study uses a multi-spacecraft assimilation approach that combines a Mars GCM with retrievals
from the ExoMars TGO and Mars Reconnaissance
Orbiter (MRO) spacecraft currently in orbit around
Mars. Retrieval datasets of water vapour vertical
profiles [12,17] and water vapour column [18] are
included from the Nadir and Occultation for Mars
Discovery (NOMAD) instrument. The Atmospheric
Chemistry Suite (ACS) retrieved temperature and
water vapour profiles in the near-infrared [13] and
mid-infrared [19,20] that are also incorporated into

the reanalysis. Mars Climate Sounder (MCS) [21]
retrieved temperature profiles are also incorporated
into the assimilation [22-24].
Differences in the orbits of these spacecraft result
in a largely complementary expanded dataset when
combined due to the minimal overlap in data at the
same local time and spatial location, with the coverage of all datasets shown in Figure 1. Data assimilation therefore is a highly efficient method in these
circumstances to provide the best possible reproduction of the global atmospheric evolution during this
time period.

Figure 1 – Coverage of the (a) water vapour and (b)
thermal observations included in the global reanalysis.
Modelling and data assimilation:
The reanalysis created through this study incorporates a Mars GCM used by the Open University
(OU) modelling group that has been developed in a
collaboration between the LMD, the OU, the University of Oxford and the Instituto de Astrofı́sica de
Andalucı́a. Physical parameterisations [25] and the
LMD photochemical module [26,27] both shared
with the LMD Mars GCM are coupled to a spectral
dynamical core and semi-Lagrangian advection
scheme [28]. Specific physical parameterisations
linked to the water cycle include the cloud microphysics package [29] that incorporates nucleation on
dust particles and supersaturation, an implementation
for radiatively active clouds [30], a semi-interactive
two-moment scheme to transport dust [31] and a
thermal plume model for better representation of
turbulence in the planetary boundary layer [32].
The reanalysis produced by this study utilises the
Mars GCM truncated at wavenumber 31 (resulting in
a 5° longitude-latitude grid in the horizontal for
physical variables) with 70 vertical sigma levels that
can extend to around 100 km.
To combine the retrieval datasets with the Mars
GCM, the Analysis Correction (AC) scheme 3[3] is
used (with necessary parameters adapted to martian
conditions) that has previously been used to investigate several martian atmosphere scientific topics.
The AC scheme is a form of successive corrections
in which analysis steps are interleaved with each
model dynamical time step. In each analysis step, the
analysis increments are spread from the observation
locations to the surrounding model grid points with
subsequent derivation of multi-variate increment
fields for dynamical balance where applicable (e.g.
after assimilating temperatures, balanced thermal

wind increments are applied).
Validation of the reanalysis dataset:
To validate the water vapour distribution produced by the reanalysis, it was compared against
SPICAM data [9] that were not assimilated. An added advantage of data assimilation is that the reanalysis can be compared to retrievals that are not necessarily in the same location as the retrievals combined
with the Mars GCM, but nevertheless are still influenced by assimilation of temperature profiles, dust
column, and potentially even water vapour profiles
in the past.
Figure 2 displays a comparison of SPICAM profiles to the reanalysis and a free running GCM at the
closest spatial location and local time. The reanalysis
is efficient at reducing the overestimation of water
vapour at higher altitudes that is a common problem
for GCMs that include supersaturation and is also in
very good agreement with the abundance and shape
of SPICAM profiles. The validation indicates the
reanalysis dataset is the optimal dataset to use for
investigating the water cycle, as it captures realistic
vertical structures in the atmospheric water profiles.

Figure 2 – Comparison of water vapour profiles
from SPICAM (red) with independent coincident
water vapour profiles in the reanalysis (blue) and a
free running GCM (black). Red horizontal bars indicate uncertainty on each altitude point of the retrieval.
Results:
The vertical distribution of water is highly variable and dependent on the season and latitudinal location, as illustrated in Figure 3 which displays the 30°
zonally averaged vertical profile of water vapour at a
selection of different latitude bands (30° latitudinal
means) over time. The global dust storm that occurred from LS = 180-225° and the intense southern
summer regional dust storm from LS = 320-335°
both provided an increased abundance of water vapour to the upper atmosphere at all latitudes including the northern polar region. During the onset of the
MY 34 global dust storm, wave activity controls the
transport of water poleward of 30°N and 30°S, with
wave activity generally acting to transport water
vapour polewards in each hemisphere. The hygro-

pause (defined as the highest point at which water is
present at 50 ppmv or above) has risen considerably
since the start of the simulation and is beyond 50 km
during the initiation phase of the global dust
storm.

Figure 3 – Zonally-averaged water vapour profiles
averaged over 30° latitudinal bands in the water reanalysis. The red line approximates the hygropause
location (50 ppmv).
Sublimation of water ice releasing vapour from
the southern polar cap in southern summer is evident
from LS = 255-315° in the 60°S-90°S latitude band.
While northern polar latitudes (90°N-60°N latitude
band) are largely absent of water vapour below 20
km during the dusty season of MY 34, water vapour
is found in a middle atmosphere layer reaching the
northern pole. The variations in water vapour above
20 km are linked to transport from mid-latitudes via
stationary and transient eddies during a global dust
storm, perihelion season and the intense MY 34
southern summer regional dust storm. Water vapour
at the start of the simulation in the 60°S-90°S latitudinal band is only seen in a band at around 40 km
altitude. This water vapour layer is linked to the anticlockwise cell transporting water vapour from lower
latitudes into this region.
New evidence is found of water vapour, in a supersaturated state, breaking into the northern winter
polar vortex. Figure 4 displays the vertical structure
of northern hemisphere water vapour and atmospher-

ic temperature during northern polar winter focused
on a 10-sol period around LS= 250°. The increased
water vapour abundance (above 50 ppmv) directly
over the north pole above around 40 km is linked to
the polar warming and meridional circulation pattern
during this time of year in which water vapour is
transported northward by transient and stationary
eddies rather than the mean meridional circulation.
This finding has implications for the surfaceatmosphere exchange of water between the northern
polar cap and the large-scale atmosphere (via
transport of water vapour) and would also have a
potential impact on how much water was present in
the past atmosphere of Mars if the breaking of supersaturated water into the northern polar vortex is happening repeatedly each Mars year. The results here
suggest that the seasonal flux of water vapour into
the northern polar cap could be larger than previously thought.

Figure 4 – Zonally-averaged latitude-altitude crosssection of water vapour vertical distribution and atmospheric temperature (white contours) at LS = 250°
in the reanalysis. Red dots indicate the presence of
supersaturated water vapour.
References:
[1] Villanueva et al. (2015), Science Advances, 7, 7.
[2] Jakosky et al. (2018), Icarus, 315, 146-157. [3]
Smith et al. (2004), Icarus, 167, 148-165. [4] Tschimmel et al. (2008), Icarus, 195, 557-575. [5] Smith
et al. (2009), J. Geophys. Res. (Planets), 114,
E00D03. [6] Sindoni et al. (2011), Planet. Space
Sci., 59, 149-162. [7] Maltagliati et al. (2011), Science, 333, 1868. [8] Trokhimovskiy et al. (2015),
Icarus, 251, 50-64. [9] Fedorova et al. (2021), J.
Geophys. Res. (Planets), 126, e06616. [10] Steele et
al. (2017), Icarus, 289, 56-79. [11] Maltagliati et al.
(2013), Icarus, 223, 942-962. [12] Aoki et al. (2019),
J. Geophys. Res. (Planets), 124, 3482-3497. [13]
Fedorova et al. (2020), Science, 367, 297-300. [14]
Montabone et al. (2014), Geosci. Data J., 1, 129139. [15] Greybush et al. (2019), Geosci. Data J., 6,
137-150. [16] Holmes et al. (2020), Planet. Space.

Sci., 188, 104962. [17] Villanueva et al. (2021), Science Advances, 7, eabc8843. [18] Crismani et al.
(2021), J. Geophys. Res. (Planets), 126,
e2021JE006878. [19] Alday et al. (2021), Nature
Astronomy, 5, 943–950. [20] Olsen et al. (2021),
Astron. Astrophys., 647, A161. [21] McCleese, D. J.
et al. (2007) JGR, 112, E05S06. [22] Kleinböhl, A.
et al. (2009) JGR, 114, E10006. [23] Kleinböhl, A. et
al. (2011) JQSRT, 112, 1568-1580. [24] Kleinböhl,
A. et al. (2017) JQSRT, 112, 511-522. [25] Forget et
al. (1999), J. Geophys. Res., 104, 24155-24176. [26]
Lefèvre et al. (2004), J. Geophys. Res., 109, E07004.
[27] Lefèvre et al. (2008), Nature, 454, 971-975.
[28] Newman et al. (2002), J. Geophys. Res., 107,
5123. [29] Navarro et al. (2014), J. Geophys. Res.
(Planets), 119, 1479-1495. [30] Madeleine et al.
(2012), Geophys. Res. Lett., 39, L23202. [31] Madeleine et al. (2011), J. Geophys. Res. (Planets), 116,
11010. [32] Colaı̈ tis et al. (2013), J. Geophys. Res.
(Planets), 118, 1468-1487. [33] Lorenc et al. (1991),
Quart. J. Royal Meteorol. Soc., 117, 59-89.