MACDA II: A NEW REANALYSIS FOR THE MARTIAN ATMOSPHERE
USING VERTICALLY-RESOLVED DUST OPACITY OBSERVATIONS
P. L. Read, A. Valeanu, Ruan Tao, Atmospheric, Oceanic & Planetary Physics, University of Oxford, UK (peter.read@physics.ox.ac.uk), L. Montabone, Space Science Institute, Boulder, CO, USA, & Laboratoire de
Météorologie Dynamique, IPSL, Sorbonne Université, Paris, France, S. R. Lewis, The Open University, UK, R.
M. B. Young, Department of Physics & National Space Science and Technology Center, UAE University, Al
Ain, United Arab Emirates.
Introduction: The dust cycle is a key component
of the Martian climate, and is extremely important for
understanding the interannual, seasonal and synoptic
evolution of the Martian environment. (e.g., Kahre et
al., 2017; Newman et al., 2002a, and references
therein). Intensive measurements of atmospheric temperature and dust extending over more than eleven
Mars years (MY) now exist with unprecedented spatial coverage, thanks to various orbital spacecraft.
Such observations have already helped to improve our
understanding of Mars' weather and climate. However, the incomplete coverage of these measurements
across the planet constrains our ability to study the
general circulation in full detail, particularly those aspects related to dust opacity.
On the other hand, numerical models provide
four-dimensional simulated data with moderate to
high temporal and spatial resolution and complete
coverage in space and time, but often fail to reproduce the dust cycle's full range of variability. Even
the most sophisticated free-running GCMs still
struggle to capture realistic interannual variability
associated with dust lifting and transport.
To aid in this task, data assimilation has become
an optimal approach to provide a solution that is consistent with both observations and modeled physical
constraints. Data assimilation corrects model-predicted variables toward observations such that the resulting solution can represent the full observed variability of the climate. Such an assimilated record is
often termed a “reanalysis” by analogy with the
practice in Earth weather and climate forecasting.
Several publicly available reanalyses of observations of the Martian atmosphere have been produced
in recent years, based mainly on remote sounding
measurements of atmospheric temperature, dust and
ice opacity and chemical constituents from various
orbital platforms (e.g. Montabone et al. 2014; Greybush et al. 2019; Holmes et al. 2020). However, almost all of these have so far mainly used measurements of column dust opacity without information on
the vertical distribution of dust. Such products provide much useful information on how dust evolves
during the Martian year, but may misrepresent some
important features, such as elevated layers of dust
(Heavens et al. 2011), and give no information on
the vertical extent of dust loading in the atmosphere.
In the present work, we have extended the Analysis Correction assimilation scheme (Lorenc et al.

1991), as used for the MACDA and OPENMARS reanalyses, to make use of both column-integrated
(CIDO) and layer-integrated dust opacity (LIDO),
such as obtained from Mars Climate Sounder limb
observations. Here we outline the new assimilation
scheme and present some results (a) that validate the
reanalysis with independent observations and (b) that
demonstrate significant improvements in the representation of the 3D distribution of dust opacity in the
Martian atmosphere, even when this distribution differs markedly from the long term climatology.
Outline of the Assimilation Scheme: In this
work the model used is based on the UK version of a
three-dimensional Martian Global Climate Model
(UK-LMD MGCM, v5.1.3) (Forget et al., 1999; Mulholland et al., 2013). The model combines a spectral
dynamical solver at triangular truncation T31, corresponding to a 96 x 48 longitude-latitude grid in real
space, a tracer transport scheme and dust lifting and
deposition routines, along with a full range of physical parameterizations.
The observations used in the present reanalysis
mainly comprise the retrievals of atmospheric temperature and dust opacity from the Mars Climate
Sounder (MCS) instrument on board the Mars Reconnaissance Orbiter spacecraft (Kleinböhl et al.
2009). These consist of vertical profiles with a vertical resolution of around 5 km and a horizontal foot

Figure 1: Spatial and temporal distribution of
available dust opacity data from Mars Climate
Sounder during MY28 and 29. The color scales show
the number of measurements in 5o Ls and 3o latitude
bins.

the layer-integrated dust opacity (LIDO) within each
Initial condition
!! ", $ , %" ", $ , &" (", $)

Dynamics
integration

Initial condition
mass mixing ratio
)" (", $)

Temperature observations & *+, (", -)
Dust opacity observations
*+,
*+,
CIDO ,#$%
(") and/or LIDO ,&'()
(", -)

! ", $ , % ", $ , &(", $)

Physics
integration

Dust
transport
scheme

Dust opacity ) ", $ ,
column dust opacity at
reference level ,#$% (")
Layer-integrated dust
opacity ,&'() (", $)

Analysis
Correction

) ", $
⟹ ) ", $ *(", $)

Layerized
dust scaling
factor
*(", $)

Corrected ! ", $ ,
% ", $ , & ", $ ,
,#$% ("), ,&'() (", $)

! ", $ , % ", $ , &(", $)

Figure 2: Sequence of operations in the new data assimilation scheme with active dust transport. Green boxes
show initial conditions and input observations, blue boxes show data generated by Mars General Circulation
Model (MGCM) integration, and red boxes show individual MGCM modules. Text in black applies to both column integrated dust opacity (CIDO) and layer-integrated dust opacity (LIDO) assimilation, text in blue applies
to CIDO assimilation only, and text in red applies to LIDO assimilation only. Only variables related to the data
assimilation scheme are included.
print of around 250 km, covering an altitude range
from near the surface up to around 80 km. An additional vertically integrated dust opacity is also derived from the dust opacity profiles to provide an estimate of the total column opacity using the methodology described by Montabone et al. (2015,2020). A
typical horizontal distribution of available data is illustrated in Figure 1 for MY28 and 29, which we use
below to validate the reanalysis and demonstrate its
capabilities.
The data assimilation scheme is based on the
analysis correction sequential estimation (AC)
scheme (Lorenc et al., 1991) but with modifications
specific to Mars (Lewis et al., 2007; Ruan et al.
2021a). The assimilation step is computationally inexpensive compared with the rest of the model, and
so is usually performed at each dynamical timestep
(typically of 3 min). Lewis et al. (2007) describe the
scheme in full detail. Temperature assimilations are
the same as in that work, except for the observational
data set used. In this work we extend the dust assimilation to incorporate advective transport of radiatively active dust in the simulation model as well as
to assimilate both CIDO and LIDO observations. An
identical free-running simulation without any assimilation, but with a fully active dust lifting and
transport cycle tuned to reproduce plausible seasonal
variations of dust loading (cf. Mulholland et al.,
2013; Newman et al., 2002), was run in parallel.
For dust assimilation, the sequence of operations
is described in detail by Ruan et al. (2021a) and illustrated in Figure 2. First the dynamics timestep is
integrated, and then the dust is advected to obtain
the three dimensional distribution of dust mass mixing ratio q(x, m). The CIDO at position x, predicted
from this distribution (τC) is obtained by summing

model layer (τLIDO). The reference column dust opacity τref at a reference pressure pref is then determined
from the model by rescaling by the ratio of local surface pressure to pref. To compare the results with observations, modeled CIDO values are rescaled to 610
Pa. The dust transport scheme transports a 3D dust
mass mixing ratio field, but the assimilated CIDO
dust observations constrain τref only, so the dust mass
mixing ratio at each model layer must be adjusted after the assimilation. This adjustment simply consists
of a multiplicative scale factor λ(x), which ensures
that the shape of the vertical dust profile at each horizontal grid point remains the same before and after
the assimilation of τref.
LIDO observations are assimilated separately and
the approach used here resembles the assimilation of
thermal profiles into the UK-LMD MGCM (Lewis
et al., 2007). First, we use the modeled dust opacities
τLIDO(m) to predict the dust opacity within each observation layer i. Model layers that overlap more
than one observed layer are split linearly in ln p
among the observed layers. The horizontal assimilation then uses these increments, summed over contributions from each observation level i, to update the
modeled dust field. Dust is transported in terms of
dust mass mixing ratio, so the assimilation needs to
correct this quantity. As dust mass mixing ratio is
proportional to LIDO, it is multiplied by a factor of
h = t’LIDO(x,m)/ tLIDO(x,m), where the primed and
unprimed quantities are the corrected and uncorrected LIDO values.
The advected dust opacity and mass mixing ratio
fields are then used to integrate the physical parametrizations. Finally, τref and T are updated using
data assimilated by the AC scheme, followed by increments to u and v in thermal wind balance.

Reanalysis verification: The new reanalysis has
been verified by comparison of the reanalysed fields
with both in-sample and out-of-sample observations,
e.g. using data from other spacecraft. Figure 3 shows
one example where the reanalysis, interpolated to the
location of the Spirit and Opportunity rovers (shown
in magenta), has been compared with predictions
from a free-running model (without assimilation –
shown in green) and the measured opacity from each
rover (at 880 nm) in blue symbols.

Figure 4: Night-time (18:00-06:00 local time) zonaltime mean dust opacity (km) during MY28 Ls =
122.5◦, with detached dust layers. From top: MCS observations, free-running model, CIDO-only reanalysis, joint CIDO/LIDO reanalysis. Time averages are
over 5o Ls.
The binned raw data in the top frame exhibits a
case with a pronounced elevated dust layer at low latitudes (Heavens et al. 2011) which is not captured either by the free-running GCM or by assimilating
CIDO observations alone. But the elevated layer is
well captured in the bottom frame, which was obtained by assimilating both CIDO and LIDO observations together. This clearly shows the importance of
assimilating the vertical structure of the dust distribution, which would otherwise lead to the reanalysis
misrepresenting the dust and corresponding heating
and cooling rates. Further details can be found in
Ruan et al. (2021a).

Figure 3: τref from the reanalysis (magenta) and
free-running model (green) compared with (a) Spirit
and (b) Opportunity Pancam observations. Each
point is averaged over one sol.
The reanalysis is clearly shown to capture most of
the variations in dust optical depth during MY28 and
29, including the large planet-encircling dust storm
around LS = 270o-300o in MY28 and smaller regional
dust storms at other times in MY29.
The advantage of assimilating LIDO profiles from
MCS as well as total column dust opacities is clearly
evident in Figure 4. This shows zonally averaged
maps of dust opacity produced by a free-running
model and assimilations that use either CIDO only
data or combined CIDO and LIDO observations from
MCS, compared with raw binned MCS observations
alone.

MACDA II: The assimilation scheme and model
described above have been used to assimilate temperature and dust observations from MCS and the earlier
dataset from the Mars Global Surveyor/Thermal
Emission Spectrometer instrument used for the original MACDA reanalysis (Montabone et al. 2014), covering the whole period from LS = 141o in MY24 until
the end of MY35.
The new dataset retains the same T31 (72 x 36
gridpoints in longitude x latitude) horizontal resolution as MACDA v1.0 and on 25 vertical levels on terrain-following s-levels. Fields and variables provided
in the new reanalysis include everything that was provided in the MACDA v1.0 dataset (Montabone et al.
2014), plus additional fields representing the 3D timevarying dust distribution, atmospheric density and
vertical velocity, together with surface wind stress
and dust lifting rates, all provided 12 times per sol.
The new MACDA II reanalysis dataset will be archived as CF-compliant netCDF files and made publicly available to registered users via the Centre for

Environmental Data Analysis (CEDA) group
(https://archive.ceda.ac.uk/). An earlier version of this reanalysis using MCS data and total column dust opacity
measurements from the THEMIS instrument on Mars
Odyssey for MY28-29 only is available on Zenodo
(Ruan et al. 2021b).
Acknowledgements: PLR and AV were also
funded by the UK Space Agency (ST/R001499/1).
SRL also thanks the UK Space Agency
(ST/R001405/1, ST/V005332/1, ST/S00145X/1 and
ST/T002913/1) and the European Commission
(Horizon 2020, Ref: 633 127) for support in related
areas. LM acknowledges support from NASA's Mars
Data Analysis Program under grant no.
NNX13AK02G. RMBY acknowledges support from
UAE University grants G00003322 and G00003407.
RMBY, PLR, and SRL were part-funded by ESA
contract 4000 114 138/15/NL/PA.

References:
Forget, F., Hourdin, F., Fournier, R., Hourdin, C.,
Talagrand, O., Collins, M., et al. (1999). Improved
general circulation models of the Martian atmosphere from the surface to above 80 km. Journal of
Geophysical Research, 104(E10), 24155–24175.
https://doi.org/10.1029/1999JE001025
Greybush, S. J., Kalnay, E., Wilson, R. J., Hoffman,
R. N., Nehrkorn, T., Leidner, M., et al. (2019). The
Ensemble Mars Atmosphere Reanalysis System
(EMARS) Version 1.0. Geoscience Data Journal,
16, 137–150. https://doi.org/10.1002/gdj3.77
Heavens, N. G., Richardson, M. I., Kleinb hl, A.,
Kass, D. M., McCleese, D. J., Abdou, W., et al.
(2011). Vertical distribution of dust in the Martian
atmosphere during northern spring and summer:
High-altitude tropical dust maximum at northern
summer solstice. Journal of Geophysical Research,
116, E01007. https://doi.org/10.1029/2010JE003692
Holmes, J. A., Lewis, S. R., & Patel, M. R. (2020).
OpenMARS: A global record of Martian weather
from 1999 to 2015. Planetary and Space Science,
188, 104962.
https://doi.org/10.1016/j.pss.2020.104962
Kahre, M. A., Murphy, J. R., Newman, C. E., Wilson, R. J., Cantor, B. A., Lemmon, M. T., & Wolff,
M. J. (2017). The Mars Dust Cycle. In R. M.
Haberle, R. T. Clancy, F. Forget, M. D. Smith, & R.
W. Zurek (Eds.), The atmosphere and climate of
mars (pp. 295–337). Cambridge University Press.
Cambridge, UK.
Kleinböhl, A., Schofield, J. T., Kass, D. M., Abdou,
W. A., Backus, C. R., Sen, B., et al. (2009). Mars
Climate Sounder limb profile retrieval of atmospheric temperature, pressure, and dust and water ice
opacity. Journal of Geophysical Research, 114,

E10006. https://doi.org/10.1029/2009JE003358
Lewis, S. R., Read, P. L., Conrath, B. J., Pearl, J. C.,
& Smith, M. D. (2007). Assimilation of thermal
emission spectrometer atmospheric data during the
Mars Global Surveyor aerobraking period. Icarus,
192, 327–347. https://doi.org/10.1016/j.icarus.2007.08.009
Lorenc, A. C., Bell, R. S., & Macpherson, B. (1991).
The Meteorological Office analysis correction data
assimilation scheme. The Quarterly Journal of the
Royal Meteorological Society, 117(497), 59–89.
https://doi.org/10.1002/qj.49711749704
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. https://doi.org/10.1016/j.icarus.2014.12.034
Montabone, L., Marsh, K., Lewis, S. R., Read, P. L.,
Smith, M. D., Holmes, J., et al. (2014). The Mars
Analysis Correction Data Assimilation (MACDA)
Dataset V1.0. Geoscience Data Journal, 1(2), 129–
139. https://doi.org/10.1002/gdj3.1310.1002/gdj3.13
Montabone, L., Spiga, A., Kass, D. M., Kleinboehl,
A., Forget, F., & Millour, E. (2020). Martian year 34
column dust climatology from Mars Climate
Sounder observations: Reconstructed maps and
model simulations. J Geophys. Res.: Planets, 125,
e2019JE006111.
https://doi.org/10.1029/2019JE006111
Mulholland, D. P., Read, P. L., & Lewis, S. R.
(2013). Simulating the interannual variability of major dust storms on Mars using variable lifting
thresholds. Icarus, 223, 344–358.
https://doi.org/10.1016/j.icarus.2012.12.003
Newman, C. E., Lewis, S. R., Read, P. L., & Forget,
F. (2002). Modeling the Martian dust cycle 1. Representations of dust transport processes. Journal of Geophysical Research, 107, 5123.
https://doi.org/10.1029/2002je001910
Ruan, T., Young, R. M. B., Lewis, S. R., Montabone, L., Valeanu, A., & Read, P. L. (2021a). Assimilation of both column- and layer-integrated dust
opacity observations in the Martian atmosphere.
Earth and Space Science, 8, e2021EA001869.
https://doi.org/10.1029/2021EA001869
Ruan, T., Read, P. L., Montabone, L., Lewis, S. R.,
Young, R. M. B., & Valeanu, A. (2021b). A vertically-resolved atmospheric dust reanalysis for
Mars Years 28-29 using Analysis Correction (Version 1). Zenodo. https://doi.org/10.5281/zenodo.5517308