IMPROVED CLIMATOLOGIES OF LOWER AND MIDDLE ATMOSPHERIC
GRAVITY WAVE ACTIVITY AT MARS.
N.G. Heavens, Space Science Institute, Boulder, CO, USA and London, UK (nheavens@spacescience.org), A.
Pankine, Space Science Institute, J.M. Battalio, Earth and Planetary Sciences, Yale University, C.J. Wright, Centre
for Space, Atmospheric and Oceanic Science, University of Bath, D.M. Kass, A. Kleinboehl, S. Piqueux, NASA Jet
Propulsion Laboratory, Caltech, J.T. Schofield, JPL, Retired.

Introduction
Atmospheric gravity waves (GW) are excited by impulses in the atmospheric circulation; like the movement
of the wind over and around topography, jet stream meanderings, and convective activity. [1, 2, 3]. GW transfer energy and momentum through stable atmospheric
regions and thereby couple different levels of the atmosphere or different mesoscale weather systems [e.g., 4,
5, 6, 1, 7, 3]. In Mars’s atmosphere, GW seem to modulate the intensity of middle atmospheric downwelling
near the poles, the escape of constituents from the upper
atmosphere, and even the formation of mesoscale cloud
trails in the middle atmosphere [8, 9, 10].
GW processes are increasingly parameterized within
Mars general circulation/global climate models (GCM)
[e.g., 11, 12, 13, 14, 8]. These parameterizations largely
derive from ones originally applied to and/or validated
at the Earth and often have been validated by or tuned
to reproduce the zonal mean temperature structure of
the middle atmosphere [e.g., 11, 8], which can be significantly affected by mechanisms other than GW [15,
16, 17]. Another approach to parameterization development and evaluation has been to resolve GW explicitly
in higher resolution simulations [18, 19, 20, 21]. Observational validation of parameterized and resolved GW
forcing, however, has been limited.
The first purpose of this abstract is to provide a short,
integrated overview of the growing body of satellite observations of GW activity in Mars’s lower and middle
atmosphere that can aid direct understanding of GW
dynamics and validate modeling. Along the way, we
will highlight important results from a project currently
funded by NASA to expand climatologies of GW activity. These results would be the focus of our conference
presentation. Where appropriate, this discussion also
will consider complementary information from observations of upper atmospheric GW activity [e.g., 22, 23,
24, 25, 26, 27, 10, 28]. Note also the increasing study
of GW in surface weather observations [e.g., 29, 30].
History and types of observations
GW are fundamentally waves whose restoring force
is buoyancy. As they propagate, they perturb the temperature, wind, pressure, and density fields locally [1].
Their periods range from the inverse of the Brunt-Väisälä

frequency to the inverse of the Coriolis frequency. Martian atmospheric scientists generally are familiar with
some aspects of GW theory, because thermal tides can
be read and mathematically treated as both planetary
waves and GW [e.g., 31]. And so GW were first recognized as not quite tidal oscillations within the entry
profiles of the Viking landers [32]. GW also were recognized as the cause of lee wave clouds [33, 34].
Quantitative study of GW in the lower and middle
atmosphere has largely focused on extracting GW signals from observations of temperature variability. Many
studies have been based on studies of individual profiles,
generally using occultations to produce temperature profiles that are loosely analogous to entry profiles, just far
more frequently and inexpensively collected. GW activity in the lower atmosphere has been sensed using
radio occultation [35, 36]. A broader vertical range has
been accessed by ground-based stellar occultation [37],
stellar occultation in Mars orbit [38], and solar occultation in Mars orbit [39]. Recent studies by [38] and [39]
rely on temperature profiles that approximate the broad
lower to upper atmosphere range (from 20 to 120/160
km), vertical resolution, and temperature uncertainty of
entry profiles. The main downside is that occultations
are usually limited to < 10 per day, making it difficult to
isolate GW sources in map view. Single profile studies
also can sense convective instabilities associated with
GW growing to an unstable amplitude, breaking, and
depositing its momentum [35, 40, 41, 38, 39].
Other studies have looked at horizontal variability in temperature/brightness temperature, usually along
the dimension of spacecraft travel. In rare cases, GW
have been imaged, either in near-infrared dayglow by
OMEGA on board Mars Express [42] or in Mars Odyssey
THEMIS Band 10 imagery, which looks near the center
of the 15 µm CO2 band [43] (Fig. 1. The 15 µm CO2
band is also regularly measured by mapping spectrometers/radiometers like MGS-TES and MRO-MCS, and
GW signals have been inferred or explicitly searched for
in data from these instruments [44, 45, 46, 47, 48]. As
we will see in the next section, looking at multiple angles and at different baselines allows different portions
of the GW spectrum to be observed. Another advantage
of studying horizontal variability is the high data density
possible, typically more than two orders of magnitude
greater than possible due to occultation [47].

Improved Gravity Wave Climatologies

Figure 2: Theoretical sensitivity of limb observations to GW
activity following [49]. (a) Sensitivity to amplitude (%) for an
observation with a vertical resolution of 1.25 km, as in [35].
Note that [35] filtered out vertical wavelengths > 10 km; (b)
Sensitivity to GW potential energy (%) for observations at two
vertical resolutions at a vertical wavelength of 10 km.

Figure 3: Probability distributions of GW activity as a function of surface roughness in MCS channels sensitive to GW
in broad vertical ranges centered at 5 km (A1, top panel), 15
km (A2, middle panel), 25 km (A3, bottom panel). The white
line is a fit line appropriate for orographic GW activity associated with wind speeds of 1.5 to 3 ms−1 , or perhaps larger,
depending on the exact form of interaction with topography.
Originally published in [48].

Figure 1: Example of GW activity in ODY-THEMIS Band 10
imagery reprocessed as described in [43](Battalio et al., this
meeting). The 25 km horizontal wavelength waves propagating along the northwest to southeast direction between 40 and
130 km on the north-south axis are most obvious, but a mixture of other wavelength modes is also present.

Improved Gravity Wave Climatologies
The observational filter
Any type of GW observation only sees a portion of
the GW spectrum at one time. It is only by combining these different sources of information that a more
complete picture emerges [e.g., 50, 51]. This concept is
known as the observational filter.
For occultation-based single profile observations (approximated as limb observations), the filter is biased
toward long horizontal wavelengths and short vertical
wavelengths [52, 49] (Fig. 2a). By applying Eq. 1
of [49], it is possible to show that GW with horizontal
wavelengths < 100 km are largely invisible to occultation profiles (Fig. 2b). Single profile observations are
not necessarily dominated by GW with horizontal wavelengths < 400 km; such GW just contribute more to
the signal. However, the invisibility of short horizontal
wavelength waves may be consequential for evaluating
GW dynamical effects; GW momentum flux is inversely
proportional to horizontal wavelength [53]. Single profile studies also generally must select some portion of
the vertical wavelength spectrum (usually < 10 or 15 km
in order to extract GW distinct from tidal signals) [35,
39]. So a short period GW population with long vertical
wavelengths will be invisible. Note that momentum flux
is proportional to vertical wavelength [53].
The main controls on multiple profile studies are angle and baseline. Nadir observations are sensitive to
long vertical wavelengths, while more glancing angles
can access shorter vertical wavelengths [48]. In the
case of MRO-MCS (and to a lesser extent for MGSTES), radiance observations are discontinuous. Such
observations create baselines along which temperature
variability can be measured. These baselines effectively
remove the sensitivity of the observations to longer horizontal wavelengths. Shorter horizontal wavelengths can
be excluded by averaging. Judicious attention to these
considerations can allow particular parts of the spectrum
to be narrowly targeted. So for example, 27 continuous
limb scans using MRO-MCS can produce weak sensitivity (no more than 30% in amplitude) to GW activity
narrowly targeted at 100 to 200 km horizontal wavelength and near 5 km vertical wavelength. However,
caution must be exercised when relying on such weak
sensitivity, because a reduced signal may be overpowered by instrumental noise.
The orographic source dominates near the surface
One important source of GW is the interaction of
wind with topography, commonly known as the orographic source. Crucially, such waves tend to have
phase velocities close to zero, making it difficult for
them to propagate through an environment with light
winds, Mars’s significant relief suggested orographic
GW would be common on Mars, a hypothesis first explored but not confirmed by [35].

Using MRO-MCS off-nadir observations to look in
the first scale height, we were able to demonstrate that
GW activity close to the surface is a strong function of
roughness in high roughness areas [48]. The slope of
this relationship is theoretically predictable and consistent with a reasonable range of surface wind speeds (Fig.
3.). This relationship breaks down by an altitude of ≈ 25
km, suggesting orographic waves are strongly filtered in
the lower atmosphere, partly explaining why this relationship was hard to see in the 10 to 30 km altitude band
sampled by [35]. But data density also was important.
≈ 6.5 × 106 measurements were used to make Fig. 3,
while [35] was based on 7917 profiles.
Seasonal shifts in GW activity
One very simple result that has fallen out of multiple datasets are seasonal shifts of GW activity between hemispheres. These datasets do not always agree,
though. [35] found that southern tropical GW activity
weakened during its summer, whereas [47, 38] found the
opposite. [47] also found that high GW activity in the
extratropics shifted between hemispheres during their
respective winters.
Dust storms radically change GW activity
One surprising result from surveying off-nadir observations from MRO-MCS was that GW activity decreased during dust storm activity, particularly during
the planet-encircling dust event of MY 34 (hereafter
34P) [47]. This effect was particularly apparent in areas
of lower GW activity in the tropics. These areas tend to
be smoother and so are likely sources of non-orographic
waves from dry convection. Areas of high GW activity
in the extratropics were minimally affected. GW activity
also seems to increase locally in the tropics [47, 48].
GCM simulations of explicitly resolved GW also
predicted such a decrease, which is caused by suppression of convection and baroclinic activity by reduced
surface heating due to high dust opacity [21]. Comparison with [47] suggested that modeling was underestimating the decrease, but re-examination of this work has
accounted for the effects of opacity raising the vertical
sensitivity range of the MCS channels [48]. Taking this
effect into account, the factor of 2 decrease predicted
by modeling now agrees with the observations, though
some discrepancies in the spatial distribution of the effect still may remain and possibly may be explained by
differences in the part of the GW spectrum being sampled by the observations and the modeling [21].
One other prediction of [21] is that dust storm activity would improve GW propagation to the middle atmosphere and result in increased middle atmospheric GW
activity there. This result could explain enhancement of
upper atmospheric GW activity during 34P [10, 9]. And
indeed [48] showed that GW propagation in the trop-

Improved Gravity Wave Climatologies
ics did improve significantly during 34P. Decreased GW
activity in the lower atmosphere resulted from reduced
source strength near the surface. A preliminary analysis of lower and middle atmospheric GW activity from
MRO-MCS limb observations only found reduced lower
atmospheric GW activity during the planet-encircling
dust event in MY 28 (28P), but we expect to present an
improved analysis at this meeting.
A mesoscale peak in GW activity
As noted above, GW with shorter horizontal wavelengths and longer vertical wavelengths should dominate
momentum transport. MGS-TES nadir observations
have enabled us to characterize the horizontal wavelength spectrum of long vertical wavelength GW. As
also found by a similar analysis by [44], we find that
GW energy seems to peak at horizontal wavelengths <
100 km. These short horizontal wavelengths also seem
to be less filtered as they propagate in the lower atmosphere. There is also a clear scale separation between
variability at < 300 km horizontal wavelength and >
1000 km horizontal wavelength population that can be
attributed to tides and planetary waves.
There are two main implications of this result. First,
observational techniques that focus on the shorter horizontal wavelength end of the spectrum provide a more
accurate view of the dynamically significant GW population. Second, the lower atmospheric wave population
peaks at somewhat shorter horizontal wavelength than
the upper atmospheric population (< 100 km vs. 100
to 300 or 500 km) as inferred by [22, 24, 54], either
suggesting strong filtering of the short end of the population or that upper atmospheric GW are a secondary
population produced by GW breaking below the upper
atmosphere.
Acknowledgments
This work was funded by NASA’s Mars Data Analysis Program (80NSSC19K1215). Work at the Jet Propulsion Laboratory, California Institute of Technology, was
performed under a contract with the National Aeronautics and Space Administration. US Government sponsorship is acknowledged.
References:

[1] D. C. Fritts and M. J. Alexander. In: Rev.
Geophys. 41.1 (2003), pp. 3571–3589. [2] S. C. R. Rafkin. In:
JGR 114 (2009), E01009. [3] A. S. Medvedev and E. Yiğit.
In: Atmosphere 10.9 (2019). [4] S. E. Koch and P. B. Dorian.
In: Monthly Weather Review 116.12 (1988), pp. 2570–2592.
[5] J. R. Holton et al. In: Rev. Geophys. 33.4 (1995), pp. 403–
439. [6] M. Yamanaka. In: Adv. Space. Res. 15.4 (1995), pp. 47–
50. [7] E. Yiğit and A. S. Medvedev. In: Adv. Space Res.
55.4 (2015), pp. 983–1003. [8] G. Gilli et al. In: JGR Planets

125.3 (2020), e2018JE005873. [9] E. Yiğit et al. In: GRL 48.5
(2021), e2020GL092095. [10] V. Leelavathi, N. Venkateswara
Rao, and S. V. B. Rao. In: JGR Planets 125.12 (2020), e2020JE006649.
[11] F. Forget et al. In: JGR 104.E10 (1999), pp. 24155–24176.
[12] A. Akingunola. PhD thesis. York University, Canada, 2009.
[13] T. Kuroda et al. In: J. Meteor. Soc. Japan Ser. II 87.5
(2009), pp. 913–921. [14] A. S. Medvedev and E. Yiğit. In:
GRL 39, L05201 (2012), p. L05201. [15] R. J. Wilson et al.
In: GRL 35, L07202 (2008), p. L07202. [16] N. G. Heavens
et al. In: JGR Planets 116.E1 (2011). [17] T. McDunn et al.
In: JGR Planets 118.2 (2013), pp. 161–178. [18] T. Kuroda
et al. In: GRL 42 (2015), pp. 9213–9222. [19] T. Kuroda et
al. In: JAS 73.12 (2016), pp. 4895–4909. [20] T. Kuroda, E.
Yiğit, and A. S. Medvedev. In: JGR Planets 124.6 (2019),
pp. 1618–1632. [21] T. Kuroda, A. S. Medvedev, and E. Yiğit.
In: JGR Planets 125.11 (2020), e2020JE006556. [22] J. E.
Creasey, J. M. Forbes, and G. M. Keating. In: GRL 33, L22814
(2006), p. L22814. [23] S. L. England et al. In: JGR Space
Physics 122.2 (2017), pp. 2310–2335. [24] A. Siddle et al. In:
Icarus 333 (2019), pp. 12–21. [25] H. N. Williamson et al.
In: Icarus 331 (2019), pp. 110–115. [26] J. Liu, S. Jin, and
Y. Li. In: JGR Space Physics 124.11 (2019), pp. 9315–9334.
[27] M. K. Elrod et al. In: Ninth International Conference
on Mars, Pasadena, California, USA, 22-25 July 2019. 2019.
[28] E. Yiğit et al. In: GRL n/a.n/a (2020), e2020GL092095.
[29] D. Banfield et al. In: Nature Geoscience 13.3 (2020),
pp. 190–198. [30] S. D. Guzewich et al. In: JGR Planets 126.8
(2021). [31] Z. Wu et al. In: Nature Communications 11.1
(2020). [32] A. Seiff. In: Adv. Space Res. 2.2 (1982), pp. 3–17.
[33] R. Kahn. In: JGR 89.A8 (1984), pp. 6671–6688. [34] G. H.
Pettengill and P. G. Ford. In: GRL 27.5 (2000), pp. 609–612.
[35] J. E. Creasey, J. M. Forbes, and D. P. Hinson. In: GRL
33, L01803 (2006), p. L01803. [36] S. Tellmann et al. In:
JGR 118 (2013), pp. 306–320. [37] W. R. Saunders, M. J.
Person, and P. Withers. In: The Astronomical Journal 161.6
(2021), p. 280. [38] H. Nakagawa et al. In: JGR Planets 125.9
(2020), e2020JE006481. [39] E. D. Starichenko et al. In: JGR
Planets 126.8 (2021), e2021JE006899. [40] N. G. Heavens
et al. In: GRL 37.18 (2010). [41] M. Vals et al. In: PSS 178
(2019), p. 104708. [42] F. Altieri et al. In: JGR 117, E00J08
(2012), E00J08. [43] J. Battalio et al. In: Bulletin of the AAS
52.6 (Oct. 26, 2020). [44] T. Imamura, Y. Kawasaki, and T.
Fukuhara. In: JAS 64.5 (2007), pp. 1717–1726. [45] C. J. Wright.
In: Icarus 219 (2012), pp. 274–282. [46] A. G. Feofilov et
al. In: Icarus 221.2 (2012), pp. 949–959. [47] N. G. Heavens et al. In: Icarus 341 (2020), p. 113630. [48] N. G. Heavens et al. In: PSJ 3.3 (2022), p. 57. [49] Q. T. Trinh et al. In:
Atmospheric Measurement Techniques 8.3 (2015), pp. 1491–
1517. [50] D. L. Wu and S. D. Eckermann. In: JAS 65.12
(2008), pp. 3695–3718. [51] M. J. Alexander et al. In: Journal of Geophysical Research: Atmospheres 113.D15 (2008).
[52] P. Preusse et al. In: Journal of Geophysical Research: Atmospheres 107.D23 (2002), CRI 6-1-CRI 6–23. [53] M. Ern
et al. In: JGR 109, D20103 (2004), p. D20103. [54] N. Terada
et al. In: JGR Space Physics 122.2 (2017), pp. 2374–2397.