MULTI-YEAR EFFECTS OF GLOBAL DUST STORMS ON THE POLAR HEAT
BUDGET AND CO2 SNOWFALL ON MARS.
P. O. Hayne, N. Alsaeed, V. Concepcion, Laboratory for Atmospheric and Space Physics, University of Colorado
Boulder (paul.hayne@colorado.edu), D. M. Kass, S. Piqueux, J. Bapst, A. Kleinboehl, NASA - Jet Propulsion
Laboratory, California Institute of Technology.

Introduction
Planetary scale dust storms are known to have measurable, though typically short-lived effects on the climate and atmospheric circulation of Mars. Occurring
on roughly decadal timescales, these global events increase planetary albedo, warm the surface at night, and
transport water to high altitudes, contributing to its loss
from the upper atmosphere [1, 2].
Here, we report multi-year impacts of dust storms
on polar temperatures, CO2 snowfall, and the extent and
retreat of the seasonal polar caps. We focus primarily on
the 2007 (Mars Year 28) global dust event (GDE) using
data from the Mars Climate Sounder (MCS) [3]. The
polar regions are fundamental to the global CO2 cycle
on Mars, controlling the ∼30% annual variation in total
atmospheric mass, and buffering mean surface pressures.
Therefore, changes in the polar regions induced by GDE
could have measurable effects on the climate elsewhere
on the planet.
Previous work demonstrated that GDE cause changes
in atmospheric and surface temperatures [e.g., 9]. During dust storms, nighttime surface temperatures are elevated due to enhanced downwelling IR radiation from
the aerosol particles. Conversely, daytime surface temperatures are reduced as a consequence of the absorbed
solar radiation. During GDE, scattering by atmospheric
dust increases the planetary albedo by 10’s of percent
[4]. We hypothesize that the enhanced planetary albedo
dominates the change in overall energy balance during
GDE, reducing the net absorbed heat of the system and
cooling the planet. In this scenario, developed further
below, polar CO2 snowfall increases through reduced
meridional sensible heat transport, along with greater
availability of condensation nuclei for CO2 cloud growth
and precipitation.

Data and Methods
We used data from the Mars Climate Sounder [3] onboard NASA’s Mars Reconnaissance Orbiter, consisting of infrared and solar radiance, retrieved atmospheric
and surface temperatures, and aerosol opacity profiles
of H2 O ice, dust, and CO2 ice [5]. Polar snow clouds
are common in polar winter, and contribute typically
∼ 10 − 20% of seasonal deposits [6, 7]. The amount
and dustiness of snowfall can significantly affect polar

heat budgets, which in turn have consequences for the
planetary climate [8].

Results
Patterns of cloud activity
CO2 cloud opacity profiles retrieved from MCS radiance
data show consistent patterns of snowfall in both polar
regions (Fig. 1). Cloud activity generally begins after
LS = 195◦ in the north, and LS = 15◦ in the south.
At the onset of winter, clouds are highly concentrated
near the surface (< 10 km altitude), with higher altitude
clouds (15 − 30 km) becoming more prominent during
mid-winter. Following the mid-winter solticial pause in
the north (LS = 270◦ ), a second flurry of cloud activity
occurs during LS = 295 − 345◦ at altitudes < 10 km. A
similar second phase does not occur in the south.

Figure 1: (left column) Median CO2 cloud opacity profiles
from MCS for MY 29 – 34 in the north (top) and south (bottom) polar winter regions 80 − 90◦ latitude. Colors indicate
aerosol opacity, and the horizontal axis is LS . Dashed lines are
pressure contours. Solid curves indicate the CO2 frost point,
TCO2 , while dotted lines indicate TCO2 + 2 K. (right column)
Difference of MY 29 cloud opacity from the 5-year median,
showing an enhancement in low-altitude clouds at both poles.

Changes in CO2 clouds following the MY 28 GDE
Following the Mars Year (MY) 28 global dust event, we
observed distinct changes in the amount of CO2 cloud
activity appearing in the MCS retrievals in the winters
of MY 29 (Fig. 1). Enhanced cloud opacity occurred
in both winter hemispheres during the first phase (LS =
195 − 265◦ ) of low-altitude activity. Although a data gap
prevents a complete analysis, the north polar region’s
second phase (LS = 295 − 345◦ ) appears to show signs
of enhanced activity as well. Notably, although the south
polar region typically does not exhibit a second phase of
low-altitude activity, the MY 29 data do show slightly
elevated CO2 cloud opacity around LS = 115 − 165◦ .
The increase in low-altitude cloud activity during
MY 29 is contrasted with a net decrease in clouds in the
middle troposphere. Around the solsticial pause in both
winter hemispheres, the polar cloud activity at altitudes
10 - 30 km is reduced or nearly the same as in other
years. Based on the difference of MY 29 cloud opacity
from the 6-MY median (MY 29 - 34), both poles show an
enhancement of ∼ 50% during their respective winters
following the MY 28 GDE.

Changes in snowfall
Effects of the MY28 GDE can also be seen in the amount
of snowfall occurring in the following winter seasons.
Surface brightness temperatures at 32 µm (T32µ ) retrieved from MCS behave as a proxy for snowfall occurrence, owing to the strong emissivity reduction of granular CO2 ice in the transparency band from ∼ 20 µm
to 40 µm [6, 8, 11, 12]. We used T32µ data for three
different north polar sites at latitudes ∼ 80 − 85◦ to determine the mean deviation of years with and without
dust storms. In MY 29 (following the GDE of MY 28),
we found a net reduction in T32µ of ∼ 2 K at two of
these sites, and no change at the third. Notably, the two
sites exhibiting these changes were at similar latitudes
(∼ 80◦ ), but separated by 90◦ of longitude. The changes
are consistent with increased snowfall at at these sites
during MY 29, relative to MY 30-32.

Effects on the polar heat budget
We measured the outgoing long-wave infrared radiation
(OLR) by integrating the top-of-the-atmosphere radiances from MCS. As shown in Figure 2, the southern
summer of MY 29 showed a measurable drop in OLR
compared to the following Mars year. This reduction
in OLR is ∼ 1013 W, and appears not to be directly
compensated by a rise in OLR during other seasons.
Furthermore, the north pole does not show any significant difference in OLR during MY 29 compared to the
following Mars year.

Figure 2: Outgoing long-wave radiation (OLR) from the top
of the Mars atmosphere, based on average MCS brightness
temperatures from 12 to 32 µm in the polar regions > 80◦
latitude. Note the reduction of OLR during southern summer
of Mars Year 29 (red circles)

The measured reduction in OLR during the south
polar summer could be offset most directly by a change
in the amount of CO2 ice deposited on the surface, either
through snowfall or direct frost accumulation. With a
latent heat L ≈ 6 × 105 J kg−1 and bulk density ρ =
(1 − ϕ)ρ ≈ 500 kg m−3 (with ϕ ≈ 0.7 the porosity)
[13], the heat deficit ∆Q × ∆t ∼ (1013 W) × (107 s)
corresponds to a difference in CO2 ice deposit thickness
∆z ∼ ∆t∆Q/(ρLAcap ) ≈ 10 cm, which is roughly 10%
of the mean seasonal cap thickness [13]. In order to
offset the heat deficit during summer, this excess ∼ 10
cm layer would accumulate during the south polar winter
of MY 29. We do not observe a measurable increase in
OLR during south polar winter of MY 29, and therefore
the heat loss causing an increase in CO2 deposition must
be offset elsewhere. For example, the total heat deficit
of ∼ 1013 W could be caused by a small increase in mean
planetary albedo of ∼ 0.1%.
We examined the possible change in planetary albedo
due to global dust storms using the optical depth data
from [14], and a two-stream radiative transfer model [6,
7]. Column dust opacities exceeding τvis = 1 covering
latitudes ±45◦ persisted during the MY 28 and MY 34
GDEs for ∼ 60 sols. During this period, our model results indicate the planetary albedo would have increased
by ∼ 10% or more (Fig. 4). On average, then, the
global planetary albedo during these years would have
been higher by ∼ 1% relative to the Mars years without
GDE. Further analysis will be carried out to improve
this estimate, but these results indicate the modeled increase in planetary albedo could be sufficient to cause
the observed polar heat deficit.

effect was ∼ 0.9 K of warming, where the nighttime
increase outweighed the daytime decrease. However,
from an energy balance perspective, the net effect is one
of planetary cooling. The increase in planetary albedo
during a GDE reduces the solar flux at the surface by
∼ 30 − 60 W m−2 (roughly 50-75% reduction in absorbed radiation) [10]. Planetary IR emission is also
reduced during GDE, because a comparable reduction
(within ∼ 1 K) of daytime temperatures corresponds to
a much larger change in thermal emission:
↑
= σ(T + ∆T )4 = σT 4 + 4σ∆T T 3 + ...
FIR
↑
∼ ∆T T 3 for ∆T ≪ T
→ ∆FIR

Figure 3: Heat deficit of the south polar region 80-90◦ S based
on (negative) difference in OLR for the MY 29 summer following the MY 28 global dust event.
0.34
0.32

Planetary Albedo

0.30
0.28
0.26
0.24

With Tday = 265 K, Tnight = 210 K, and ∆T = 15 K, the
↑
changes in daytime and nighttime IR fluxes are ∆Fday
≈
↑
−2
−2
−63 W m
and ∆Fnight ≈ +32 W m respectively.
↑
The net change is therefore ∆Fnet
≈ −31 W m−2 , implying planetary cooling. With Mars’ hemispheric surface area of ∼ 7×1013 m2 , the total reduction in absorbed
power could be as large as ∼ 1015 W. Thus, the observed
decrease in MY 29 summer OLR (and corresponding enhancement in CO2 deposition in the preceding winter)
could be explained by the net reduction in absorbed solar
radiation by the planet during the global dust storm. Sensible heat transport into the polar regions could therefore
have been reduced overall during MY 29 as a result of
the preceding year’s dust event.

0.22
0.20

Conclusions

0.18
0

2

4

6

8

10

Solar-band Dust Optical Depth, τvis

• Atmospheric retrievals from MCS show an increase in low-altitude polar cloud opacity of ∼
50% in both winter seasons (i.e., a full Mars Year)

following MY 28 global dust storm.
Figure 4: Modeled planetary albedo using the two-stream
delta-Eddington approximation and dust solar-band singlescattering albedo (ϖ = 0.94) and asymmetry parameter
(g = 0.7) from [15], as a function of atmospheric column
dust optical depth. The solid line indicates the approximate
global Bond albedo of Mars (under dust-free atmospheric conditions) of ∼ 0.25 [16]. Dashed lines indicate the approximate
range given variations in surface albedo at regional (∼1,000
km) scales.

Discussion
Previous analysis of the MY 34 GDE using the LMDMGCM and MCS data assimilation [9] found that increased dust visible optical depth reduced daytime surface radiative fluxes and lowered daytime surface temperatures, whereas nighttime temperatures were increased
due to enhanced atmospheric IR optical depth. The net

• Surface brightness temperatures from indicate en-

hanced snowfall in the northern winter of MY 29.
• Outgoing longwave radiation was reduced by ∼
1013 W during the south polar summer of MY 29

compared to years that did not follow a GDE.
• We show that the observed changes can be under-

stood in terms of changes to the planetary heat
budget driven by a reduction in absorbed solar radiation due to enhanced planetary albedo during
periods of high atmospheric dust opacity.
• Further work, including analysis of MCS data

from other GDEs and model comparisons, will
elucidate the connections between the dust cycle,
CO2 cycle, and planetary heat budget.

References
[1] Heavens, N. G. et al. (2018), Nature Astron. 2(2),
126-132. [2] Chaffin, M. S. et al. (2021), Nature Astron.
5(10), 1036-1042. [3] McCleese, D. J. et al. (2007),
JGR, 112(E5). [4] Mallama, A. (2007), Icarus, 192(2),
404-416. [5] Kleinböhl, A. et al. (2009), JGR, 114(E10).
[6] Hayne, P. O. et al. (2012), JGR, 117 (E8). [7] Hayne,
P. O. et al. (2014), Icarus, 231, 122-130. [8] Gary-Bicas
et al. (2020), JGR, 125(5), e2019JE006150. [9] Streeter,

P. M., et al. (2020), GRL, 47 (9), e2019GL083936. [10]
Read, P. L. et al. (2016), Quart. J. Royal Met. Soc.,
142(695), 703-720. [11] Kieffer, H. H. et al. (2000),
JGR 105(E4), 9653-9699. [12] Hansen, G. B. (1999),
JGR, 104(E7), 16471-16486. [13] Aharonson, O. et al.
(2004), JGR, 109(E5). [14] Montabone, L. et al. (2015),
Icarus, 251, 65-95. [15] Wolff, M. J., et al. (2009), JGR,
114(E2). [16] Christensen, P. R. et al. (2001), JGR,
106(E10), 23823-23871.