Constraining the warming effect of high-altitude water ice clouds on early Mars in a 3D
moist general circulation model with a simplified cloud microphysical scheme
F. Ding1,*, R. Wordsworth1,2, K. Loftus2, K. Steakley3, M. Kahre3, R. Haberle3. 1School of Engineering and Applied
Sciences, Harvard University. 2Department of Earth and Planetary Sciences, Harvard University. 3NASA Ames
Research Center. *fding.dfdfdf@gmail.com.
Introduction: Geologic evidence suggests >102-yrlong lake-forming climates persisted on Mars 3-4 Ga
[e.g., 1-3]. These early warm climates cannot be explained by the greenhouse effect of CO2 and water
vapor alone [4]. Recently, a warming mechanism for
early Mars based on high-altitude water ice clouds was
proposed, for situations where the surface water inventory is limited and far away from tropical regions [5].
However, microphysical representations of clouds and
precipitation remain one of the main uncertain factors
for climate modeling of terrestrial planets, including
present-day Earth [6]. Here we use a three-dimensional
moist general circulation model (3D moist GCM) with
a simplified cloud microphysical scheme to constrain
the potential warming effect of high-altitude water ice
clouds in a physical parameter space.

ble water, and surface water distributed within 10°
around the north pole.

3D Moist GCM and Model Setup: We have developed a 3D moist general circulation model to simulate
diverse planetary atmospheres based on the "vertically
Lagrangian" finite-volume dynamical core [7] of the
Geophysical Fluid Dynamics Laboratory Flexible
Modeling System (FMS), which solves the atmospheric primitive equations in spherical coordinates. The
dynamical core uses a cubed-sphere gridding technique
for atmospheric dynamics [8,9] which improves both
the computational performance and accuracy compared
to conventional latitudinal-longitudinal gridding. We
have previously used this GCM to simulate and elucidate multiple moist climate states on arid rocky
exoplanets [10-12]. Our GCM uses a line-by-line approach to describe the radiative transfer, which maintains both flexibility and accuracy for simulating diverse planetary atmospheres. The physical schemes for
moist processes are very similar to those used in [13],
including a moist convection scheme, a large-scale
condensation scheme, and a planetary boundary
scheme. The cloud microphysical scheme in our GCM
is intentionally simple: the radius of cloud particles is
fixed (default value of 10 m), and in addition to dynamical transport, the tendency of cloud mass is only
subject to water vapor condensation (the cloud source)
and precipitation (the cloud sink). The cloud sink term
is characterized by a constant timescale that can be
interpreted as the lifetime of cloud particles. Other
baseline parameters of our simulations are: 75% of
present-day solar luminosity, zero eccentricity, obliquity of 25°, an atmosphere with 1 bar CO2 and condensa-

Figure 1: Global annual mean and surface temperature
as a function of the lifetime of cloud particles. The
climate states diagnosed in Figure 2 are marked by
green arrows.
Simulated Climate as a Function of Cloud Lifetime:
Two distinct climate regimes emerge as the lifetime of
cloud particles is varied in the GCM simulations. Figure 1 shows the global annual mean surface temperature as a function of cloud lifetime. When this timescale is shorter than 10 days, the climate remains cold
and insensitive to cloud lifetime, because cloud particles are removed rapidly from the atmosphere. Diagnostics of the cloud mass budget when the cloud lifetime is 2.5 hours (Figure 2) show it is dominated by
local condensation and precipitation. The dynamical
transport of clouds from the northern polar region to
other regions is so weak that the ice clouds are still
confined near the north pole. So in this climate regime,
the climate is nearly as cold as in the cloud-free run.
When the cloud lifetime is longer than 10 days (roughly the dynamical timescale to transport cloud particles
from the north pole to the equator), ice clouds start to
accumulate in the tropical upper atmosphere. The climate is warmed by the greenhouse effect of highaltitude ice clouds and is very sensitive to the cloud
lifetime. Figure 2 confirms that the dynamical
transport of clouds is comparable to local condensation
rate when the cloud lifetime is 250 hours. When this
timescale is ~100 days, the global annual mean surface
temperature is warmed to the freezing point by the

greenhouse effect of the water ice clouds in the upper
atmosphere.

Figure 2: Water cloud mass concentration (upper
panels) and zonal-mean vertically-integrated cloud
mass budget (lower panels) when the cloud lifetime is
2.5 hours (left) and 250 hours (right).
Conclusions and Future Directions: Our GCM simulations with a simplified cloud microphysical scheme
suggest that warming early Mars with high-level ice
clouds requires a cloud lifetime that is longer than the
inter-hemispheric dynamical transport timescale. To
make the planet warm enough for low-latitude lakes
for centuries or longer, the cloud lifetime should be
longer than 50 days. However, the falling timescale of
10 m particles, as estimated by Stokes' law and the
gravitational settling velocity, is ~30 days. In addition,
microphysical processes such as autoconversion of
cloud ice to form snow and accretion of cloud ice by
snow may occur even faster than gravitational sedimentation, limiting the warming effect of high-level
ice clouds further.
Currently, we are performing a transformed Eulerianmean analysis to calculate meridional transport timescales directly, implementing cloud evaporation in the
model for a CO2 atmosphere following the thermodynamic-heat conduction and vapor diffusion calculation
in [14] with some simplifications, and testing the effects of varying the surface water distribution. In addition, we plan to perform selected intercomparison
simulations between the FMS, LMDZ and NASA
Ames early Mars climate models.

References: [1] Grotzinger, J. P. et al. (2014). Science,
343(6169), 1242777. [2] Haberle, R. M. et al. (2017),
Cambridge University Press. [3] Kite, E. S. (2019).
Space Science Reviews, 215(1), 1-47. [4] Wordsworth,
R. et al. (2013). Icarus, 222(1), 1-19. [5] Kite, E. S. et
al. (2021). PNAS, 118(18). [6] Zhao, M. et al. (2016).
Journal of Climate, 29(2), 543-560. [7] Lin, S. J.
(2004). Monthly Weather Review, 132(10), 2293-2307.
[8] Ronchi, C., Iacono, R., & Paolucci, P. S. (1996).
Journal of computational physics, 124(1), 93-114. [9]
Putman, W. M., & Lin, S. J. (2007). Journal of Computational Physics, 227(1), 55-78. [10] Ding, F., &
Wordsworth, R. D. (2020). ApJL, 891(1), L18. [11]
Ding, F., & Wordsworth, R. D. (2021). PSJ, 2(5), 201.
[12] Ding, F., & Wordsworth, R. D. (2022). ApJL,
925(1), L8. [13] Merlis, T. M., & Schneider, T. (2010).
JAMES, 2(4). [14] Loftus, K., & Wordsworth, R. D.
(2021). JGR: Planets, 126(4), e2020JE006653.