AN IMPROVED MODEL OF WATER ICE SUBLIMATION ON MARS:
VALIDATION AT THE PHOENIX LANDING SITE
A. R. Khuller, School of Earth and Space Exploration, Arizona State University, Tempe, Arizona, USA
(akhuller@asu.edu), G. D. Clow, Institute of Arctic and Alpine Research, University of Colorado, Boulder, Colorado, USA.

Introduction:
While numerous authors [e.g., 1, 2] have modeled
the sublimation rate of exposed H2O ice at the surface
of Mars, these models have not been directly validated
with in-situ data. Furthermore, although analog studies of martian ice have been performed [3, 4], these
studies fail to accurately replicate martian ice conditions in terms of dust content, physical scale and
boundary layer atmospheric conditions.
Here we present results from a one-dimensional
atmospheric surface layer model that calculates turbulent fluxes over planetary surfaces to predict the sublimation rates of ices (CO2, H2O, CH4 etc.). We have
validated our model against terrestrial and martian insitu data. We have compared our results with the predictions of [1], which are commonly used by martian
ice studies today.
Methods: We have developed a model of the atmospheric surface layer that is based on previous terrestrial/martian atmospheric surface layer work [5, 6].
The model requires the following input data: (1) surface pressure, (2) surface roughness, (3) temperature
(at two heights), (4) relative humidity (at two heights)
and (5) wind speed. To validate the model, we used
the only martian in-situ measurements of martian H2O
ice available at present, from the Phoenix mission,
which landed in Utopia Planitia (68.2°N, 125.8°W) in
2008. The lander’s robotic arm dug trenches around
the lander to excavate buried H2O ice present a few
centimeters below the surface [7]. This trenching revealed light-toned, friable H2O ice on sol 20 of the
mission (Fig. 1), with an albedo matched by 350 𝜇m
H2O snow with 0.015% dust in it [8].
This ice was left undisturbed for the remainder of
the mission. Within the first four days (between sols
20 and 24), a few 1.5 - 2 cm sized, light-toned clods
sublimated away (yellow ellipse in Figure 1). The
light-toned patches sublimated far more slowly, with
shadow measurements indicating losses of ~3 mm
over 42 sols [7]. However, no measurements of temperature were made of the ice itself, and the lower atmospheric temperature measurements were impacted
by the lander itself (but not at the 2 m height; [9]).
Thus, to validate our model, we simulated temperatures at the surface of the ice by using the Mars Climate Database (MCD; [10, 11]).
To do so, we ran the MCD for the Phoenix mission
landing site (with properties similar to [12]) to obtain
2 m air temperatures (black curves in Figure 2) and

also soil surface temperatures (blue curves). We then
added the mean temperature difference between the
MCD surface and MCD 2 m temperatures to the 2 m
air temperatures measured by the lander (red curves)
to obtain simulated soil surface temperatures (green
curves).

Figure 1. Evolution of ice exposed at the Phoenix
landing site over 42 sols (adapted from [7]).
We assumed that these simulated soil surface temperatures are representative of the ice clods’ surface/skin temperatures, despite the differences in

albedo and thermal inertia, due to their small size and
the radiating nearby trench walls resulting in enhanced warming. Relative humidity at the surface of
the ice was assumed to be equal to 100% and interpolated values of relative humidity measured by the
lander were used at ~2 m. The wind speed was assumed to be 4 m/s throughout [13]. The sensitivity of
the modeled net sublimation was also analyzed.

conditions, leading to errors in the resulting fluxes of
sensible and latent heat of up to a factor of three depending on the stability of the atmosphere [5]. Additionally, thus far, these equations have not been validated/tested using in-situ data on Mars.
Figure 3 shows that these equations consistently
underestimate (black dashed lines) the net sublimation of the 1.5 – 2 cm ice clods observed at the Phoenix landing site. No set of conditions we tested in our
sensitivity analyses was able to match the observed
sublimation.
Thus, our improved model of ice sublimation represents a step forward in martian ice modeling studies,
building on past work. We are in the process of applying this model to other exposures of ice on Mars, such
as within the mid-latitude mantle (e.g., [15]).

Figure 2. Temperatures between sols 20 and 62 at 0
m (simulated; green, and modeled by MCD; blue) and
2 m (measured; red, and modeled by MCD; black).
Results and Discussion:
Our Model Results
The modeling results for the base case (~0.7 cm;
black triangles in Figure 3) are within a factor of two
of the observed sublimation of the 1.5 - 2 cm clods
over four sols (dashed red lines). Our sensitivity analysis indicates that higher values of surface roughness
and surface temperatures (due to the ~13 cm deep
trench walls near the clods) would increase the net
sublimation during this period, whereas lower wind
speeds caused by the trench walls obstructing the local airflow would reduce sublimation.
To model the light-toned patches (shallow portion
of the trenches, Fig. 1), we ran KRC, a planetary thermal model [14] using the ice’s measured albedo and
calculated thermophysical properties. Due to their
small areal extent, we assumed that the atmospheric
properties above the interfacial sublayer immediately
adjacent to the ground are the same above the patches
as for the surrounding terrain. The resultant net modeled sublimation was significantly lower than for the
clods, which is consistent with the observed ~3 mm
sublimation over 42 sols. The lower rate of sublimation for the patches relative to the clods is likely due
to colder surface temperatures, lower surface roughness and less exposed surface area.
Comparison with Commonly Used Martian Ice
Sublimation Model Predictions [1]
Almost all existing models that simulate the stability and sublimation rate of exposed ice on Mars use
versions of equations similar to those presented by
[1]. These equations ignore the important effects of
atmospheric stability and assume neutral atmospheric
stability under sheared (“forced” convection

Figure 3. Sensitivity of the net sublimation (blue
curves) over the four sols during which the 1.5-2 cm
clods disappeared (dashed red lines) to: a) surface
roughness, b) the simulated surface temperature relative to the base value and c) the wind speed at 2 m.
The predictions of [1] for the same inputs are shown
as black dashed lines for comparison.

References: [1] Dundas, C. M. and Byrne, S.
(2010). Icarus, 206(2), 716-728. [2] Ivanov, A. B.
and Muhleman, D. O. (2000). Icarus, 144(2), 436448. [3] Douglas, T. A. and Mellon, M. T. (2019). Nature comm., 10(1), 1-9. [4] Chittenden, J. D. et al.
(2008). Icarus, 196(2), 477-487. [5] Brutsaert W.
(1982) Springer 1567-7419 [6] Clow G. & Haberle R.
(1990) Mars, 14-91, 111. [7] Smith, P. H. et al.
(2009). Science, 325(5936), 58-61. [8] Khuller, A. R.
et al. (2021). JGR, 126(9), e2021JE006910. [9] Davy,
R. et al. (2010). JGR, 115(E3). [10] Forget, F. et al.
(1999). JGR, 104(E10), 24155-24175. [11] Millour,
E. et al. (2018). MEEMA, 68. [12] Savijärvi, H. and
Määttänen, A. (2010). QJRMS, 136(651), 1497-1505.
[13] Holstein‐Rathlou, C. et al. (2010). JGR, 115(E5).
[14] Kieffer, H. H. (2013). JGR, 118(3), 451-470.
[15] Khuller, A. R and Christensen, P. R.
(2021). JGR, 126(2), e2020JE006539.
Acknowledgements: We are immensely grateful
to Sarah MacDonald, Steve Warren, Phil Christensen,
Tyler Richey-Yowell, Jeff Plaut, Serina Diniega, Jay
Serla, Alex Huff and Sarah Braunisch for very helpful
feedback and advice.