MSL FROST DETECTION CAMPAIGNS
G.M. Martínez, Lunar and Planetary Institute/USRA, Houston, USA (gmartinez@lpi.usra.edu), R.V. Gough,
CIRES and University of Colorado, Boulder, USA, W. Rapin, IRAP, CNRS-UPS, Toulouse, France, P.-Y.
Meslin, IRAP, CNRS-UPS, Toulouse, France, O. Gasnault, IRAP, CNRS-UPS, Toulouse, France, S. Schröder,
DLR, Berlin, Germany, T.H. McConnochie, Space Science Institute, USA, H. Savijärvi, Finnish Meteorological Institute, Helsinki, Finland, E. Fischer, University of Michigan, Ann Arbor, USA, S. Guzewich, NASA Goddard Spaceflight Center, Greenbelt, USA, C.E. Newman, Aeolis Research, USA, A. R. Vasavada, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, USA, M. de la Torre-Juárez, Jet Propulsion
Laboratory, California Institute of Technology, Pasadena, USA, R. Wiens, Purdue University, West Lafayette,
USA, and N. Lanza, LANL, Los Alamos, USA.,
Introduction: We report results of the frost detection campaigns performed by the Curiosity rover
during the first 3332 sols of the Mars Science Laboratory (MSL) mission. Also, we describe the observational strategy and target selection of these campaigns to support NASA’s Mars 2020 and future
missions in the search for frost on Mars.
We use environmental measurements from the
Rover Environmental Monitoring Station (REMS) to
predict frost formation, and ChemCam Laser Induced Breakdown Spectroscopy (LIBS) to detect
enhanced hydration levels. Only when both instruments return positive results, we interpret that the
formation of frost is likely.
Out of 14 campaigns (Table 1), only on sol 2548
(Ls ~ 89°) in Martian Year (MY) 35 both REMS and
ChemCam LIBS returned results consistent with
frost formation. However, we note that these results
were not unambiguous, and that future lab studies
will improve our understanding of ChemCam sensitivity to frost.

REMS and ChemCam Instruments: REMS is
the weather station onboard MSL. It includes six
sensors measuring atmospheric pressure, UV fluxes,
wind velocity (which stopped working at around sol
1500), air temperature, ground temperature, and
relative humidity [1–3]. The REMS sampling strategy consists of 5-min-long hourly samples at 1 Hz,
with interspersed full hour sample periods at 1 Hz to
cover every time of the sol over a period of a few
sols.
The LIBS instrument, part of the ChemCam instrument package, is able to perform remote and
sensitive elemental analysis that includes the identification of submicron-thick frost layers [4–5].
Environmental Context: Fig. 1 shows REMS
measurements of ground temperature (T g), relative
humidity (RH), and water vapor volume mixing ratio
(VMR) for the first 3332 sols of MSL mission. In
every MY, the ground temperature shows the coldest
annual values and the RH shows the highest annual
values between late southern fall and early winter.

Table 1. List of frost/hydration detection campaigns by the Curiosity rover during the first 3332 sols of the MSL
mission. Only on sol 2548 in MY 35, both REMS and ChemCam returned results indicating the likely formation
of frost on a soil patch informally named Kilpatrick. ‘Maybe’ indicates positive results without enough certainty.

The nighttime VMR shows annual maximum values
at Ls ~ 160°, when the water vapor column abundance at Gale is highest [5].
The RH reached saturation levels in MY 35, although fog has not been detected by the mission.

VMR×P, where es is the saturation vapor pressure
over ice [6], e is the water vapor pressure at 1.6 m,
and P is the atmospheric pressure. In establishing
this comparison, and due to the lack of humidity
measurements at the ground, we implicitly assume
that the water vapor content is constant in the first
1.6 m. Uncertainties derived from this assumption
will be assessed as part of an ongoing manuscript.
Fig. 2 shows the seasonal evolution of the maximum and minimum Tg (black), along with Tf values
calculated when the RH at the ground (~e/es(Tg)) is
highest (blue), which typically occurs right before
sunset (LMST ~0500–06:00). Most frost events are
predicted to occur seasonally within the Ls 90°–110°
period, when the ground temperature falls below the
frost point.
We note that Tg measurements in the second half
of MY 35 and in MY 36 are subjected to strong
electronic noise at nighttime, and thus the uncertainty in minimum Tg values is larger.

Figure 2. Seasonal evolution of the maximum and
minimum ground temperature measured by REMS
(black), and of the frost point calculated when the
RH at the ground, RHg ~e/es(Tg), is highest (blue).
The ground temperature falls below the frost point
on several sols during fall and winter (mostly in the
Ls 90°–110° period), and on a few sols during summer and spring, suggesting potential frost events.

Figure 1. Interannual and seasonal evolution of the
daily maximum, mean and minimum ground temperature (top), daily maximum RH (middle), and
nighttime VMR (bottom). The daily maximum RH is
generally achieved between 04:00 and 06:00 LMST,
when the temperature is coldest. VMR values are
obtained at the same time as the RH shown above.
To predict frost events with REMS, we compare
Tg with the frost point (Tf) calculated as es(Tf) = e =

Target Selection: Three types of targets have
been considered as part of our Frost Detection Campaigns: (1) soil/sand, (2) rocks, and (3) ventifacted
rock fingers. Each type presents pros and cons for
the potential formation of frost.
Soil/sand has low thermal inertia, and therefore it
can get colder overnight. However, this type of terrain is porous, and thus water vapor might diffuse to
some depth before forming frost. Rocks are not porous but have high thermal inertia, and thus stay
warmer overnight. Finally, ventifacted rock fingers
have relatively low thermal inertia and are not porous, but it is challenging to estimate their temperature (the REMS ground temperature sensor has a
field of view of a few m2, while the size of ventifacts
is of the order of cm).

Because frost at the Viking 2 landing site preferentially formed and persisted on soils as opposed to
rocks (Fig. 3), it was decided to select this type of
terrain in the campaigns of MYs 35 and 36 (Table 1).

ing dawn vs daytime, suggesting there was no detectable change in water content overnight. Results
for Kittybrewster, not shown, are similar to those of
the Kinnordy target (i.e., no H enhancement).

Figure 3. Color image by Viking Lander 2 camera
showing frost preferentially on the soil. Credit:
NASA/JPL.
Observational Strategy of MSL Frost Campaign in MY 35: Three targets were analyzed by
ChemCam LIBS during this campaign: Kilpatrick on
sol 2548 at Ls = 89° (Fig. 4); Kinnordy on sol 2565
at Ls = 97°; and Kittybrewster on sol 2582 at Ls =
104°. All targets met the following selection criteria:
(1) a fine-grained soil patch, (2) larger than 5 cm in
size and (3) more than 3 m from the rover. This was
decided to ensure thermal insulation from higher
thermal inertia rocks, and to avoid thermal contamination from the rover.
For each target, a ChemCam raster was collected
a few minutes before sunrise (~05:50 LMST), when
temperature is at its coldest and RH at its highest,
and another was collected around 13:00, when temperature is at its warmest and RH at its lowest. Each
raster consisted of 10 points spaced along a line (Fig.
4, top), with 10 shots on each point. The hydrogen
emission peak at 656.6 nm is extracted from the
spectra [5,7] and several signal normalizations are
tested to compare day/dawn pairs of observations.
Contemporaneously to ChemCam, REMS measurements of Tg, RH and P were collected to document the rapidly changing environmental conditions
near sunrise. From these measurements, the frost
point at the surface was determined assuming constant mixing ratio in the lower 1.6 m of atmosphere.
Results of MSL Frost Campaign in MY 35:
Fig. 5 shows the ground temperature (red) and frost
point (blue) for Kilpatrick (top) and Kinnordy (bottom) on sols 2548 and 2565, respectively. For
Kittybrewster, not shown, the diurnal evolution of
both quantities was similar to that of Kinnordy. Only
at Kilpatrick the environmental conditions were
favorable for the formation of frost, where the
ground temperature fell well below the frost point
between 05:00 and 06:00 LMST.
Fig. 6 shows ChemCam LIBS normalized H signals
at Kilpatrick (top) and Kinnordy (bottom), both at
daytime (left; ~13:00 LMST) and dawn (right;
~05:40 LMST). Only for Kilpatrick there was enhanced average H signal in shots at dawn relative to
daytime, above point-to-point signal variability. For
Kinnordy, there is no difference in the H signal dur-

Figure 4. (Top) ChemCam Remote Micro-Imager
image of Kilpatrick target on sol 2548. (Middle)
Mastcam image of the surroundings. (Bottom)
Navcam image with REMS’ field of view of ground
temperature on sol 2548.
Conclusions: Out of 14 frost detection campaigns performed along the MSL traverse, only on
sol 2548 (Ls ~ 89°) in MY 35 both REMS and
ChemCam LIBS returned results consistent with the

Temperature (K)

formation of frost. REMS observations suggested
frost formation between 05:00 and 06:00 LMST and
ChemCam LIBS showed enhanced average H signal
in shots at dawn relative to daytime, above point-topoint signal variability. We interpret these results as
likely frost formation. However, we are not able to
determine the phase or distribution of this transient
water, and adsorbed water layers on soil grains are
also a possible cause.
Point-to-point variability and signal normalization are the main limitations of these LIBS measurements. Future lab studies should improve our
understanding of ChemCam sensitivity to frost.
270
260
250
240
230
220
210
200
190
180
170
160

Tg
Tf

Sol 2548
Kilpatrick

Potential Frost
Event
0

2

4

6

8

280

Temperature (K)

Tg
Tf

Sol 2565
Kinnordy

260

10 12 14 16 18 20 22 24
LMST

240
220

200
180
160
140

0

2

4

6

8

10 12 14 16 18 20 22 24
LMST

Figure 5. Diurnal evolution of ground temperature
(red) and frost point temperature (blue) on sols 2548
(top) and 2565 (bottom). Between 05:00 and 06:00
LMST on sol 2548, the ground temperature fell well
below the frost point, indicating favorable conditions
for frost formation. Sunrise was ~05:50 LMST on
both sols.

Figure 6. Comparison of ChemCam LIBS hydrogen
signal collected at targets Kilpatrick (a) and
Kinnordy (b) during the day (left) and dawn (right).
References: [1] Gómez‐ Elvira, Javier, et al. "Curiosity's rover environmental monitoring station:
Overview of the first 100 sols." Journal of Geophysical Research: Planets 119.7 (2014): 1680-1688. [2]
Martínez, G. M., et al. "The modern near-surface
Martian climate: a review of in-situ meteorological
data from Viking to Curiosity." Space Science Reviews 212.1 (2017): 295-338. [3] Newman, Claire E.,
et al. "MarsWRF convective vortex and dust devil
predictions for Gale Crater over 3 Mars years and
comparison
with
MSL‐ REMS
observations." Journal of Geophysical Research: Planets 124.12 (2019): 3442-3468. [4] Meslin, P-Y., et
al. "Soil diversity and hydration as observed by
ChemCam at Gale Crater, Mars." Science 341.6153
(2013): 1238670. [5] Schröder, S., et al. "Hydrogen
detection with ChemCam at Gale crater." Icarus 249
(2015): 43-61. [6] McConnochie, Timothy H., et al.
"Retrieval of water vapor column abundance and
aerosol properties from ChemCam passive sky spectroscopy." Icarus 307 (2018): 294-326. [7] Rapin,
William, et al. "Quantification of water content by
laser induced breakdown spectroscopy on
Mars." Spectrochimica Acta Part B: Atomic Spectroscopy 130 (2017): 82-100.