OBSERVED SEASONAL VARIATIONS OF HYDROGEN ABUNDANCE
IN THE SHALLOW SUBSURFACE OF MARTIAN TROPICS AND
IMPLICATIONS FOR WATER CYCLE.
M. A. Kreslavsky , Earth and Planetary Sciences, University of California – Santa Cruz, Santa Cruz, CA, USA
(mkreslav@ucsc.edu).

Introduction:
Ground ice is known to be abundant in the very
shallow subsurface at high latitude of Mars and
slightly deeper at midlatitudes. At low latitudes some
modest spatially varying amount of hydrogen is present as a component of hydrated salts or as an unstable, currently sublimating deposits.
Significant amount of information about H distribution in the shallow subsurface came from two
neutron detectors onboard Mars Odyssey orbiter,
Mars Odyssey Neutron Spectrometer (MONS)
[Feldman et al., 2002], and High-Energy Neutron
Detector (HEND) [Mitrofanov et al., 2002]. They are
based on different measurement approaches, however, they measure the same physical quantity, local
neutron flux as a function of neutron energy.
With HEND data, significant seasonal variations
of neutron spectra in martian tropics have been reported and interpreted as seasonal changes in ice
abundance in the upper 10-30 cm of the surface
[Kuzmin et al, 2005, 2007, 2015]. The reported effect is unexpectedly and alarmingly strong, however,
potentially explainable. This effect has been neither
confirmed nor disproved with MONS data, partly
because the MONS data reduction procedure would
artificially suppress seasonal variations, if they indeed exist. On the other hand, the reported processing of HEND data ignored the effects of seasonal variations of the atmospheric mass and a number
of possible biases. If HEND-detected seasonal ice
redistribution [Kuzmin et al, 2005, 2007, 2015] is
reliably confirmed, this would have a profound effect on our understanding of the martian surface layer processes and properties; this also would require
significant changes in the present widely used global
climate models.
I tried to independently analyze both MONS and
HEND data to figure out, if the effect reported by
Kuzmin et al. is caused by seasonal changes in ice
abundance in the shallow subsurface, or it is an observational artefact. Here I present my preliminary
results.
Reanalysis of neutron spectrometer data:
The physical mechanism of sensitivity of neutron
data to ice in the shallow subsurface is the following.
Galactic cosmic rays (GCR) split nuclei in the shallow subsurface; this process released a few highenergy neutrons per split; the neutrons elastically (=
without nuclear reactions, with energy conservation)
collide with other nuclei and, collision by collision,
give part of their energy to them until the energy

reduced to typical thermal energy; some neutrons
occasionally leak from the surface to space. Light
nuclei, and especially the lightest ones, H nuclei,
reduce neutron energy more effectively, therefore, if
H is more abundant, leaked high-energy neutrons are
less abundant.
Neutron detector data are difficult to reduce.
Neutron leakage flux depends on GCR flux and
spectrum that vary at time scales form hours to decades. In addition, detector sensitivity and rate of
background detections also vary in a bizarre way.
Fig. 1 shows the 10 martian year (MY) long history
of high-energy neutron fluxes registered by MONS
(counts of “category 2 events” in the instrument terminology) averaged over two latitudinal zones (10°N
– 30°N, blue, and 10°S – 30°S, red) and over several
days long time periods. I carefully removed all data
associated with various kinds of irregularities. It is
seen that the systematic difference between the zones
(lower flux = higher H abundance in the N zone) is
tiny in comparison to long-term variations of the
detected flux. MONS and HEND teams used data
reduction approaches, although different, both based
on assumption that there are no variations in true
neutron “albedo” in some specific areas. I try to reprocess the data without such assumptions to obtain
the most objective information about seasonal flux
variations. This work is in progress, and technical
details will be reported elsewhere.
My preliminary results suggest that seasonal variations in neutron “albedo” are a real effect. Fig. 2
(red crosses) presents the ratio of raw fluxes (N zone
/ S zone) from Fig. 1 for 10 martian years. The scattering is comparable to what would be expected from
the pure Poisson noise of individual neutron counts.
Individual consecutive measurements in these zones
are separated by minutes; within these short intervals, variability of the GCR flux is very minor, while
rare sudden jumps in detector gain were filtered out.
Systematic seasonal variations of the ratio are clearly
seen. Blue line in Fig. 2 shows the least-square fit
with a periodic function smoothed down to 4 Fourier
harmonics (with periods from 1 MY to ¼ MY). The
first two harmonics are well above the noise level (~
5 “sigma”). To obtain this fit I used the LombScargle algorithm from [Scargle, 1989] with some
errors fixed. My analysis of possible observational
biases preliminarily shows that these periodic variations objectively characterize the ratio of neutron
“albedo”.

Despite inherently wide field of view of the neutron detectors, seasonal frost does not affect the
measurements in Fig. 2, because all seasonal frost is
always beyond the Odyssey’s horizon in the considered latitude zones. Seasonal variations of atmospheric pressure are known to affect the high-energy
neutron flux. On average, the N zone has a lower
elevation and hence a thicker atmosphere than the S
zone, therefore, variations in atmospheric pressure
can potentially have a minor second-order effect on
the neutron flux ratio. In Fig. 3 the single one-year
period of the fit curve from Fig. 2 is compared
against the seasonal atmospheric pressure curve (orange dots), which is arbitrarily stretched and shifted,
and also inverted (upside down) to show the expected signature of the second-order effect on the
N/S ratio. It is seen that the pressure effect likely
contributes to the observed seasonal variations, but
cannot account for the whole signal. (In principle,
the pressure effect can be quantitatively obtained
from the observations, but this work is still in progress at the abstract submission date). The residual
variations of the flux ratio should be attributed to
variations in neutron “albedo” of the surface, which
suggests variations of H abundance in the upper decimeter(s) of the surface. The observed varied H is
almost certainly is in the form of water molecules,
therefore the observed variations imply seasonal
cyclic transport of H2O to and from the uppermost
decimeters of the surface.
HEND data have a better signal-to-noise ratio
and more data irregularities; they yield a generally
similar result, which again suggests that the observed
trend is objective.
The absolute amount of H2O abundance change
cannot be deduced from the data alone without involvement of some modeling of neutron interaction
with the surface nuclei. Order of magnitude estimations by Kuzmin et al. [2007, 2015] are equivalent to
a few mm equivalent layer of ice coming to and going from a 10 – 30 cm thick surface layer. Application of the same estimation method to MONS data
(Fig. 2) gives a part of a mm ice equivalent.
Implications for martian water cycle:
Exchangeable H2O in the upper decimeters of the
surface can be in different physical forms: as ice,
brine, molecules adsorbed at the soil particle surfaces, crystalline water. Kuzmin et al. [2007] argued
that the observed H variations are due to seasonal
accumulation of ice due to vapor exchange with the
atmosphere. Straightforward calculations [Kreslavsky, 2022] with LMD Mars Climate Database as a
proxy for martian temperature and humidity conditions showed seasonal ice condensation with such a
mechanism only above ~30° latitude and seasonal
brine-forming deliquescence only above ~40° latitude. Those calculations did not account for slopes; it
is likely that on slopes of proper orientation ice formation could potentially occur at lower latitudes.
These considerations, however, have two significant

caveats. First, the calculations [Kreslavsky, 2022]
predict formation of seasonal ice in N in late summer, therefore the iciest season is early fall, significantly earlier than the observed neutron minimum.
Similar predicted seasonality is likely for all other
H2O forms (adsorbed and crystalline water), if they
are formed due to vapor exchange with the atmosphere. Second, the observed amount of variable H2O
is alarmingly high. For a martian regolith analog and
martian conditions, Hudson et al. [2007] obtained
diffusion coefficient of H2O vapor about 4 cm2s-1.
With this diffusion coefficient, the characteristic
diffusion time scale through the tens of cm thick
regolith layer is several hours. This means that diffusion is not a significant barrier for the seasonal exchange of H2O, and the amount of exchange is limited by ability of the atmosphere to transport H2O
vapor from polar summer sources to low latitude
boundary layer. If exchange of a mm of H2O equivalent between the surface and the atmosphere indeed
occurs, this means that the water cycle in all martial
climate models is modeled incorrectly, and significant changes are needed in the models.
Another possible explanation of the observed
variability of H abundance is H2O vapor exchange
between shallow and deeper subsurface. The high
effective diffusivity of H2O vapor [Hudson et al.
2007] makes likely movement of large amounts of
H2O. The predicted seasonal phasing of H2O
transport up and down through the seasonal thermal
skin is roughly consistent with the observed phase of
the annual harmonic in Fig. 2 and 3. The caveat is
that shallow ground ice is known not to be stable at
those latitudes [e.g., Schorghofer & Aharonson,
2005], therefore the variable shallow ice could only
exist if there is a transient ice deeper in the subsurface (meters), which is in the process of ongoing
removal. In this case, the presence of variable ice in
the upper decimeters again requires a significant flux
of H2O vapor from the shallow subsurface to the
atmosphere, which in turn means the need of proper
changes in the water cycle modeling.
Conclusions and prospects:
Reanalysis of neutron spectrometer data confirmed seasonal variations of H2O abundance in the
shallow (decimeters) subsurface of Mars. The seasonal changes are likely caused by seasonal vertical
redistribution of transient, unstable ice in the subsurface in response to seasonally changing temperature
gradients. This process is inevitably associated with
seasonal release of H2O vapor in the atmosphere.
Future work includes accomplishment of the independent neutron data processing and derivation of
seasonal atmospheric corrections from the data
alone. Theoretical modeling of ice migration in the
regolith is needed for better understanding of processes operating in the shallow subsurface of Mars
and quantification of atmospheric water sources.
Inclusion of regolith vapor sources and sinks is essential in future Mars climate models.

References:
Feldman, W. C., et al. (2002) Global Distribution of
Neutrons from Mars: Results from Mars Odyssey,
Science, 297, 75-78.
Hudson, T. L., et al. (2007) Water vapor diffusion in
Mars subsurface environments, JGR, 112, E05016.
Kreslavsky, M.A. (2022) Seasonal Brine Formation
in Shallow Subsurface on Mars, LPSC 53, #2628.
Kuzmin, R. O., et al. (2005) Seasonal Redistribution
of Water in the Surficial Martian Regolith: Results
of the HEND Data Analysis, LPSC 36, #1634.
Kuzmin, R. O., et al. (2007) Seasonal redistribution
of water in the surficial Martian regolith: Results
from the Mars Odyssey high-energy neutron detector (HEND), Solar System Res. 41, 89-102.
Kuzmin, R. O., et al. (2015) Analysis of the Seasonal

Fig.1

Fig.2

Fig.3

Variations of the Water Equivalent of the Hydrogen Amount in the Subsurface Regolith on Mars
Based on the HEND Data Accumulated During
Five the Martian Years, LPSC 46, #2007.
Mitrofanov, I. et al. (2002) Maps of Subsurface Hydrogen from the High Energy Neutron Detector,
Mars Odyssey, Science, 297, 78-81.
Scargle, J. D. (1989) Studies in Astronomical Time
Series Analysis. III. Fourier Transforms, Autocorrelation Functions, and Cross-Correlation Functions of Unevenly Spaced Data. Astrophys. J.
343:874–887
Schorghofer, N. & Aharonson, O. (2005). Stability
and exchange of subsurface ice on Mars. JGR 110.
doi:10.1029/2004je002350