DISTINGUISHING CAVES FROM NON-CAVE ANOMALIES USING THERMAL INFRARED:
LESSONS FOR THE MOON AND MARS. J. J. Wynne1,2, T. N. Titus3, M. D. Jhabvala4, G. E. Cushing3, N. A. Cabrol1,5,
and E. A. Grin1,5. 1SETI Carl Sagan Center, 515 N. Whisman Road, Mountain View, CA 94043 (firstname.lastname@example.org);
2Merriam-Powell Center for Environmental Research, Department of Biological Sciences, Northern Arizona University, Box
6077, Flagstaff, AZ 86011; 3U.S. Geological Survey, Astrogeology Branch, 2255 North Gemini Dr., Flagstaff, AZ 86001;
4NASA Goddard Space Flight Center, Instrument Systems and Technology Division, Code 550, Greenbelt, MD 20771; 5NASA
Ames Research Center, Space Science Division, MS 245-3, Moffett Field, CA 94035.
Overview: Caves on Earth are often microclimates
which contain evidence of extant life. On other solar
system bodies, such as Mars, these features may be
excellent places to search for extinct/ extant lifeforms.
For the Moon and Mars, caves may also provide pro-
tection or habitats for future human exploration.
The Earth-Mars Cave Detection Project has entered
its fourth year; Phase 2 of this effort has launched its
first year. Our research to date has demonstrated the
viability of the thermal detection of caves on Earth [1-
3] and Mars [4,5], as well as provided theoretical justi-
fication for lunar cave detection . On Earth, caves
are detectable when differences in thermal radiance
between cave entrance and surface are greatest [1-3].
On Mars, features associated with speleogenesis [e.g.
6,10,11] and actual cave-like features [e.g. 4,5] have
been confirmed. While no cave entrances have been
observed on the Moon, sinuous rills (features largely
accepted as lava channels), collapsed lava tubes and
collapse pit features [6-9], have been identified.
While these efforts have improved our understand-
ing of cave thermal behavior [e.g. 1-5], we have yet to
address how cave thermal behavior may influence cave
detection using thermal infrared (TIR) imaging. One of
the most critical concerns facing remote sensing of
caves is how to differentiate a cave from a false posi-
tive (non-cave anomaly) . Before NASA will target
caves on the Moon or Mars as potential sites for ro-
botic exploration, a high level of certainty must be
obtained that the feature of interest is indeed a cave.
Importance of Martian Caves: (A) Caves may be
important in detecting evidence of extraterrestrial life
because they offer protection from inhospitable surface
conditions [1-5]. (B) A manned mission to Mars will
require access to significant H2O deposits for drinking
water, oxygen and hydrogen fuel. If water deposits
exist, caves may provide the best access to these re-
sources . (C) Future human exploration and possi-
ble establishment of a permanent settlement on Mars
will require construction of living areas sheltered from
harsh surface conditions. Caves with a protective rock
ceiling would provide an ideal environment where
these shelters may be built .
Importance of Lunar Caves: These features may be
valuable as potential shelters for human habitation [14-
15] because this buffered environment could protect
astronauts from the inhospitable lunar surface.
Objective: This work aims to improve our under-
standing of thermal signature strength as it relates to
differentiating caves from non-cave anomalies in the
Fig. 1: Pisgah lava beds, Mojave Desert, CA. [A] Color visi-
ble image containing cave entrance (red circle) and anomaly
(blue circle). [B] IR image acquired at 0510 hr overlaid on
the visible image. Cave entrance appears as a warmer fea-
Results: Using the QWIP (Quantum Well
Photodetector) camera, we collected thermal imagery
of the Pisgah lava beds, Mojave Desert, CA (Fig. 1).
Imagery was collected every 10 m over ~24 hrs, 07-08
April 2008 (122 images captured). Within the field of
view are two features, a cave and tunnel (i.e., anom-
aly). Using all 122 images, we ran a Principal Compo-
2451.pdf40th Lunar and Planetary Science Conference (2009)
nents Analysis (PCA) to investigate for potential dif-
ferences between our cave and non-cave anomalies.
For the dataset presented, our results suggest PCA
was a useful tool in discerning a cave from an anom-
aly. Scatter plot (Fig. 2) shows a clear separation be-
tween these two features. Plotting Eigenfunction
weight against time of day (Fig. 3), we observe the
cave as most discernable from the anomaly between
~0500-1000hr and ~1300-1400hr.
Discussion: While the results presented are en-
couraging, our findings represent only one example.
Thermal signal strength of the entrance, and thus de-
tectability, is likely driven by volume, horizontal
length, depth from surface, percentage of rock ob-
structing entrance, slope, aspect, topographic rough-
ness, and geologic substrate [1,3]. We suggest these
factors will also influence signal strength of non-cave
anomalies. We had the luxury of a large dataset con-
taining data points over a diel window (122 images).
Imagery captured via a satellite platform for a lunar or
martian mission would be limited and may represent
only a couple of data points. Additional work is re-
quired to better understand and discern thermal signa-
tures associated with cave entrances and anomalies.
Because detectability of caves on the Earth, Moon
and Mars is likely driven by conduction, locating
caves on these planetary bodies is possible . Unlike
on Earth, groundtruthing potential cave targets on the
Moon and Mars is not possible without considerable
expense. As Phase 2 in the Atacama and Mojave De-
serts continue, we will improve both our cave detec-
tion capabilities and our ability to distinguish caves
from non-cave anomalies. The analytical tools devel-
oped and lessons learned from terrestrial applications
will ultimately be used for interpreting and evaluating
exploration targets on the Moon and Mars.
Acknowledgements: Special thanks to M. Allner
for data collection assistance, and J. Blue, R. Fergason,
R. Hayward and J. Richie for providing comments on
previous versions of this abstract. This study was sup-
ported by NASA Exobiology grant EXOB07-0040,
NASA Spaceward Bound!, USGS-Southwest Biologi-
cal Science Center, and KAOEF.
References:  Wynne, J.J. et al. (2008) Earth Planet. Sci. Lett.
272: 240-250;  Wynne, J.J. et al. (2008) LPSC 39th, #2459; 
Wynne, J.J. et al. (2007) LPSC 38th, #2378;  Cushing, G.E. et al.
(2007) GRL 34, L17201;  Cabrol et al. (2009) LPSC 40th, #1040;
 Halliday, W.R. (2007) J. Caves Karst Stud. 69: 103–113; 
Guest, J.E. (1972) Stud. Speleol. 2: 161–175;  Greeley, R. (1977)
Atti Del Seminaro Sulle Grotte Laviche: 181–192;  Greeley, R.
(1983) IV Symposium Internazionale di Vulcanospeleogia, Catania,
Sicily: 15;  Ferrill, D.A., et al. (2003) LPSC 34th, # 2050; 
Wyrick, D. et al. (2004) JGR 109, E06005;  Baker, V.R. et al.
(1993) Ed. J.S. Lewis, Resources of Near-Earth Space (UofA Press,
Tucson), p. 765-798;  Boston, P.J. et al. (2003) Grav. Space
Biol. Bull. 16: 121-131;  Horz, F. (1985) Lunar bases and space
activities of the 21st century (A86-30113 13–14), LPI, pp. 405–412;
 Billings, T., Godshalk, E. (1998) Workshop on New Views of
the Moon, LPI, # 6049.
Fig. 2: [A] Scatter plot of the 2nd and 3rd principle compo-
nents. Output can be used to differentiate between the cave
(red), non-cave anomaly (blue), and high thermal inertia
basalt (green). [B] Visible image with 3rd principle compo-
nent output overlaid; colors match those used in A.
Fig. 3: Cave entrance has a strong 3rd Eigenfunction, sug-
gesting warm temperatures at dawn and cool temperatures in
afternoon. The 2nd Eigenfunction is phase-shifted, allowing
the differentiation between cave and anomaly.
2451.pdf40th Lunar and Planetary Science Conference (2009)