Icarus 204 (2009) 443–457

Contents lists available at ScienceDirect

Icarus
journal homepage: www.elsevier.com/locate/icarus

Subsurface structure of Planum Boreum from Mars Reconnaissance Orbiter

Shallow Radar soundings
Nathaniel E. Putzig a,*, Roger J. Phillips a,b, Bruce A. Campbell c, John W. Holt d, Jeffrey J. Plaut e,
Lynn M. Carter c, Anthony F. Egan a, Fabrizio Bernardini f, Ali Safaeinili e,1, Roberto Seu g
a
Southwest Research Institute, Boulder, CO 80302, USA
b
Washington University, St. Louis, MO 63130, USA
c
Center for Earth and Planetary Studies, Smithsonian Institution, Washington, DC 20013, USA
d
Jackson School of Geosciences, University of Texas, Austin, TX 78712, USA
e
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA
f
A&C2000 s.r.l., Viale Cristoforo Colombo 436, 00145 Rome, Italy
g
Dipartimento InfoCom, Università di Roma ‘‘La Sapienza,” I-00184 Rome, Italy

a r t i c l e i n f o a b s t r a c t

Article history: We map the subsurface structure of Planum Boreum using sounding data from the Shallow Radar (SHA-
Received 27 April 2009 RAD) instrument onboard the Mars Reconnaissance Orbiter. Radar coverage throughout the 1,000,000-
Revised 22 July 2009 km2 area reveals widespread reflections from basal and internal interfaces of the north polar layered
Accepted 22 July 2009
deposits (NPLD). A dome-shaped zone of diffuse reflectivity up to 12 ls (1-km thick) underlies two-
Available online 30 July 2009
thirds of the NPLD, predominantly in the main lobe but also extending into the Gemina Lingula lobe
This article is dedicated to the memory of across Chasma Boreale. We equate this zone with a basal unit identified in image data as Amazonian
Ali Safaeinili, 1964–2009. Dr. Safaeinili was sand-rich layered deposits [Byrne, S., Murray, B.C., 2002. J. Geophys. Res. 107, 5044, 12 pp.
a key contributor to the design and doi:10.1029/2001JE001615; Fishbaugh, K.E., Head, J.W., 2005. Icarus 174, 444–474; Tanaka, K.L., Rodri-
operation of planetary radars and served as guez, J.A.P., Skinner, J.A., Bourke, M.C., Fortezzo, C.M., Herkenhoff, K.E., Kolb, E.J., Okubo, C.H., 2008. Icarus
the SHARAD Instrument Scientist.
196, 318–358]. Elsewhere, the NPLD base is remarkably flat-lying and co-planar with the exposed surface
of the surrounding Vastitas Borealis materials. Within the NPLD, we delineate and map four units based
Keywords:
on the radar-layer packets of Phillips et al. [Phillips, R.J., and 26 colleagues, 2008. Science 320, 1182–
Mars
1185] that extend throughout the deposits and a fifth unit confined to eastern Gemina Lingula. We esti-
Radar observations
Mars, Polar caps mate the volume of each internal unit and of the entire NPLD stack (821,000 km3), exclusive of the basal
Mars, Polar geology unit. Correlation of these units to models of insolation cycles and polar deposition [Laskar, J., Levrard, B.,
Mustard, J.F., 2002. Nature 419, 375–377; Levrard, B., Forget, F., Montmessin, F., Laskar, J., 2007. J. Geo-
phys. Res. 112, E06012, 18 pp. doi:10.1029/2006JE002772] is consistent with the 4.2-Ma age of the oldest
preserved NPLD obtained by Levrard et al. [Levrard, B., Forget, F., Montmessin, F., Laskar, J., 2007. J. Geo-
phys. Res. 112, E06012, 18 pp. doi:10.1029/2006JE002772]. We suggest a dominant layering mechanism
of dust–content variation during accumulation rather than one of lag production during periods of
sublimation.
Ó 2009 Elsevier Inc. All rights reserved.

1. Introduction their correspondence with layered units observed in surface

images are key to assessing the geologic history of the layers and
Sounding-radar instruments in orbit around Mars (Picardi et al., to evaluating their implications for the past and present climate
2005; Seu et al., 2007b) are yielding a wealth of subsurface data, of Mars (e.g., Fishbaugh and Head, 2005; Phillips et al., 2008; Mil-
especially from the polar regions, where icy layered deposits ex- kovich et al., 2009), much as they are for Earth (e.g., Morse et al.,
tend to depths of several kilometers (Plaut et al., 2007; Seu et al., 1998; Welch and Jacobel, 2003).
2007a; Phillips et al., 2008). Here, we describe radar mapping of The most prominent features in the polar regions of Mars are
the subsurface structure of ice-rich layers in Planum Boreum, at stacks of finely layered materials up to 4-km thick that are nearly
the north pole of Mars. The geometry of the radar reflectors and centered on each pole and are cut by large, arcuate chasmata and
smaller reentrant and interior troughs. These materials are com-
posed primarily of water ice with a variable amount of lithic inclu-
* Corresponding author. Address: Southwest Research Institute, 1050 Walnut
sions (Picardi et al., 2005; Plaut et al., 2007), whose contrast with
Street, Suite 300, Boulder, CO 80302, USA. Fax: +1 303 546 9687.
E-mail address: [email protected] (N.E. Putzig). ice reveals the layered nature of the deposits in optical images,
1
Deceased. both on the periphery and within the chasmata where surface

0019-1035/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.icarus.2009.07.034
444 N.E. Putzig et al. / Icarus 204 (2009) 443–457

exposures occur (Clifford et al., 2000, and references therein. See, In the following section, we describe the SHARAD observations
however, Herkenhoff et al., 2007, who suggest that the apparent and the methods we use for analyzing them. In Section 3, we dis-
brightness of layers is controlled more by topography, texture, cuss our rationale for assigning and interpreting radar units and
and surficial frosts than by internal composition). High-resolution their boundaries, highlighting features of Planum Boreum revealed
images of Planum Boreum indicate that a major division exists be- by the radar (Figs. 2–8). In Section 4, we explain our methods for
tween the upper stack of layered deposits (referred to hereinafter interpolation, extrapolation, and conversion to depth of the unit
as the ‘‘NPLD”) and a lower basal unit (BU), where the lower unit boundaries, which were interpreted originally in delay time. These
is typically darker and its layers less continuous (Byrne and Mur- steps yield isopach and elevation maps (Figs. 9 and 10) as well as
ray, 2002; Fishbaugh and Head, 2005). From the image data, the precise volume estimates for the NPLD as a whole and for its con-
BU is inferred to extend beneath most of the main lobe of the polar stituent radar units (Table 1). In Section 5, we present a discussion
cap but generally not beneath the Gemina Lingula lobe (Fig. 1). This of our results and their implications for the history of the NPLD and
inference led Fishbaugh and Head (2005) to suggest that the loca- the Martian climate. We also briefly discuss smaller-scale features,
tion of Chasma Boreale (Fig. 1) may have been controlled by that of such as unconformities and structures associated with troughs. In
the BU. On the basis of morphological differences, Tanaka et al. Section 6, we summarize our most important findings and outline
(2008) subdivide the BU into the Rupes Tenuis unit, a terraced low- avenues for future research.
er sequence interpreted to be of Early Amazonian age, and the Pla-
num Boreum cavi unit, a more resistant upper sequence
2. Observations
interpreted to be of middle to late Amazonian age. We see no fea-
tures in the radar data that would allow differentiation of these
The SHARAD instrument is a chirped-pulse sounding radar with
geologic units, so we use the collective term ‘‘basal unit” or ‘‘BU”
a 10-MHz bandwidth centered at 20 MHz, which yields a free-
when referring to these materials. ‘‘Planum Boreum” refers to the
space range resolution of 15 m (8.4 m in water ice with a real per-
polar-plateau landform, which consists of the BU and the NPLD.
mittivity of 3.15). The altitude of the MRO orbit varies between
The Mars Advanced Radar for Subsurface and Ionospheric
about 250 and 320 km, allowing a lateral resolution at the surface
Sounding (MARSIS) instrument onboard the Mars Express orbiter
of 3–6 km (1–2 Fresnel zones), which can be reduced in the azi-
obtained the first subsurface profiles from Planum Boreum, which
muth (along-track) direction to 0.3–1.0 km with synthetic-aper-
showed a strong radar return from the base of the deposits (Picardi
ture focusing. See Seu et al. (2007b) for a detailed discussion of
et al., 2005; Phillips et al., 2008, SOM; Selvans et al., 2009). More
the SHARAD experiment.
recently, observations from the Shallow Radar (SHARAD) instru-
A SHARAD observational campaign for the north polar region
ment onboard the Mars Reconnaissance Orbiter (MRO) show a
has yielded dense coverage in this region, with observations ac-
weaker, diffuse radar return from the base, but many more internal
quired on more than 1300 orbits through March 2009. Typically,
reflections from overlying layered materials (Phillips et al., 2008).
SHARAD observations of the polar region are taken over a 700-s
Taken together, these results exemplify the complementary nature
period and span 20° of latitude on either side of the pole
of the two instruments, with MARSIS providing greater depth of
(2000 km), but the duration and coverage of individual data takes
penetration and SHARAD finer vertical resolution (10 m in water
may vary substantially. We selected 358 observations for this
ice, vs. 100 m for MARSIS) (Seu et al., 2007b). The work presented
study, providing the coverage shown in Fig. 1. We present portions
here will focus on the SHARAD observations, which provide greater
of six representative observations where they extend across Pla-
detail of internal structures and thus allow a more direct compar-
num Boreum in Figs. 2–8.
ison to image-based geologic interpretations.
Subsurface radar data are typically displayed as radargrams,
with distance along track shown horizontally and power vs. delay
time displayed vertically in an image format (e.g., Figs. 2–8, top pa-
Olympia Planum nel). While delay time is roughly analogous to depth, such radar-
grams exhibit a geometric distortion due to the combination of
surface topography and the difference in signal speed in free space
Fi
g.

vs. that in typical geological materials (e.g., regolith or water ice).

7
57

Furthermore, reflections from off-nadir surface features may arrive

77
02

at delay times similar to those from true subsurface interfaces, and

a careful analysis of the sources of such surface clutter is required.
main lobe Fi
Fig.
6 57 g. We now consider approaches to accounting for these effects.
890 8
2 52
16
To address the geometric distortion, one may convert the delay-
Planum 02 time radargram to depth on a sample-by-sample basis as:
Boreum
Ddi ¼ cDt=2e1=2
i ð1Þ
Chasma Boreale
where Ddi is the depth interval for the ith sample in delay time, c is
the speed of light in free space, Dt is the delay-time sampling inter-
Fig. 2

03
20 Gemina Lingula val, and ei is the real permittivity of the traversed medium, with the
52
.5 divisor of two accounting for two-way travel (i.e., radar transmit
02
Fig. 4

Fig
529701
6247

and receive paths). To create depth-converted radargrams, we first

100 km
identify the samples associated with the surface return and then ap-
562002

Vastitas
Fig. 3

ply Eq. (1), assuming values for ei of 1.0 above the surface and 3.15
Borealis 75
°N
below, where the latter value is typical of pure water ice under
Martian surface conditions (Johari, 1976; Mätzler and Wegmüller,
Fig. 1. Polar stereographic location map of Mars’ north polar region, showing 1987; Grima et al., 2009). Where the initial surface return is off-na-
SHARAD ground tracks (tangent to limit of MRO orbit at 87.4°N) for observations
used in this study, overlain on MOLA shaded relief. Bold lines are labeled with
dir, a distortion in the depth-converted radargram can occur (e.g.,
observation numbers corresponding to radargrams shown in Figs. 2–8. Zero see Figs. 3 and 8). Our assumption of nearly pure water ice in the
longitude is at bottom-center of figure. NPLD is based on estimates of their lithic content derived from
 N.E. Putzig et al. / Icarus 204 (2009) 443–457 445

Fig. 2. SHARAD observation 529701, crossing the topographic saddle between the main lobe and Gemina Lingula. Delay-time radargram (top panel) is overlain with
interpreted boundaries for radar units A–G, dashed where extrapolated. Extrapolated boundaries may not align with radar reflectors (see text). Unit A extends beneath Unit B,
but the contact is not typically evident in SHARAD data. Red arrow shows lateral extent of apparent angular unconformity at 2.5 ls delay time in Unit G. Depth-converted
radargram (middle panel), which is created by assuming a subsurface real permittivity of 3.15 (nearly pure ice), shows proper geometric relationships at 45:1 vertical
exaggeration, with packets of Phillips et al. (2008) in red and approximate correlations to geological units of Tanaka (2005) in yellow (VBF: Vastitas Borealis Formation; BU:
basal unit; LLD: lower layered deposits; ULD: upper layered deposits). Colored arrows show reflectors corresponding to unit boundaries in upper panel. Layer discontinuities
occur progressively closer to the southern extent of the BU with depth (white arrows). MOLA elevation map (bottom panel) shows location of ground track.

the attenuation of radar signals (Phillips et al., 2008; Picardi et al., 2005; Holt et al., 2008; Plaut et al., 2009). Extra caution is advised
2005; Grima et al., 2009). when interpreting radar observations made in the vicinity of areas
The potential for surface clutter may be addressed qualitatively with sparse elevation data, such as the regions poleward of 87° lat-
by examining a map of surface topography in the vicinity of the itude, as surface features missing from the elevation maps may
radar ground track (e.g., Figs. 2–8, third panel). Radar signals yield clutter in the radar data that does not appear in the simulated
reflected from spacecraft-facing facets of topographic highs or radargrams. We point out an example in Fig. 7, where troughs
depressions on either side of the ground track may produce a re- poleward of 87°N that are seen in the Viking image mosaic (fourth
turn that is delayed in time relative to the nadir surface return in panel) are missing from the MOLA map (third panel), and these
proportion to the difference in distance from the spacecraft. Exam- troughs produce surface clutter visible in the radargram (first pa-
ples of strong surface clutter are identified in Fig. 6, where returns nel) but not in the MOLA-generated simulation (bottom panel).
from surface troughs offset from the ground track appear as dip-
ping reflectors subsequent to the surface returns. For subtle fea-
tures, a more quantitative approach to characterizing clutter may 3. Radar unit delineation
be taken by producing simulated radargrams, which can be gener-
ated from topographic information such as an elevation map of In Planum Boreum, SHARAD radargrams show a packet of clo-
Mars Orbiter Laser Altimeter (MOLA) data. Such synthetic radar- sely spaced, strong reflections to 5 ls delay time below the sur-
grams are used routinely in distinguishing probable subsurface face. These reflections are typically conformal with the surface
detections from surface clutter in terrestrial studies (e.g., Holt return, but occasionally include apparent angular unconformities
et al., 2006) and in MARSIS and SHARAD data (e.g., Picardi et al., at depth (e.g., see Fig. 2). The packet of strong reflectors is generally
446 N.E. Putzig et al. / Icarus 204 (2009) 443–457

Fig. 3. Radargrams for SHARAD observation 624702 in delay time (top panel) and depth (middle panel; vertical exaggeration 100:1), extending from Olympia Planum,
across the main lobe, just east of the head of Chasma Boreale, and across Gemina Lingula. MOLA elevation map (bottom panel) shows location of ground track. Brace in depth
radargram shows zone of distortion due to off-nadir surface returns. Radar units, depth-conversion methods, and MOLA scale same as in Fig. 2.

underlain by a zone of much weaker or negligible returns 2–3 ls zone, which we interpret as a likely extension of the Vastitas Bore-
in round-trip delay time. At later delay times, several similar se- alis Interior Unit (ABvi) (Tanaka et al., 2008)—often referred to as
quences of packet–inter-packet reflections follow, but the reflec- the Vastitas Borealis Formation (VBF) (Tanaka and Scott, 1987)—
tors are typically reduced in number, frequency, and relative beneath Gemina Lingula. Unit B, the thicker diffusely reflective
backscattered power. Much of this behavior may be attributable zone, has been correlated with the BU discussed earlier (Phillips
to attenuation of the radar signal by transmission and path losses et al., 2008). Although not evident in most of the SHARAD data
rather than actual increases in layer thickness or reductions in shown here, Unit A extends beneath Unit B and sometimes dis-
dielectric contrast (e.g., see Nunes and Phillips, 2006). The individ- plays a weak reflection under the main lobe; this same contact
ual reflectors within packets and the repeated packet sequences yields the strong basal reflections seen by MARSIS (Picardi et al.,
represent two periodicities, which Phillips et al. (2008) related to 2005; Phillips et al., 2008, SOM; Selvans et al., 2009). Hints of a re-
global climatological cycles of Mars driven by orbital and rotational turn from this depth are seen in SHARAD observation 522003
dynamics (Laskar et al., 2002). In designating the units within the (Fig. 5). Additionally, a reflector likely corresponding to Unit B ap-
zone of packet sequences for this study, we chose the top of each pears only rarely in radargrams over Olympia Undae (not shown
packet as the unit boundary, with the inter-packet zone included here). Given these inconsistent returns, we did not attempt to
in the base of the same unit. We found that the uppermost reflec- map Units A and B in these areas, nor do we map them where they
tions in the packets are relatively consistent and traceable across occur at the surface beyond Planum Boreum. Overlying Units A and
many radargrams, whereas the lowermost reflections in the pack- B is a set of quasi-parallel reflecting horizons 200–300-m thick
ets occur intermittently, implying a less distinct transition. Be- that appear to extend throughout the NPLD, which we designate
neath these packet–inter-packet sequences, SHARAD radargrams as Unit C. The upper boundary of this unit is identified by the pres-
typically show zones of reflections that are diffuse (but signifi- ence of a ‘wedge’ of nearly parallel, reflecting interfaces up to
cantly above the instrument noise background) with durations of 300-m thick that is confined to the eastern part of Gemina Lin-
5–10 ls beneath most of the main lobe and 1–3 ls beneath Gemin- gula and defined here as Unit D (Figs. 2, 3 and 7). Another set of
a Lingula, portions of the main lobe in the eastern hemisphere, and near-parallel reflectors, often associated with an inter-packet zone
the topographic saddle between the two lobes. The duration of the of modestly diffuse returns at its base, appears to be ‘draped’ over
latter zone of diffuse energy is similar to that observed just below those of Units C and D, and we designate these as Unit E. This unit
the surface of the Vastitas Borealis materials that surround Planum is in turn overlain by a packet–inter-packet sequence identified
Boreum. here as Unit F. Finally, Unit F is itself overlain by the uppermost
We designated a working set of radar-based geologic units packet–inter-packet sequence that extends to the surface and is
(Fig. 2), beginning with Unit A as the thinner diffusely reflective designated as Unit G.
 N.E. Putzig et al. / Icarus 204 (2009) 443–457 447

Fig. 4. Radargrams for SHARAD observation 562002 in delay time (top panel) and depth (middle panel; vertical exaggeration 100:1), extending from Olympia Planum
across the main lobe, Chasma Boreale, and Gemina Lingula. MOLA elevation map (bottom panel) shows location of ground track. Radar units, depth-conversion methods, and
MOLA scale same as in Fig. 2.

Using commercial seismic-data analysis software (provided by RAD observation 577702 (second panel in Fig. 7), one sees the
SeisWare International Inc.), we delineated the unit contacts, top of Unit B rising gradually before a ‘precipitous’ drop of
beginning with radargrams that cross the topographic saddle be- 600 m near 85°N, 260°E. While the elevation change appears
tween the main lobe and Gemina Lingula. The radar units de- quite abrupt on the vertically exaggerated depth radargram, the
scribed above are best resolved in this area, which is actual slope is only 3–4° (see discussion in Section 5). Continuing
topographically smooth over hundreds of kilometers, having no on, the boundary again takes on a dome-like profile as it ap-
troughs and few surface undulations (Rodriguez et al., 2009), and proaches the periphery of Planum Boreum near eastern Olympia
thus we chose to begin correlations of unit boundaries here. Pro- Planum. In addition, the character of the radar returns changes in
cessing of the SHARAD data employs a delay-time datum that var- this zone such that the backscattered power at the boundary is
ies from one radargram to another. Unit correlation and datum greater and the diffuse power normally associated with the under-
adjustment are handled within the data analysis software, which lying Unit B is diminished. The radargram for SHARAD observation
allows the display on each radargram of interpreted horizons from 578902 demonstrates similar behavior across this region, including
other, intersecting radargrams, and the interactive adjustment of topographic and reflectivity anomalies in Unit B (see Fig. 6), and
the radargram datums. Designated unit boundaries were trans- MARSIS data show disruptions in this same region (Selvans et al.,
ferred in this fashion from one radargram to another, with datums 2009).
adjusted accordingly, and the units were systematically extended
as far as possible throughout Planum Boreum (see solid colored
lines in the top panels of Figs. 2–8). 4. Mapping
As demonstrated by the depth-converted versions of the radar-
grams (second panels of Figs. 2–8), many of the reflector undula- Once the unit boundaries were delineated on the selected de-
tions seen in the delay-time radargrams are actually due to the lay-time radargrams, we used the data analysis software to map
combination of complex topographic relief at the surface and the them to a regular grid with cells of 2.7 km  2.7 km. First, we inter-
difference in radar wave speeds of the atmosphere and subsurface polated the surface horizon, mapping it everywhere poleward of
materials. The true geometry of the subsurface returns is shown to 75°N. Next, we interpolated the tops of the basal Units A and B,
be much smoother in the depth-converted radargrams—for exam- which together completely cover the region at the base of the NPLD
ple, those returns corresponding to the top of Unit A are nearly out to their intersections with the surface grid. Then, as we inter-
flat-lying everywhere and those associated with the top of Unit B polated the internal boundaries (tops of Units C–F), we also extrap-
appear as a smoothly varying dome, with one notable exception. olated them to the periphery of the NPLD to their intersections
From right to left across the depth-converted radargram for SHA- with either the surface grid or the basal grid. We differenced the
448 N.E. Putzig et al. / Icarus 204 (2009) 443–457

Fig. 5. Radargrams for SHARAD observation 522003 in delay time (top panel) and depth (middle panel; vertical exaggeration 100:1), extending from Olympia Mensae
across the main lobe, Chasma Boreale, and Gemina Lingula. MOLA elevation map (bottom panel) shows location of ground track. Arrow indicates rarely seen returns from
below Unit B that are interpreted as possible Top Unit A. Radar units, depth-conversion methods, and MOLA scale same as in Fig. 2.

radar-unit delay-time grids, converting them to thickness using Eq. The NPLD isopach map, which represents all of the material
(1) and assuming a subsurface real permittivity of 3.15 to produce overlying Units A and B (i.e., Units C–G collectively), demonstrates
the isopach maps shown in Fig. 9. From these maps, we computed that much of the difference in elevation between Gemina Lingula
the average thickness, extent by area, total volume, and percentage and the main lobe (see Surface elevation map in Fig. 10) can be
of volume in the extrapolated region for each of the radar units and attributed to the BU deposits (Unit B), with the stack of NPLD lay-
for the NPLD as a whole, and we report these values in Table 1. ers in Gemina Lingula approaching the same maximum thickness
Additionally, we counted the typical number of radar layers within of 2 km that it does in the main lobe. At their highest point, the
each unit and we report estimates for their average thicknesses, BU deposits stand about 1 km above the surrounding terrain (see
also in Table 1. The total NPLD volume of 821,000 km3, which rep- Base NPLD map in Fig. 10), which presumably represents their
resents a global-equivalent layer of 5.7 m, is 27% smaller than the maximum thickness above the flat-lying Vastitas Borealis deposits.
estimate by Smith et al. (2001) of 1,140,000 km3 (where they as- The isopach maps for Units C, E, and F show that these three
sumed no flexure and pure water ice). While they used a slightly units are of similar thickness and geographic distribution to each
larger area (1,120,000 km2 vs. our 949,000 km2), most of the other, each relatively uniform in thickness in the interior of Planum
319,000-km3 difference in volume is due to our exclusion of the Boreum and tapering quickly at the periphery. This similarity sug-
BU from these calculations. An effort to better constrain the vol- gests that the deposition of these units was a repetitious process
ume of the BU from MARSIS observations is ongoing (Selvans that did not vary greatly in its intensity. In contrast, Unit D appears
et al., 2009). wedge-shaped in its thickness variations and is geographically re-
We created an elevation map for the base of the NPLD together stricted to a substantially smaller area in eastern Gemina Lingula.
with the surrounding surface (Base NPLD map in Fig. 10) by first Relative to Units C, E, and F, the uppermost sequence—Unit G—is
converting the surface delay-time grid to elevation (Surface map thicker, has a more dome-like profile, and is restricted to a some-
in Fig. 10) and then subtracting the NPLD isopach map from it. what smaller region, tapering out further up in Planum Boreum.
By progressively adding each of the isopach maps for Units C–F An important caveat is that the distal geographic extent of the
to the Base NPLD elevation map, we obtained the set of elevation internal unit boundaries is controlled to a large extent by our
maps for these internal unit boundaries shown in the other panels rather simplistic extrapolation of the boundaries out to the periph-
of Fig. 10. In each map shown in Figs. 9 and 10, the region poleward ery of the deposits. It is a distinct possibility that these extrapola-
of the oval near 87.4°N has no coverage by SHARAD and hence the tions do a poor job of tracking actual depositional or erosional
values there are entirely interpolated. This region is included in the boundaries—for instance, one might argue that the highly reflec-
area and volume estimates shown in Table 1 and represents about tive zones near the periphery actually correspond largely to Unit
13% of the volume for each radar unit and for the NPLD as a whole. G rather than to Units E, F, and C (e.g., see Fig. 5 near 84°N,
 N.E. Putzig et al. / Icarus 204 (2009) 443–457 449

Fig. 6. Radargrams for SHARAD observation 578902 in delay time (top panel) and depth (middle panel; vertical exaggeration 100:1), extending from Rupes Tenuis across
the main lobe to 62°E. MOLA elevation map (bottom panel) shows location of ground track. Pronounced surface clutter comes from trough walls running subparallel to
ground track. Vertical arrow shows location of anomalous topography and reduced internal reflectivity of Unit B (see discussion in text). Radar units, depth-conversion
methods, and MOLA scale same as in Fig. 2.

100°E and Fig. 6 near 84°N, 280°E). In that case, the distal extent of high permittivity suggests a minimum dust loading of 85%,
the interior units is likely to be substantially different than por- approaching 100% if reasonable values of attenuation in the ice
trayed in the isopach maps (Fig. 9), wherein the extent of Unit G are included. Many of the other, fainter reflections in this region
ought to be greater and that of Units E, F, and C ought to be lesser. could be explained by a thin layer with 30% dust content.
Nevertheless, the accuracy of the maps is high in their interiors The fact that the strong echoes exhibit relatively modest power
where we explicitly chose the boundaries, and we believe that they reduction with increasing depth indicates that the total column
are broadly representative of the actual thicknesses and volumes of loss due to attenuation by lithic materials (dust or sand) is very
the materials associated with these units. In any case, the esti- low—modeling suggests at most a few percent of the total NPLD
mates of thickness, extent, and volume for the NPLD as a whole volume (Picardi et al., 2005; Nunes and Phillips, 2006). Thus, each
are robust. of the intermediate-power reflections (those 6 dB down from the
surface return) might arise from one of two mechanisms: (1) a sin-
gle layer with a high concentration of dust or sand no more than
5. Discussion 1 m in thickness (based on a 2-km layer with 40–50 bright
reflectors and a 1–2% column abundance of lithic material); or
For a particular target medium, a reflecting horizon in SHARAD (2) alternating, parallel bands of clean and moderately lithic-rich
data includes the radar returns from all dielectric interfaces within ice within the vertical-resolution cell. In the latter case, it is the
the vertical-resolution cell (8.4 m in pure water ice). The strong, net reflectivity of these densely spaced layers that creates the ob-
parallel reflections that characterize the packet regions are only served bright return. For example, if the echo power is an incoher-
somewhat weaker than the radar echoes from the surface. For ent sum of reflections, then about four 1-m layers with 30% dust
example, the echo power of the brightest subsurface reflector be- in ice could produce echos with a 6-dB reduction in power rela-
low the saddle region in observation 529701 (Fig. 2), occurring at tive to the surface.
about 190-m depth, is only 2 dB down from that of the surface Given that image data clearly show pervasive layering at scales
return. Other intermediately bright reflections in this zone, which well below the SHARAD vertical-resolution cell size, we favor the
are more typical of Unit G throughout Planum Boreum, are 6 dB multi-layer model as representative of at least the intermediate
down from the surface return. The reflectivity of the surface echo reflections from the banded packets. However, SHARAD data alone
is 0.08, so that of the brightest subsurface return is at least 0.05 cannot resolve the nature of any given bright reflector. The multi-
(ignoring attenuation), which requires a real permittivity of 7.7 layer model may introduce complications with regard to interfer-
in contrast with the assumed value of 3.15 for clean ice. Such a ence between returns from interfaces separated by fractions of
450 N.E. Putzig et al. / Icarus 204 (2009) 443–457

Fig. 7. Radargrams for SHARAD observation 577702 in (a) delay time and (b) depth (vertical exaggeration 100:1) extend from Olympia Planum across the main lobe and
Gemina Lingula. (c) MOLA gridded elevation map and (d) Viking MDIM mosaic show location of ground track. (e) Simulated delay-time radargram from MOLA data and a
geometrical optics, facet-based scattering model with incoherent summation (Holt et al., 2006) identifies as surface clutter many returns with steep apparent dip. Other
returns (e.g., yellow arrow in (a)) do not appear in the simulation but are attributable to troughs (e.g., yellow arrow in (d)) poleward of the 87°N limit of the MOLA data
(dashed black semicircles). Red shading in (e) shows region affected by the no-data zone. White arrow in (b) identifies a large elevation anomaly at top of Unit B. Radar units,
depth-conversion methods, and MOLA scale same as in Fig. 2.

the wavelength in the medium. While we cannot determine the suggested that its location may have been influenced by the south-
dielectric structure of the NPLD at scales below the SHARAD verti- ern boundary of the BU, but its extension into Gemina Lingula re-
cal resolution, it is plausible that bright echos arise from alternat- vealed by the radar data indicates that the relationship between
ing, closely spaced layers of clean ice and layers of modest dust the BU and Chasma Boreale is not so straightforward. The possibil-
abundance. ity remains that pre-existing topographic features adjacent to and/
Radargrams that cross Chasma Boreale (Figs. 4 and 5) clearly or atop the BU played a pivotal role in the formation of the Chasma
show that Unit B (i.e., the BU) extends southward into the Gemina (Holt and Safaeinili, 2009).
Lingula lobe. These radar results confirm an earlier suggestion by In the earlier image-based analyses of the BU, it was thought
Tanaka et al. (2008) based on surface images that the BU is not en- that these deposits extended nearly everywhere beneath the main
tirely confined to the main lobe as was originally thought (Byrne lobe, but the SHARAD data show that the BU is absent below a
and Murray, 2002; Fishbaugh and Head, 2005). In discussing sce- large portion of the main lobe to the east of the saddle region
narios of Chasma Boreale formation, Fishbaugh and Head (2005) (20°E–110°E) and south of 85°N (compare the boundary be-
 N.E. Putzig et al. / Icarus 204 (2009) 443–457 451

Fig. 8. Radargrams for SHARAD observation 521602 in delay time (top panel) and depth (middle panel; vertical exaggeration 100:1), extending from Olympia Planum
across the main lobe to 72°E. MOLA elevation map (bottom panel) shows location of ground track. Radar returns are highly disrupted and interpreted unit boundaries are
mostly extrapolated in this region that borders Olympia Planum. The morphology of the basal unit is inaccurate in the depth-converted radargram due to off-nadir surface
returns. Radar units, depth-conversion methods, and MOLA scale same as in Fig. 2.

tween Units A and B with the BU boundary of Fishbaugh and Head, (identified by the white arrow). Such a remnant deposit may con-
2005, on the surface elevation map in Fig. 10). SHARAD radargrams tain less internal contamination by lithic materials than elsewhere
clearly and consistently show the absence of Unit B in this region of in Unit B and perhaps a thick lag deposit on its surface, which could
the main lobe (e.g., Figs. 5, 6 and 8). Either the BU was never depos- explain the reduced power in diffuse radar returns from within the
ited in this region or it was eroded away prior to the deposition of unit and increased power of the radar return from the top of the
Unit C. unit (e.g., see Figs. 6 and 7). Another possibility is that the anoma-
The elevation anomaly in Unit B that we mention in Section 3 lous feature is an erosional remnant of a pedestal preserved by the
and identify in SHARAD observations 578902 (Fig. 6) and 577702 ejecta of a large impact into Unit B that was subsequently buried
(Fig. 7) provides an intriguing puzzle. The fact that this same fea- by deposition of the overlying units. An analogous feature might
ture occurs at similar delay times on all radargrams crossing this be the irregular, 700-m-high plateau around Escorial crater that
region and the lack of surface features that might cause clutter at is currently exposed at the surface and thought to be made up of
these delay times or induce large errors in the depth conversion Rupes Tenuis materials (Tanaka et al., 2008).
provides confidence that the returns are indeed attributable to a With the exception of Unit D, all of the NPLD radar units extend
subsurface elevation anomaly atop Unit B. In map view (see Base across much of the main lobe and Gemina Lingula, indicating that
NPLD map in Fig. 10), this feature appears as a longitudinally elon- deposition occurred regionally throughout most of the NPLD for-
gated peak at the highest elevation of Unit B, with relatively steep mation history. The series of elevation maps in Fig. 10 suggests that
slopes to the south and shallower slopes to the west, north, and the highest point (or at least the highest point south of the unob-
east. While it is tempting to suggest a normal fault in the BU that served area poleward of 87°N) remained near 85°N, 260°E for a
was subsequently buried by the NPLD as an explanation, when one long period of time, presumably established during an erosional
accounts for the vertical exaggeration of 100:1 in the depth event represented by the base of the NPLD and persisting through
radargrams, the true slope of the putative fault plane is actually the deposition of Units C–F. While there is some evidence for inter-
quite low (3–5°), and normal faults of such low angle are rare. nal unconformities that may represent erosional periods (e.g.,
Perhaps a more likely explanation is that this portion of Unit B rep- within Unit G in Fig. 2), the units and their internal layers are
resents a remnant of an earlier polar ice deposit, and the 600-m- remarkably conformable to each other in most areas. Any erosion
high slope is analogous to scarps of present-day Planum Boreum. that may have occurred must have either stripped materials off
For example, Fig. 7 shows an interior surface scarp of similar height uniformly or been restricted geographically, perhaps largely con-
and slope angle about 80 km south (left) of the anomalous feature fined to the periphery of Planum Boreum.
452 N.E. Putzig et al. / Icarus 204 (2009) 443–457

Fig. 9. Polar stereographic isopach (thickness) maps for the entire NPLD and for Units C–G from SHARAD data, assuming subsurface real permittivity of 3.15 (nearly pure
water ice). Contour intervals are 200 m for NPLD map (left scale bar) and 100 m for individual unit maps (right scale bar). On each map, a dashed yellow outline encompasses
basal Units A (lower right section) and B (upper left section). Shaded oval extends to northernmost extent of MRO ground tracks at 87.4°N.
 N.E. Putzig et al. / Icarus 204 (2009) 443–457 453

Fig. 10. Polar stereographic elevation maps of the base of NPLD (i.e., top of Units A and B), the top of Units C–F, and the surface, derived from SHARAD data. A subsurface real
permittivity of 3.15 (nearly pure water ice) is assumed. Contour interval is 250 m. On each map, a dashed yellow outline encompasses basal Units A (lower right section) and
B (upper left section). Shaded oval extends to northernmost extent of MRO ground tracks at 87.4°N. Black dashed line in surface map (lower right panel) is limit of basal unit
as mapped by Fishbaugh and Head (2005).
454 N.E. Putzig et al. / Icarus 204 (2009) 443–457

Table 1
NPLD radar-unit number of layers, thicknesses, areas, and volumes.

Unit # Layers Mean unit thickness (m) Mean layer thickness (m) Area (km2) Volume (km3) Extrapolated volume (%)
G 21 432 21 537,000 232,000 19
F 12 254 21 672,000 171,000 34
E 10 281 28 815,000 229,000 45
D 4 192 48 71,000 14,000 29
C 6 227 38 765,000 174,000 44
NPLD 48a 864 18 949,000 821,000 35b
a
Does not include layers in the geographically isolated Unit D.
b
Average for internal units; total NPLD volume uses surface and basal picks with no extrapolation.

Unit D represents a potentially important exception to the Modeling of polar-ice accumulation that incorporates the evo-
above generalizations. It is likely that these deposits have been lution of orbital characteristics together with effects from the peri-
greatly eroded on the periphery together with the overlying units odic formation of lag deposits was performed by Levrard et al.
(Holt and Safaeinili, 2009), and it is possible that detailed mapping (2007), who obtained a maximum age of 4.2 Ma for the NPLD
of sub-unit layering may reveal a greater lateral extent for Unit D. (see base graphs in Fig. 11). Their results suggest that periods when
In either case, it is evident from our mapping that Unit D is more the amplitude of insolation variation is low, which correspond to
variable in thickness and restricted in extent than the overlying periods of lower-amplitude obliquity variations (cf. second panel
units and thus represents substantially different depositional in Fig. 12, taken from Fig. 2d of Laskar et al., 2002), have relatively
and/or erosional patterns early in the history of the NPLD. rapid accumulation of ice in the polar region. The intervening
Estimates for the age of the NPLD based on correlations of lay- times, when insolation varies more widely, are predicted to yield
ering seen in image data with Mars’ orbital characteristics or with alternating periods of accumulation and sublimation that cause
other data range from a few million years to hundreds of millions the concentration of lithic materials in lag deposits (see gray shad-
of years (Laskar et al., 2002; Milkovich and Head, 2005; Tanaka ing in bottom panel of Fig. 12). In this scenario, the periods of low-
et al., 2008). The lower estimates are generally based on climate amplitude oscillations would correspond to the zones of low radar
modeling studies (Laskar et al., 2002; Milkovich and Head, 2005; power defining the base of the radar units (i.e., the inter-packet
Levrard et al., 2007), whereas the higher estimates are constrained zones), with the ice experiencing less sublimation—and thus less
by sparse but relatively large craters that may represent impacts in concentration of lithic materials—during the lower insolation
some of the lower sequences of the NPLD and by an unconformity peaks, thereby producing lower dielectric contrasts between suc-
in the upper part of the NPLD (Tanaka, 2005; Tanaka et al., 2008, cessive deposits. At other times, each interval between the higher
2009). local maxima in the insolation function might then correspond to

Fig. 11. Proposed correlation of radar Units C, E, F, and G with the summer-solstice insolation function and modeled polar-ice thickness of Levrard et al. (2007; their Figs. 3
and 4). Packets (P) of strong reflections in the radar, shown in darker shading of E–G, correspond to low-amplitude oscillations in insolation, and zones of weaker reflections
(C, lighter shading of E–G) correspond to high-amplitude oscillations and short-period lows in insolation. See discussion in text.
 N.E. Putzig et al. / Icarus 204 (2009) 443–457 455

Fig. 12. Various correlations of radar Units C, E, F, and G with obliquity, eccentricity, and insolation functions of Laskar et al. (2002; their Fig. 2d–f). Levrard et al. (2007)
predicted layering by lag formation during local insolation maxima (darker gray shading on bottom panel) and little or no lag forming when oscillations are suppressed
(lighter gray shading), with the latter presumably leading to inter-packet zones (IP). However, no such zone appears to exist in the uppermost part of the NPLD. Phillips et al.
(2008) suggested two alternative correlations, one in which inter-packet zones correspond to lows in eccentricity (top panel) and another where they correspond to short-
period lows in obliquity (second panel). We modify their second scenario by correlating packet zones (P) to low-amplitude oscillations of insolation (third panel). See
discussion in text.

individual radar-detected layers in the strongly reflective portions et al., 2003) deliver lithic material to the polar region and concen-
of the units (i.e., the packet zones) (see color shading in bottom pa- trate it as lags during excursions in insolation (Levrard et al., 2007).
nel of Fig. 12). The insolation function shows about 45 local max- In the second scenario, the inter-packet zones are correlated to
ima in the 4.2-Ma period, which appears to correspond closely to phases of short-period minimum obliquity at 0.8, 2.0, and
our estimate of 48 NPLD radar layers (Table 1). However, the fact 3.2 Ma when ice may be accumulating more rapidly at the poles
remains that the most recent 300,000 years are a period of low- (Levrard et al., 2007). The first scenario extends the age of the
er-amplitude oscillations in insolation and yet there is no zone of deposits to 6.5–6.8 Ma, well before the excursion to low mean
low radar power at the surface of the NPLD as might be expected, obliquity of the last 5 Myr, a period which the models suggest
but rather a strongly reflective sequence. We therefore discount are not compatible with large deposits of polar ice at or near the
this scenario in favor of others that do not require formation of surface. The timing in the second scenario is more favorable, but
the inter-packet zones during periods with low-amplitude varia- it must contend with short-period excursions to higher obliquity
tions in insolation. during the inter-packet periods and secularly variable insolation
Considering these and other results, Phillips et al. (2008) sug- during the periods of packet formation.
gested two other possible correlations between the SHARAD pack- To address the latter issue, we modify the second scenario of
et–inter-packet sequences and climate forcing due to orbital Phillips et al. (2008) by confining the period of packet formation
characteristics (upper two panels of Fig. 12). In the first scenario, to the lows in the oscillation of the insolation function of Levrard
the inter-packet zones correspond to lows in the 2.4-Myr cycle of et al. (2007), accepting their maximum age constraint for the NPLD
eccentricity at 1.4, 3.8, and 6.2 Ma (Laskar et al., 2002) and of 4.2 Ma (see color shading in Fig. 11). The dielectric contrasts
represent accumulation of ice with little dust, consistent with the between layers in the packet zones would then be driven by vari-
expectation of low dust-storm activity at times of low obliquity ations in dust content during ice accumulation rather than by the
and eccentricity (Toon et al., 1980). Meanwhile, the packets form production of lags during sublimation, in keeping with the mecha-
during high obliquity when more prevalent dust storms (Haberle nism for layering preferred by Laskar et al. (2002). In this revised
456 N.E. Putzig et al. / Icarus 204 (2009) 443–457

scenario, the longer periods with high-amplitude oscillations of from the periphery and troughs where the layers outcrop (e.g., Clif-
insolation correspond to the weakly reflective inter-packet zones ford et al., 2000; Tanaka et al., 2008), they also provide a substan-
at the base of the radar units. These periods still encompass the tially more detailed view of that structure and reveal many new
obliquity minima, but nevertheless remain more difficult to ex- features. The unveiling by radar sounding of the basal unit’s true
plain, as one might expect significant lag production leading to geographic extent—present in Gemina Lingula and absent in the
large dielectric contrasts during the short-period excursions to eastern main lobe—and of its subsurface topography—a generally
high insolation values. However, such may not be the case if sub- dome-like structure, with an anomalous elongated peak beneath
limation processes are less efficient than predicted or eolian activ- the western main lobe—raises interesting new questions about
ity removes any lag materials prior to the return to accumulation the history of these deposits and their influence on the subsequent
of relatively clean ice at low obliquity. Experiments with an atmo- evolution of the NPLD. A wedge of layers discovered in eastern
spheric general circulation model (Haberle et al., 2003) suggest Gemina Lingula hints at a dramatic shift in the north-polar ice
that, during times of high obliquity, atmospheric dynamics are depocenter early in NPLD history, attended by a restricted area of
favorable for lifting dust at high latitudes. deposition. The periodic nature of the layers and packets of layers
With regard to the higher age estimates, Tanaka et al. (2009) seen in the radar data provides a tantalizing new piece to the puz-
suggest a correspondence between the high obliquity period prior zle of the age of the NPLD. Constrained by clear evidence of sur-
to 4.5 Ma and an unconformity near the top of the NPLD that is face-conformal, strongly reflective layers in the upper few
too shallow to be resolved by SHARAD. In this scenario, nearly all hundred meters of the NPLD, we have proposed a seemingly coun-
of the radar-detected layers pre-date the precisely established terintuitive correlation between the radar layering pattern and
obliquity history (unconstrained prior to 10–20 Ma; see Laskar modeled cycles of insolation over the past five million years. We
et al., 2004) and no correlation of insolation history to the bulk expect that more rigorous testing of this idea will supply fresh
of NPLD layering is possible. Nevertheless, the cyclic and relatively insights.
conformal nature of the radar-detected layering could still be use- The advent of this rich dataset sheds new light on several
ful for constraining possible obliquity histories at earlier times. For important and long-standing mysteries and simultaneously opens
example, if it can be shown by other means that the layering is a host of intriguing new avenues for investigation. Our mapping of
likely to have been emplaced in the last few 100 Ma and not earlier, the NPLD base confirms that it is extremely flat beneath Gemina
then those orbital solutions that retain high obliquity over ex- Lingula and the eastern main lobe (Fig. 10), implying an extraordi-
tended periods of time prior to the 4.5-Ma transition can be dis- nary lack of significant isostatic compensation as discussed by Phil-
counted (e.g., 250-Ma obliquity histories in panels 1, 2, 10, 11, lips et al. (2008). This result has far-reaching ramifications for
13, and 15 in Fig. 11 of Laskar et al., 2004, would be more likely regional and global heat flow and the nature of Martian mantle
than the others). materials, which in turn impact a host of studies concerning crustal
In the absence of more definitive information, we cannot rule evolution, tectonics, and the distribution of liquid water and ice in
out the possibility that the NPLD layering is a function of much the subsurface. Our choice of radar units within the NPLD was
longer or shorter climate cycles (e.g., Tanaka, 2005) or even other, based on the packet–inter-packet sequences, whose cyclic patterns
potentially chaotic processes that are not linearly related to orbital are suggestive of climatological processes, and we anticipate that
cycles (e.g., Perron and Huybers, 2009). The age of the NPLD might important revelations about the climate history of Mars will be
then be substantially younger or older than the 4.2 Ma suggested realized. Mapping of these units provides important new con-
here. To the degree that the insolation models are robust, a dra- straints on the internal structure of the NPLD, but other approaches
matically greater age will require some mechanism to protect the such as unconformity-based mapping (e.g., Holt and Safaeinili,
deposits from entirely sublimating away. A simple means of pres- 2009) will add crucial details to the investigation of these deposits.
ervation is the production of thick lag deposits to preserve the ice While the troughs and surface undulations obscure our view of the
through extended periods of high insolation. However, the fact that subsurface in many areas by introducing unwanted clutter into the
SHARAD obtains strong reflections through to the base of the NPLD radar data, they are an integral part of the NPLD and have provided
stack limits the present-day lithic fraction to no more than several a window into the interior for image-based studies (e.g., Rodriguez
percent (Picardi et al., 2005; Nunes and Phillips, 2006; Phillips et al., 2009). In some areas, the radar exposes structural features
et al., 2008), and the relatively uniform and conformal nature of beneath the troughs and surface undulations (e.g., Fig. 2), and map-
the layering in the inter-trough regions of the NPLD suggests that ping of these features promises to impart key information about
the deposits have not been subjected to widespread erosion over trough migration and development (e.g., Smith et al., 2009). An
long periods of time. In contrast, the south polar layered deposits important next step in all of these investigations is the synthesis
(SPLD)—which have substantially older surface cratering ages of of the radar products with a host of other data sets and modeling
10–100 Ma (Herkenhoff and Plaut, 2000)—exhibit much more var- results, which taken together will provide an impressive set of
iable radar reflectivity, including apparent internal and near-sur- tools for unraveling the geology of Planum Boreum and Mars’ cli-
face angular unconformities suggestive of more widespread mate history.
erosion (Plaut et al., 2007; Seu et al., 2007a; Milkovich et al.,
2009; Phillips et al., 2009). These radar results are consistent with Acknowledgments
the identification of several local and regional unconformities
within the SPLD on the basis of surface imagery and elevation data The Shallow Radar (SHARAD) instrument was provided to NA-
(Kolb and Tanaka, 2006, 2009). SA’s Mars Reconnaissance Orbiter (MRO) mission by the Italian
Space Agency (ASI), and its operations are led by the InfoCom
Department, University of Rome ‘‘La Sapienza.” Thales Alenia Space
6. Conclusions Italia is the prime contractor for SHARAD. The MRO mission is
managed by the Jet Propulsion Laboratory, California Institute of
The mapping work described here represents the first detailed Technology, for the NASA Science Mission Directorate, Washing-
examination of the internal structure of Planum Boreum that ton, DC. Lockheed Martin Space Systems, Denver, CO, is the prime
incorporates radar soundings. While the radar results generally contractor for the orbiter. We are grateful to all the people at these
confirm expectations of the internal structure that have been institutions who are instrumental in the ongoing success of the
developing over decades of studying images and other surface data SHARAD experiment. We thank Prateek Choudhary and Isaac
 N.E. Putzig et al. / Icarus 204 (2009) 443–457 457

Smith for providing simulated radargrams and helpful discussion Milkovich, S.M., Head, J.W., 2005. North polar cap of Mars: Polar layered deposit
characterization and identification of a fundamental climate signal. J. Geophys.
of surface clutter features. We also thank Doug Paul, Kerry Befus,
Res. 110, E01005, 21 pp. doi:10.1029/2004JE002349.
Ed VanWieren, and Ed Chow of SeisWare International Inc. for ac- Milkovich, S.M., Plaut, J.J., Safaeinili, A., Picardi, G., Seu, R., Phillips, R.J., 2009.
cess to and support of the SeisWare interpretation software, which Stratigraphy of Promethei Lingula, south polar layered deposits, Mars, in radar
serves an integral role in our mapping efforts. Reviews by Ken Ta- and imaging data sets. J. Geophys. Res. 114, E03002, 21 pp. doi:10.1029/
2008JE003162.
naka, Brian Davis, and an anonymous reviewer greatly improved Morse, D.L., Waddington, E.D., Steig, E.J., 1998. Ice age storm trajectories inferred
the manuscript. from radar stratigraphy at Taylor Dome Antarctica. Geophys. Res. Lett. 25,
In dedicating this work to the late Dr. Ali Safaeinili, we wish to 3383–3386.
Nunes, D.C., Phillips, R.J., 2006. Radar subsurface mapping of the polar layered
salute our friend and colleague for his outstanding contributions to deposits on Mars. J. Geophys. Res. 111, E06S21, 16 pp. doi:10.1029/
the field of planetary radar science. His technical expertise, insight- 2005JE002609.
ful analysis, and ability to catalyze international teamwork will Perron, J.T., Huybers, P., 2009. Is there an orbital signal in the polar layered deposits
of Mars? Geology 37, 155–158.
continue to serve as an inspiration to all who knew him. Phillips, R.J., and 13 colleagues, 2009. Subsurface structure of the south polar
layered deposits, Mars. Lunar Planet. Sci. XL, 2007 (abstract).
References Phillips, R.J., and 26 colleagues, 2008. Mars north polar deposits: Stratigraphy, age,
and geodynamical response. Science 320, 1182–1185.
Picardi, G., and 33 colleagues, 2005. Radar soundings of the subsurface of Mars.
Byrne, S., Murray, B.C., 2002. North polar stratigraphy and the paleo-erg of Mars. J.
Science 310, 1925–1928.
Geophys. Res. 107, 5044, 12 pp. doi:10.1029/2001JE001615.
Plaut, J.J., and 33 colleagues, 2007. Subsurface radar sounding of the south polar
Clifford, S.M., and 52 colleagues, 2000. The state and future of Mars polar science
layered deposits of Mars. Science 316, 92–95.
and exploration. Icarus 144, 210–242.
Plaut, J.J., Safaeinili, A., Holt, J.W., Phillips, R.J., Head III, J.W., Seu, R., Putzig, N.E.,
Fishbaugh, K.E., Head, J.W., 2005. Origin and characteristics of the Mars north polar
Frigeri, A., 2009. Radar evidence for ice in lobate debris aprons in the mid-
basal unit and implications for polar geologic history. Icarus 174, 444–474.
northern latitudes of Mars. Geophys. Res. Lett. 36, L02203, 4 pp. doi:10.1029/
Grima, C., Kofman, W., Mouginot, J., Phillips, R.J., Hérique, A., Biccari, D., Seu, R.,
2008GL036379.
Cutigni, M., 2009. North polar deposits of Mars: Extreme purity of the water ice.
Rodriguez, J.A.P., Tanaka, K.L., and Berman, D.C., 2009. Depression systems in
Geophys. Res. Lett. 36, L03203, 4 pp. doi:10.1029/2008GL036236.
Western Planum Boreum, Mars: Distributions, orientations, and cross-cutting
Haberle, R.M., Murphy, J.R., Schaeffer, J., 2003. Orbital change experiments with a
relationships. Lunar Planet. Sci. XL, 2371 (abstract).
Mars general circulation model. Icarus 161, 66–89.
Selvans, M.M., Aharonson, O., Plaut, J.J., Safaeinili, A., 2009. Structure of the basal
Herkenhoff, K.E., Byrne, S., Russell, P.S., Fishbaugh, K.E., McEwen, A.S., 2007. Meter-
unit of the north polar Plateau of Mars, from MARSIS. Proc. IEEE Radar Conf.,
scale morphology of the north polar region of Mars. Science 317, 1711–1715.
3206 (abstract).
Herkenhoff, K.E., Plaut, J.J., 2000. Surface ages and resurfacing rates of the polar
Seu, R., and 51 colleagues, 2007a. Accumulation and erosion of Mars’ south polar
layered deposits on Mars. Icarus 144, 243–253.
layered deposits. Science 317, 1715–1718.
Holt, J.W., Peters, M.E., Kempf, S.D., Morse, D.L., Blankenship, D.D., 2006. Echo source
Seu, R., and 11 colleagues, 2007b. SHARAD sounding radar on the Mars
discrimination in single-pass airborne radar sounding data from the Dry
reconnaissance orbiter. J. Geophys. Res. 112, E05S05, 18 pp. doi:10.1029/
Valleys, Antarctica: Implications for orbital sounding of Mars. J. Geophys. Res.
2006JE002745.
111, E06S24, 13 pp. doi:10.1029/2005JE002525.
Smith, D.E., Frey, H.V., Carvin, J.B., Zwally, H.J., Lemoine, F.G., Rowlands, D.D.,
Holt, J.W., and Safaeinili, A., 2009. Northern polar layered deposits, Mars: Structural
Abshire, J.B., Afzal, R.S., Sun, X., Zuber, M.T., 2001. Mars orbiter laser altimeter:
relationships between Gemina Lingula, the main lobe, and Chasma Boreale from
Experiment summary after the first year of global mapping of Mars. J. Geophys.
SHARAD radar stratigraphy. Lunar Planet. Sci. XL, 1721 (abstract).
Res. 106, 23689–23722.
Holt, J.W., Safaeinili, A., Plaut, J.J., Head, J.W., Phillips, R.J., Seu, R., Kempf, S.D.,
Smith, I.B., Holt, J.W., Christian, S.W., and Safaeinili, A., 2009. Evidence for spiral
Choudhary, P., Young, D.A., Putzig, N.E., 2008. Radar sounding evidence for
trough migration and evolution from SHARAD radar observations of
buried glaciers in the southern mid-latitudes of Mars. Science 322, 1235–1238.
stratigraphy within the northern polar layered deposits, Mars. Lunar Planet.
Johari, G.P., 1976. The dielectric properties of H2O and D2O ice Ih at MHz
Sci. XL, 1423 (abstract).
frequencies. J. Chem. Phys. 64, 3998–4005.
Tanaka, K.L., 2005. Geology and insolation-driven climatic history of Amazonian
Kolb, E.J., Tanaka, K.L., 2006. Accumulation and erosion of south polar layered
north polar materials on Mars. Nature 437, 991–994.
deposits in the Promethei Lingula region, Planum Australe. Mars 2, 1–9.
Tanaka, K.L., Rodriguez, J.A.P., Fortezzo, C.M., and Seelos, F., 2009. Obliquity-driven
Kolb, E.J., and Tanaka, K.L., 2009. Bed structures of polar layered deposit outcrops in
periods of extended erosion and deposition in the geologic record of Planum
Australe Mensa: Implications for the formational history of Planum Australe,
Boreum, Mars. Lunar Planet. Sci. XL, 2476 (abstract).
Mars. Lunar Planet. Sci. XL, 2230 (abstract).
Tanaka, K.L., Rodriguez, J.A.P., Skinner, J.A., Bourke, M.C., Fortezzo, C.M., Herkenhoff,
Laskar, J., Correia, A.C.M., Gastineau, M., Joutel, F., Levrard, B., Robutel, P., 2004. Long
K.E., Kolb, E.J., Okubo, C.H., 2008. North polar region of Mars: Advances in
term evolution and chaotic diffusion of the insolation quantities of Mars. Icarus
stratigraphy, structure, and erosional modification. Icarus 196, 318–358.
170, 343–364.
Tanaka, K.L., Scott, D.H., 1987. Geologic map of the polar regions of Mars. USGS
Laskar, J., Levrard, B., Mustard, J.F., 2002. Orbital forcing of the martian polar layered
Miscellaneous Investigations Map I–1802–C.
deposits. Nature 419, 375–377.
Toon, O.B., Pollack, J.B., Ward, W., Burns, J.A., Bilski, K., 1980. The astronomical
Levrard, B., Forget, F., Montmessin, F., Laskar, J., 2007. Recent formation and
theory of climatic change on Mars. Icarus 44, 552–607.
evolution of northern martian polar layered deposits as inferred from a Global
Welch, B.C., Jacobel, R.W., 2003. Analysis of deep-penetrating radar surveys of West
Climate Model. J. Geophys. Res. 112, E06012, 18 pp. doi:10.1029/2006JE002772.
Antarctica, US-ITASE 2001. Geophys. Res. Lett. 30, 1444, 4 pp. doi:10.1029/
Mätzler, C., Wegmüller, U., 1987. Dielectric properties of fresh-water ice at
2003GL017210.
microwave frequencies. J. Phys. D: Appl. Phys. 20, 1623–1630.