SHARAD soundings and surface roughness at past, present, and proposed landing sites on Mars:
Re fl ections at Phoenix may be attributable to deep ground ice
Nathaniel E. Putzig1, Roger J. Phillips2,3, Bruce A. Campbell4, Michael T. Mellon1, John W. Holt5, and T. Charles Brothers5
1Department of Space Studies, Space Science and Engineering Division, Southwest Research Institute, Boulder, Colorado, USA,2Planetary Science Directorate, Space Science and Engineering Division, Southwest Research Institute, Boulder, Colorado, USA,3Department of Earth and Planetary Sciences, Washington University in St. Louis, St. Louis, Missouri, USA,
4National Air and Space Museum, Smithsonian Institution, Washington, District of Columbia, USA,5Institute for Geophysics and Department of Geological Sciences, Jackson School of Geosciences, University of Texas at Austin, Austin, Texas, USA
Abstract
We use the Shallow Radar (SHARAD) on the Mars Reconnaissance Orbiter to search for subsurface interfaces and characterize surface roughness at the landing sites of Viking Landers 1 and 2, Mars Pathfinder, the Mars Exploration Rovers Spirit and Opportunity, the Phoenix Mars lander, the Mars Science Laboratory Curiosity rover, and three other sites proposed for Curiosity. Only at the Phoenix site do wefind clear evidence of subsurface radar returns, mapping out an interface that may be the base of ground ice at depths of ~15–66 m across 2900 km2in the depression where the lander resides. At the Opportunity, Spirit, and candidate Curiosity sites, images and altimetry show layered materials tens to hundreds of meters thick extending tens to hundreds of kilometers laterally. These scales are well within SHARAD’s resolution limits, so the lack of detections is attributable either to low density contrasts in layers of similar composition and internal structure or to signal attenuation within the shallowest layers. At each site, we use the radar return power to estimate surface roughness at scales of 10–100 m, a measure that is important for assessing physical properties, landing safety, and site trafficability. The strongest returns are found at the Opportunity site, indicating that Meridiani Planum is exceptionally smooth. Returns of moderate strength at the Spirit site reflect roughness more typical of Mars. Gale crater, Curiosity’s ultimate destination, is the smoothest of the four proposed sites we examined, with Holden crater, Eberswalde crater, and Mawrth Vallis exhibiting progressively greater roughness.1. Introduction
The selection of landing sites on Mars has evolved into a highly collaborative process involving several years of effort by a broad range of scientists and engineers [Golombek et al., 2012]. Subsequent to landings, the evaluation of site characteristics with orbital observations continues throughout the landed missions and often extends many years beyond them [Golombek et al., 2008]. Since orbital data are typically acquired with spatial resolution that is coarser by an order of magnitude or more than that of ground observations, accurate interpretation of orbital data relies heavily on the observations made by landed spacecraft. At the same time, choosing new landing sites is driven largely by the interpretation of the orbital data sets, taking into consideration a variety of sometimes conflicting constraints such as landing safety, site trafficability, and geologic interest. Historically, site evaluations have relied predominantly on orbital observations of the surface such as those from visible imagers, spectrometers, and a laser altimeter to assess parameters critical to landing-site selection and operations, including surface slopes, atmospheric conditions, mineralogy, grain sizes, rock abundance, stratigraphy, and geologic structures. More recently, other remote- sensing instrumentation such as gamma ray and neutron spectrometers and radars have been put into orbit at Mars and are acquiring data that include new information about surface and subsurface properties. Integration of these data more fully into site evaluations will allow tighter constraints on stratigraphy and geologic structures. In addition, radar observations can be used to evaluate surface roughness on scales intermediate to those available from visible imagery and laser altimetry.
Journal of Geophysical Research: Planets
RESEARCH ARTICLE
10.1002/2014JE004646
Key Points:
•Possible base of ground ice is mapped with Shallow Radar at Phoenix Mars site
•Roughness at scales of 10 to 100 m is assessed at 10 landing sites on Mars
•High-resolution surface-elevation mod- els are critical for radar interpretation
Correspondence to:
N. E. Putzig, [email protected]
Citation:
Putzig, N. E., R. J. Phillips, B. A. Campbell, M. T. Mellon, J. W. Holt, and T. C. Brothers (2014), SHARAD soundings and surface roughness at past, present, and proposed landing sites on Mars: Reflections at Phoenix may be attributable to deep ground ice,J. Geophys. Res. Planets,119, 1936–1949, doi:10.1002/
2014JE004646.
Received 21 APR 2014 Accepted 26 JUL 2014
Accepted article online 1 AUG 2014 Published online 21 AUG 2014
In this work, we focus on results from the Shallow Radar (SHARAD) instrument onboard the Mars Reconnaissance Orbiter (MRO), which has obtained observations over each of the past landing sites and several other sites proposed for the Mars Science Laboratory Curiosity rover. SHARAD has proven highly capable of detecting subsurface interfaces, most dramatically within the icy layered deposits in the polar regions [Phillips et al., 2008;Seu et al., 2007a] but also within volcanic deposits [Campbell et al., 2008;Carter et al., 2009b] and at the base of glacial ices [Holt et al., 2008;Plaut et al., 2009] in the midlatitudes. At Meridiani Planum, the landing site of the Mars Exploration Rover Opportunity, images of the surface both from orbit and from the rover show clear evidence for layering on the order of meters to tens of meters in thickness that extends laterally over several hundred kilometers. Layering at similar scales is also evident at the landing site for Curiosity (Gale crater) and three other sites that were proposed for that mission (Eberswalde crater, Holden crater, and Mawrth Vallis). At the Phoenix Mars landing site in the northern plains, images of the surface do not show direct evidence of layering, but they do show polygonal terrain consistent with the widespread presence of near-surface ground ice extending to depths of at least 10 m [Mellon et al., 2008b].
This interpretation is strongly supported by the detection of excess hydrogen throughout the high latitudes by the Mars Odyssey Gamma Ray Spectrometer (GRS) suite and modeling of thermal observations [Mellon et al., 2008a]. With sufficient dielectric contrasts at layer interfaces, these scales should allow detections by SHARAD, so each of these sites was targeted with high priority by the SHARAD Team and extensive coverage is available. Since there is no clear evidence for layering on these scales at the landing sites of the Mars Exploration Rover Spirit (in Gusev crater), Mars Pathfinder, and Viking Landers 1 and 2, targeting these other sites was not deemed a high priority by the SHARAD Team and only limited coverage is available.
In 2008, the SHARAD Teamfirst made note of low-power apparent subsurface returns at the Phoenix site [Safaeinili et al., 2008]. Subsequently, similar, slightly shallower returns were identified across the northern plains [Putzig et al., 2009a] and the southern highlands [Putzig et al., 2009b], suggesting widespread detection of the base of ground ice. However, later analysis showed that most of the widespread shallow returns occur at delay times corresponding to those expected for sidelobes of the surface return [Putzig et al., 2012]. While the returns do have about twice the expected power of the sidelobes, the close correspondence in delay time casts doubt on a subsurface interpretation. At the Phoenix site, the delay times are typically larger and the power of returns is greater, so a subsurface source for the late returns is much better supported, as we discuss in subsequent sections.
Surface roughness is important both as a factor in evaluating spacecraft landing safety and the ability for rovers to move around on the surface and as an indicator of morphology to constrain geologic properties.
Measures of roughness depend on the horizontal length scale at which topography is sampled, with scales of order 1–10 m of primary concern to landing and roving safety and scales of order 10–1000 m of primary concern to geomorphology. Image data such as that from the MRO High Resolution Imaging Science Experiment (HiRISE) provide a means of evaluating roughness at thefiner scales, either by interpreting features in individual images (e.g., using shadows to estimate heights of rocks or other relief) or by using stereo pairs to obtain surface-elevation models. For coarser scales, individual observations by the Mars Orbiter Laser Altimeter (MOLA), which were typically spaced at ~300 m, have been used collectively to evaluate roughness at scales of 600 m to 20 km [Kreslavsky and Head, 2000, 2002]. Because the power of radar returns from a surface depends on the distribution of slope facets at scales on the order of the signal wavelength, SHARAD data can be used to estimate roughness at intermediate scales of 10–100 m, complementary to the estimates provided by images and MOLA data. In this work, we extract roughness for each landing site from a global data set of a SHARAD-derived roughness parameter [Campbell et al., 2013], mapping it at two scales to assess the regional and local distributions of roughness.
2. Radar Observations
The SHARAD sounding radar emits an 85μs chirped pulse with a 10 MHz bandwidth centered at 20 MHz, yielding range resolutions of 15 m in free space, 8.5 m in water ice with real part of dielectric permittivity (“real permittivity”hereinafter) of 3.15, and 5.3 m in basalt with real permittivity of 8. Correlation of returned signals with the transmitted pulse produces a band-limited waveform with thefirst and second sidelobes expected at delays of 0.24μs and 0.42μs and powers of20 dB and34 dB, respectively, relative to peak returns. In ground processing, a Hann weighting function is applied to the data, providing some suppression of the sidelobe power without greatly reducing the range resolution. The altitude of the MRO orbit varies
between about 250 km and 320 km, allowing SHARAD a lateral resolution at the surface of ~3–6 km (1–2 Fresnel zones), which can be reduced in the azimuth (along-track) direction to 0.3–1.0 km with synthetic-aperture focusing. SeeSeu et al. [2007b] for a detailed discussion of the SHARAD experiment.
For this work, we applied the radar processing technique ofCampbell et al. [2011] to the SHARAD data prior to searching for subsurface returns. Roughness data for each site were taken from the global data set ofCampbell et al.
[2013]. We refer to each collection of contiguous SHARAD returns along an orbit segment as an“observation”, presenting them as radargrams, which are image-format displays of returned power with the horizontal axis representing distance along track and the vertical axis representing the signal delay time. Along-track resolution is matched to that of a MOLA elevation map [Smith et al., 2001] at 128 pixels per degree (ppd) (~460 m per pixel), and the power is sampled at 37.5 ns in delay time. From top to bottom, a radargram usually shows (1) lower-power background noise associated with transmission through the atmosphere, (2) higher-power returns from points on the surface closest to the spacecraft (often from the nadir points directly below the spacecraft, but sometimes from high-standing off-nadir structures), and (3) later returns that may be from either subsurface interfaces or off-nadir structures that are farther away from the spacecraft (termed“surface clutter”). Because of the latter ambiguity, common practice in evaluating subsurface radar data is to generate corresponding clutter simulations using surface-elevation models of the observed region and the spacecraft trajectory. We used an incoherent, geometrical optics model that takes into account the effects of along-track focusing [Holt et al., 2006; P.
Choudhary et al., Surface clutter and echo location analysis for the interpretation of SHARAD data from Mars, IEEE Geoscience and Remote Sensing Letters, manuscript in revision, 2014] and gridded MOLA topography [Smith et al., 2001] or Mars Express High Resolution Stereo Camera (HRSC) DT4 products [Neukum et al., 2004]. This method maximizes the possible surface clutter, rather than attempting to precisely simulate the radar data. To the extent that actual surface structures are well represented in the models and simulations, late returns appearing in the data but not in the simulations may be interpreted as subsurface returns. Where returns are identified as subsurface returns, delay times can be converted to depths on a sample-by-sample basis as
Δdi¼½Δt cεi½ (1)
whereΔdiis the depth interval for thei-th sample in delay time,Δtis the delay-time sampling interval,cis the speed of light in free space, andεiis the real permittivity of the traversed medium, with the factor of 1/2 accounting for two-way travel (i.e., radar transmit and receive paths).
The SHARAD signal is subject to distortion and attenuation by the ionosphere of Mars, effects largely confined to dayside observations that can be corrected in processing [Campbell et al., 2011, 2014]. In addition to providing additional coverage, the inclusion of dayside observations is important to our analysis because they form a data grid with the crossing nightside observations, thereby allowing confident mapping of specific subsurface returns. During SHARAD operations, it was established that an increase in the signal-to-noise ratio up to 6 dB can be achieved by acquiring observations with the spacecraft in a rolled configuration and (on the nightside) positioning the solar array in a specific orientation [Campbell et al., 2014]. We included these
Table 1. SHARAD Observation Counts and Peak Roughness
Landing Site
Area of Interest (AOI) for Subsurface Returns
Observation Count
Peak Roughness Subsurface Return AOI
Roughness AOIs
3 × 3° 0.7 × 0.7° 3 × 3° 0.7 × 0.7°
Viking 1 21.2–24.2°N, 310.3–313.3°E 16 16 3 2.75 2.75
Viking 2 46.8–49.8°N, 132.5–135.5°E 27 15 2 3.50 3.75
Pathfinder 17.6–20.6°N, 325.3–328.3°E 12 6 1 3.25 3.25
Spirit 13.2–15.7°S, 174.3–176.8°E 10 13 5 3.50 3.00
Opportunity 4.0°S–9.0°N,9.5–9.0°E 68 17 3 2.50 2.50
Phoenix 67.0–69.5°N, 229.5–238.5°E 125 103 32 3.25 3.25
Curiosity 4.0–6.6°S, 136.7–138.9°E 29 13 4 3.25 4.25
Holden
23.5–27.2°S, 324.5–327.2°E 28 38 9 3.50 3.50
Eberswalde 68 26 3.75 3.75
Mawrth 22.2–25.8°N, 339.6–342.2°E 52 69 21 4.00 4.00
AOI for Roughness Observation Count Peak Roughness
Midlatitudes 70°S–70°N, 0–360°E 8000 3.25
rolled observations in our search for subsurface returns, but they are omitted in the derivation of roughness to avoid complicating comparisons of changes in relative surface reflectivity [Campbell et al., 2013].
SHARAD operates in a targeted mode, with large targets established for the polar regions (spanning 20° of latitude on either side of the pole, ~2400 km) and smaller targets at other sites of interest, including some of the past, present, and proposed landing sites. Early in the MRO mission (2007), targets were established surrounding the landing sites for Opportunity, Phoenix, and the fourfinalist sites for Curiosity. Extensive coverage by SHARAD has been acquired at each of these sites. While targets were not created specifically for the landing sites of Viking Landers 1 and 2, Pathfinder, and Spirit, regional targets do cover these sites and some observations have been acquired near each of them. For the purpose of searching for subsurface returns, we used a variety of criteria to establish areas of interest (AOIs) at each site. For the sites of Viking Landers 1 and 2 and Pathfinder, we used the same 3×3° areas for both the subsurface- return and regional-roughness investigations. At Spirit, we used the bounds of Gusev crater to establish the AOI. At Opportunity, we chose an AOI bounding the extents of layered materials that had previously been mapped in the surrounding region [Hynek, 2004]. At the Phoenix site, we adopted as AOI the prelanding study area known as Region D, box 1 [Arvidson et al., 2008]. For the actual and proposed Curiosity sites, we used the prelanding study areas covering the proposed landing ellipses, roving areas, and their surroundings [Golombek et al., 2012]. In assessing subsurface returns, we treated the Holden and Eberswalde sites together due to their adjacent locations. Table 1 lists the AOIs for our subsurface-return investigations at each site together with the number of SHARAD observations examined.
Our roughness analysis employs the results of Campbell et al. [2013], who established a roughness parameterρas
ρ¼Pi=P0 (2)
wherePiis the radar power integrated over a range of incidence angles andP0is the peak radar power (i.e., that at zero incidence angle). In this formulation,ρis independent of reflectivity and modulated only by surface roughness on the order of the radar wavelength (15 m for SHARAD). In calculating the roughness parameter,Campbell et al.
Figure 1
[2013] restricted incidence angles to 0–1.5°, and the roughness measure includes any subsurface returns to depths of ~35–60 m. For this work, we extracted roughness parameter values from the database of 8000 SHARAD observations on MRO orbits 2000 through 25,999. We then selected results at each landing site, mapping them at 50 ppd to assess regional (3 × 3°) roughness and at 200 ppd to assess local (0.7 × 0.7°, 40 × 40 km) roughness. The numbers of SHARAD observations within the roughness AOIs for each site at these two scales are listed in Table 1.
3. Results
In this section, we describe our findings at the landing sites for the identification of SHARAD subsurface returns and for the analysis of SHARAD-derived roughness. We found clear evidence for subsurface returns only at the Phoenix site, but we include a discussion of the nondetections at the other sites, emphasizing the results at Gale crater where early work suggestive of subsurface returns was later discounted using clutter simulations created from a high- resolution surface-elevation model.
3.1. Subsurface Detections at the Phoenix Mars Landing Site We examined SHARAD observations taken on 125 orbits through July 2011 that crossed the region 67.0–69.5°N, 229–238°E surrounding the Phoenix site (corresponding to Region D, box 1 ofArvidson et al. [2008]). Figure 1 presents a subset of this region where we have identified and mapped subsurface returns in 81 radargrams using SeisWare™ seismic-data analysis software. The ground-track map (Figure 1a) shows the grid of nightside and dayside coverage that we used in our search for subsurface returns. We highlight tracks for four rolled observations in red and for two observations nearest the landing site in yellow. The base map, consisting of a daytime Mars
Figure 1.Maps of the area around the Phoenix Mars landing site in the northern plains of Mars. Base map in each case is a THEMIS daytime IR mosaic [Christensen et al., 2013]. (a) SHARAD ground tracks through July 2011, with nightside and dayside acquisition trajectories indicated by arrows. Red ground tracks are for observations acquired with the spacecraft in a rolled configuration to improve the signal-to-noise ratio. Solid lines indicate ground tracks of labeled observations nearest the lander that are shown in Figures 2–4. (b) Delay time to subsurface returns mapped on 81 SHARAD observations of the Green Valley where the Phoenix lander resides, interpolated to infill between ground tracks. While the modal value is near that expected for the second sidelobe of the surface return, delay times range broadly from 0.18 to 0.78μs. The material—likely dominated by ground ice—between the surface and the subsurface return over the mapped area of 2912 km2occupies 64–99 km3for real permittivities of 3.15–8.0. (c) Power in decibels of the subsurface return relative to that of the surface return on 81 SHARAD observations. All power values are well above that expected for the second sidelobe of the surface return, and the modal power of9 dB is well above that expected for thefirst sidelobe.
Figure 2.Portion of SHARAD nightside rolled observation 16378–02, inverted to extend from south to north across the Phoenix landing site and the Green Valley. (a) Uninterpreted radargram. (b) Interpreted radargram, showing delay times for the MOLA modeled surface in blue and for the subsurface return in yellow, with the latter delayed ~0.4μs from the surface return.“×”symbols show corresponding delay times from crossing dayside observations. Interpretation of subsurface returns is discontinued where it is obfuscated by off-nadir surface returns. Inset shows map view of ground track. (c) Clutter simulation produced from a swath of the MOLA modeled surface along the ground track of the observation. Returns arriving late or early relative to the nadir surface (yellow line) are off-nadir surface returns that may also appear in the SHARAD radargram (white arrows). No late returns occur where the subsurface return is identified in the SHARAD radargram (yellow arrows).
Odyssey Thermal Emission Imaging Spectrometer (THEMIS) mosaic, shows the relatively smooth,flat-lying depression (informally known as Green Valley) that extends north and west of the landing site as well as nearby Heimdal crater. The subsurface returns mapped are relatively low power, even for nightside observations acquired with the spacecraft in a rolled orientation that provides the highest possible signal-to-noise ratio (e.g., Figure 2a). As discussed in section 2, having a grid of data is critical to correlating subsurface returns between adjacent tracks (see yellow × symbols in Figure 2b), and here it is also important to confident interpretation of the low- power signals within each track. In addition, we use clutter simulations to identify returns from off-nadir surface features (white arrows in Figures 2b and 2c) that might otherwise be mistaken for subsurface returns and to ensure that returns identified as subsurface returns are not in fact off-nadir features present in the modeled surface (cf. yellow line in Figure 2b and yellow arrows in Figure 2c).
Given the relatively sparse coverage of rolled observations (red ground tracks in Figure 1), mapping of the laterally extensive subsurface returns found in the vicinity of the Phoenix site requires the inclusion of
nonrolled tracks. Nightside nonrolled tracks (e.g., Figure 3) exhibit lower signal-to-noise ratio than rolled tracks (cf. Figures 2a and 3a), but the subsurface returns have sufficient power to identify them. Dayside nonrolled tracks (e.g., Figure 4) provide still lower signal-to-noise ratio, but as already mentioned, they provide a critical link between adjacent nightside tracks.
We found that the subsurface returns shown in Figures 2–4 extend contiguously across 81 of the 125 radargrams included in our Phoenix study area. We discontinued our tracing of the subsurface returns either where the power fell to the noise level or the feature is interrupted or comingled with off-nadir surface returns within each radargram (e.g., Figure 2b). To the east and west of the mapped area (see Figure 1b), a shallow, late return persists in the radargrams but its delay time approaches that expected for Figure 3.Portion of SHARAD nightside nonrolled observation 5724–01,
inverted to extend from south to north across the Phoenix landing site and the Green Valley. (a) Uninterpreted radargram. (b) Interpreted radargram, showing delay times for the MOLA modeled surface in blue and for the subsurface return in yellow, with the latter delayed ~0.4μs from the surface return.“×”symbols show corresponding delay times from crossing dayside observations. Inset shows map view of ground track. Nonrolled observations typically have lower signal-to-noise ratio than rolled ones, but power of subsurface returns remains high enough to distinguish from sidelobes.
Figure 4.Portion of SHARAD dayside nonrolled observation 12903–01 extend- ing from south to north across the Phoenix landing site and the Green Valley.
(a) Uninterpreted radargram. (b) Interpreted radargram, showing delay times for the MOLA modeled surface in blue and for the subsurface return in yellow, with the latter delayed ~0.4μs from the surface return.“×”symbols show corresponding delay times from crossing nightside observations. Inset shows map view of ground track. Dayside nonrolled observations typically have lower signal-to-noise ratio than either nightside or rolled ones, but power of subsur- face returns remains high enough to distinguish from sidelobes, and the crossing trajectories are critical for tracking features between observations.
sidelobes of the surface return, so we discontinued mapping of the subsurface return in those areas. In addition to tracing the subsurface returns, we determined the delay to the surface along each ground track using MRO ephemeris data and the 128 ppd MOLA elevation map, displayed as a blue line in Figures 2b, 3b, and 4b. Once the surface and subsurface-return boundaries were delineated on the radargrams, we used the analysis software to calculate the delay time from the surface to the subsurface returns, which we binned along track and interpolated between tracks in an regular grid of
500 m × 500 m bins to obtain the map in Figure 1b. Across the mapped area, the delay times vary between 0.18μs and 0.78μs, with a modal value of 0.41 μs. Delay time of the subsurface returns can be translated to depth via equation (1) using an appropriate value for the real permittivity of the subsurface materials. In some studies, it has been possible to constrain the dielectric using structural considerations [e.g., Campbell et al., 2008;Carter et al., 2009a;Phillips et al., 2011]. That method cannot be applied to SHARAD data at the Phoenix site, since there are no indications of structures exposing subsurface layering that may be present. However, the presence of a substantial quantity of ground ice extending to depths up to a few tens of meters is well supported by
observations from the Phoenix lander, GRS data, and thermal modeling [Mellon et al., 2009], and it is therefore reasonable to assume a real
permittivity consistent with a mixture of water ice and basaltic regolith.
Considering end-member cases, if the materials at depth are entirely water ice (with real permittivity of 3.15), then the depth to the detected interface ranges from 15 to 66 m, with a modal depth of 34 m. If the materials are purely basaltic (with a real permittivity of 8), then the depth range is 9 to 41 m with a modal value of 22 m.
Integrating the depths (using the end-member modal values) over the entire mapped area of 2912 km2yields a volume of 64–99 km3for the interval from the surface to the mapped subsurface returns.
As discussed in section 1, somewhat shallower and lower-power returns across other areas of the northern plains were found to coincide closely with the delay times of thefirst and second sidelobes of the surface return (0.24μs and 0.42μs), with about twice the expected power. For SHARAD returns at the Phoenix site, Figure 5.Portion of SHARAD nightside rolled observation 21762–01, inverted
to extend from south to north across Gale crater and the Curiosity landing site. (a) Uninterpreted radargram. (b) Interpreted radargram, showing delay times for the MOLA modeled surface in blue and for the putative subsurface returns in yellow, with the latter delayed ~2–3μs from the surface return. Inset shows ground track with putative subsurface returns identified along adja- cent tracks over a MOLA color shaded relief base map. Curiosity landing ellipse is shown with a black line. (c) Clutter simulation produced from a swath of the MOLA modeled surface along the ground track of the observation. Returns arriving late or early relative to the nadir surface (yellow line) are off-nadir surface returns, many of which also appear in the SHARAD radargram (white arrows). No late returns occur where the subsurface returns are identi- fied in the SHARAD radargram (yellow arrows). (d) Clutter simulation produced from a swath of the HRSC modeled surface along the ground track of the observation. Many more off-nadir surface returns are identifiable relative to that for the MOLA surface, including ones corresponding to the putative subsurface returns identified in the SHARAD radargram (yellow arrows).
the delay times to the subsurface return are more variable and predominantly greater than that of thefirst sidelobe, but the modal delay time is very near that of the second sidelobe, so one must include consideration of the power, which we do for the collection of radargrams without interpolation (Figure 1c). The subsurface-return power has a modal value of9 dB relative to that of the surface return, withallvalues well above that expected for the second sidelobe (34 dB) and most values above that expected for thefirst sidelobe (20 dB). In addition, we conducted processing tests using different weighting functions (uniform, Hann, Hamming, and Chebychev) to assess their relative effects, and in no case was the power of the putative detections substantially suppressed. These results lend greater confidence to an interpretation of the returns as having a subsurface source.
3.2. Nondetections of Subsurface Returns at Gale Crater and Other Landing Sites
Our search for subsurface returns at landing sites other than that of the Phoenix Mars lander has so far proven fruitless. In the other cases where the region surrounding the landing site is relativelyflat lying (the Viking Lander 1 site in Chryse Planitia, the Viking Lander 2 site in Utopia Planitia, the Mars Pathfinder site in Ares Vallis, the Spirit site in Gusev crater, and the Opportunity site in Meridiani Planum), the available SHARAD observations show a distinct surface return with no apparent subsurface returns within the study areas (see section 2 and Table 1). The lack of detections in these regions may be due in part to the sparse coverage at each of these sites except for Meridiani Planum, which was targeted extensively by SHARAD. There, we find some of the highest-power radar returns from the surface of Mars, but the subsequent records show very low power and no strong indicators of subsurface returns throughout the region (see, however,Watters et al. [2014]). In the cases where the region surrounding the landing site has substantial topographic relief (the actual Curiosity site in Gale crater and the proposed Curiosity sites in Holden crater, Eberswalde crater, and Mawrth Vallis), returns subsequent to the nadir surface return are common, but wefind that they are largely explained as surface returns from off-nadir topographic features. Typically, the off-nadir topography is well characterized by the MOLA modeled surface, but this is not the case for Gale crater. In SHARAD observations of Gale crater (e.g., Figure 5a), we identified a laterally extensive feature at delays of
~1–3μs below the northern craterfloor (Figure 5b). Clutter simulations produced from the MOLA modeled surface show extensive off-nadir returns in the area, but none of them correspond to that of the SHARAD feature we identified (Figure 5c), and thus, our preliminary analysis supported the interpretation of the feature as a subsurface return [Putzig et al., 2012]. However, we implemented clutter simulations using surface-elevation models produced from HRSC images [Neukum et al., 2004], and simulations produced from a Gale crater HRSC surface-elevation model at 75 m per pixel reveal that the feature we identified can in fact be attributed to an off-nadir return (Figure 5d). This example demonstrates that high-resolution topography is essential to distinguishing possible subsurface returns from surface clutter, especially in limited areas with rough terrain and few observations.
Figure 6.Map of roughness parameter [Campbell et al., 2013] derived from SHARAD nonrolled observations on MRO orbits 2000–25,999 between 70°S and 70°N and binned at 10 ppd in a Mollweide projection centered on 0° longitude. White boxes indicate landing sites for Viking 1 (V1), Viking 2 (V2), Pathfinder (Pa), Opportunity (O), Spirit (S), Phoenix (Ph), Curiosity (C), and the proposed MSL sites in Holden crater (H), Eberswalde crater (E), and Mawrth Vallis (M).
3.3. SHARAD-Derived Roughness at All Landing Sites
In Figure 6, we present a map of the roughness parameter [Campbell et al., 2013] between 70°S and 70°N, produced from 8000 SHARAD nonrolled observations on MRO orbits 2000–25,999 and binned at 10 ppd.
Locations of the landing sites analyzed in this work are indicated with white boxes. An example of how we specify areas for regional and local analyses is presented in Figure 7a, using the Opportunity site in Meridiani Planum. Results from SHARAD are infilled (Figure 7c) and compared to those obtained previously from MOLA at the 0.6 km scale (Figure 7d) [Kreslavsky and Head, 2002]. While the specific values from the two roughness techniques are not directly comparable, the patterns of smooth and rough terrains are largely consistent. In general, the region around the Opportunity landing site is shown to be extremely smooth on the scales of ~10 m and ~1 km provided by SHARAD and MOLA, respectively.
To compare radar roughness of the landing sites included in this study, we created histograms of the roughness parameter in both roughness AOIs at each site and for the entire midlatitude map (Figure 8).
Peak (modal) values for each site can be discerned from the histograms and are also listed in Table 1. At both regional and local scales, the Opportunity site is smoothest with a peak value of 2.5 and the Mawrth site is roughest with a peak value of 4.0. Of the actual past landing sites, only those of Viking 2 and Spirit exhibit regional peak roughness greater than that of the entire midlatitude band. The sites show diverse distributions of roughness, with Opportunity and Viking 1 having the narrowest distributions and Mawrth, Eberswalde, and Curiosity having the broadest distributions. For the most part, local and regional results are consistent with each other, with the notable exception of the Curiosity site. Here, the local roughness results are limited to 4 SHARAD tracks that disproportionately sample the Gale crater rim and Figure 7.Maps of Meridiani Planum and surrounding areas. X symbols mark the Opportunity landing site in each map.
(a) SHARAD roughness parameter at 10 ppd, with white boxes showing the regional (3 × 3°) and local (0.7 × 0.7°) areas of interest used for histograms in Figure 8. (b) Shaded relief map of MOLA elevation [Smith et al., 2001]. (c) Same as Figure 7a with nearest-neighbor mean infilling applied. (d) MOLA 0.6 km scale roughness fromKreslavsky and Head[2002], where values in map bins represent typical slopes in radians (M. Kreslavsky, personal communication, 2013).
Aeolis Mons (Mount Sharp) and are thus biased toward higher values. Future efforts to include additional coverage in roughness parameter can be expected to reduce the local peak value for the Curiosity site.
4. Discussion
The Phoenix Mars mission was motivated in large part by the inference from GRS results of abundant water ice in the near-subsurface all across the northern plains of Mars. Selection of possible landing sites began by choosing a band of latitudes where the ice was believed to be shallow enough to be accessible by the lander’s scoop but deep enough to retain a surface layer of dry soil. Initially, the Phoenix team considered potential landing sites over a broad range of longitudes. When images of the sites began to arrive from HiRISE, the discovery of widespread high rock abundance raised attendant concerns about landing safety for the spacecraft.
Eventually, the landing site was chosen based on that region having relatively low rock
abundance in HiRISE imagery and meeting the scientific objectives of the mission. Early SHARAD observations were limited and did not provide discriminating information toward landing-site selection. The subsurface detections presented in this work are unique across the entire zone of latitudes that were being considered, and in retrospect, the results may well have provided some scientific motivation toward choosing this same location.
The simplest explanation for the SHARAD subsurface returns mapped at the Phoenix site is that they represent the base of ground ice in this area, but if that is the case, then it is more difficult to explain why the base of ground ice is not detected more broadly in the northern plains. An ongoing study does suggest the presence of a widespread deposit of water ice in Arcadia Planitia [Bramson et al., 2014], but the SHARAD detections there are deeper and have higher power that is suggestive of relatively pure ice, perhaps deposited as snowfall. More broadly, prior thermal modeling and observational data showing polygonal patterns at scales of ~5–20 m have suggested that the base of ground ice at the Phoenix site and across the northern plains extends to depths of at least ~10 m [Mellon et al., 2008b]. A broad range of mechanisms for emplacing ground ice on Mars have been proposed, from freezing of an ocean or ancientflood waters [Carr, 1996] to snow accumulated at times of higher obliquity [Head et al., 2005]—perhaps coeval with dust and thereby forming the observed high-latitude mantle deposits [Mustard et al., 2001]–to vapor diffusion into an initially dry regolith. At the Phoenix site, the lower power of the SHARAD detections suggests that the ice is largely interstitial to the regolith, favoring the vapor-diffusion mechanism. In that case, ground ice may be emplaced by some combination of atmospheric water vapor diffusing downward along seasonal thermal waves due to insolation and ground-water vapor diffusing upward along the Martian geothermal gradient [Clifford, 1993;Mellon and Jakosky, 1993, 1995]. The level where the vapor-density gradient from each source converges is the maximum depth to which diffusion of atmospheric water vapor alone may emplace ground ice. Below this point, any additional ground ice would necessarily be due to underplating from upward diffusion of ground water. Using the thermal model ofMellon et al. [2004] and making reasonable assumptions of physical properties, wefind a mean-annualflux-convergence depth of
~15–20 m at the Phoenix latitude (upper curve in Figure 9a), somewhat shallower than the SHARAD
1000
100
10
1
2 3 4 5 6 7 8 9
roughness parameter
count in bins of 0.25count in bins of 0.25
(b) (a)
1000
100
10
1
Figure 8.Histograms of the SHARAD roughness parameter in bins ofρ= 0.25. Black curves show midlatitude values mapped at 10 ppd (Figure 6) and scaled by a factor of 50.
Color curves show values mapped at (a) 50 ppd in 3 × 3°
regions and (b) 200 ppd in 0.7 × 0.7° local areas surrounding each site. Legend is sorted in ascending order of peak rough- ness in Figure 8a. See Figure 6 for site locations.
detections. Greater depths are achieved for lower geothermal gradients, an assumption that is supported by the lack of basal deflection observed beneath Planum Boreum [Phillips et al., 2008]. One might also argue that the maximum rather than the meanflux
convergence could control the base of ground ice (lower curve in Figure 9a), but then the lack of regionally broader SHARAD
detections remains unexplained. One possibility is that as ground ice has been emplaced over time by vapor diffusion and freezing within the upper few m throughout the northern plains, it has reached greater depths here due to a higher rate of aggradation of surface sediments in the low-lying Green Valley. Another possibility is that certain features or conditions unique to the Phoenix region allowed vapor diffusion and deposition of ground ice at greater depths than elsewhere.
One of those features might be a localized source of ground water at depth. In either case, one might expect tofind a SHARAD basal-ice detection only in areas with such a deeper base of ground ice, since sidelobes of the surface return would obfuscate returns from shallower (~10–15 m) depths (Figure 9b). A third possibility is that the ground ice extends to similar depths more broadly, but its base is only detected here because of reduced loss of the radar signal resulting from some combination of the uniquely low rock abundance in this area and a localized lack of radar-attenuating materials at depth.
We cannot rule out the possibility that ground ice is shallower at the Phoenix site as well as elsewhere and that the base of ground ice is not the source of the radar detection. Rather, it may be due to some other geologic contact with a high dielectric contrast, such as sediments over a buried lavaflow. In seeking alternative explanations, we found a geographic correspondence between the SHARAD subsurface returns at the Phoenix site and a geomorphic unit associated with the outer ejecta blanket of Heimdal crater [Heet et al., 2009] (Figure 10). Heimdal has been classified as a double-layered ejecta crater, for which the outer eject layers typically range from a few meters to tens of meters in thickness [Boyce and Mouginis-Mark, 2006].
However, the idea that the subsurface returns could be the base of an ejecta blanket that extends to the Figure 9.(a) Modeled depths of the mean-annual (solid curve) and maxi-
mum-annual (dashed curve) convergence point between downward diffusion of atmospheric water vapor and upward diffusion of ground- water vapor at the latitude of the Phoenix lander (68.2°N). Theflux-conver- gence point controls the maximum depth to which atmospheric vapor may deposit ground ice. Although earlier studies have suggested a Martian geothermal gradient of 30 mW m2K1(bold vertical line), SHARAD results [Phillips et al., 2008] suggest a much lower value (e.g., 10 mW m2K1, dashed vertical line). The thermal model [Mellon et al., 2004] assumes a surface thermal inertia of 200 tiu and an albedo of 0.2. (b) Cartoon cross section depicting different scenarios for the SHARAD detection at the Phoenix land- ing site. The detected interface (dark grey line at depth of ~30 m) may correspond to a relatively deep base of ground ice or to some other geologic contact with a dielectric contrast (e.g., sediments over bedrock such as a lava flow). Overlain curve depicts the theoretical power response of the surface alone to the SHARAD signal and indicates that later returns from a possible shallower base of ground ice as might occur elsewhere in the northern plains (light grey line at ~10 m depth) may be obfuscated by sidelobes (S1, S2) of the surface return (S0) while returns from the deeper interface could arrive sufficiently later in delay time to be distinguishable.
surface can be dismissed solely on geometric considerations, as the size of Heimdal is insufficient by well over an order of magnitude to account for the volume of material overlying the mapped subsurface returns.
Another consideration is that the geomorphic unit is characterized by low rock abundance inferred to be a result of the impact process [Heet et al., 2009], which could allow for reduced scattering of the radar signal and thereby explain the geographically restricted detection of an interface at depth that may actually extend well beyond the bounds of the unit and the detection area. Given all of these considerations, we favor a scenario of a locally deeper base of ground ice that may extend beyond the mapped area, with the radar reflections made more detectable in the mapped area by the relatively low rock abundance of Heimdal’s outer ejecta blanket.
The lack of subsurface detections at the other landing sites may be due to a variety of factors. Despite the presence of layering on scales nominally resolvable by SHARAD at many of the sites, it is possible that there is not sufficient dielectric contrast between the layers to yield a detectable reflection. Another potential explanation is that the radar signal is being attenuated within the uppermost layers, due to either absorption or scattering.Stillman and Grimm[2011] have suggested that surfaces older than Middle Amazonian have been subject to a larger extent than more recent ones to processes leading to the presence of adsorbed water as well as pervasive fracturing, features which may lead to increased absorption and scattering of radar signals and thereby account for the fact that SHARAD subsurface detections have largely been restricted to younger terrains. While they implied that even low-temperature alteration by water (such as influvial and outwash deposits) will preclude detections, SHARAD returns apparently from the base of the Early Amazonian Vastitas Borealis Formation [Campbell et al., 2008] suggest that may not be the case, andEhlmann et al. [2011] have suggested that hydrothermal alteration is needed to produce clay minerals (e.g., smectite) that are sufficiently attenuating to the radar. In any case, crater counting yields age estimates of 0.6 Ga for Heimdal ejecta materials, making the Phoenix site the youngest surface of any landing site to date [Heet et al.,
Unit map: NASA/JPL-Caltech/Washington Univ. St. Louis/JHU APL/Univ. of Arizona
SHARAD detection area
Heimdal Phoenix
Figure 10.Map of geomorphic units ofArvidson et al. [2008] in the region surrounding the Phoenix Mars landing site, overlain with an outline (red) showing the extents of the SHARAD subsurface returns mapped in Figure 1. The SHARAD detections largely overlap with the Lowland Bright geomorphic unit that extends over much of the outer ejecta blanket of Heimdal crater (dashed black line).
2009]. While the counts for the eroded portions of the outer ejecta unit yield a significantly older age of 3.0 Ga [Heet et al., 2009], the Heimdal impact may have produced temperatures and pressures sufficient to vaporize near-surface water ice, shatter rocks, and otherwise alter surface materials [Wrobel et al., 2006] in ways that could reduce their attenuation of radar signals.
SHARAD roughness at the Phoenix site is about average for the northern plains (see Figure 6), but landing- safety concerns were of a different scale, focused on meter-scale boulders found in greater abundance at the other sites being considered. The scales of roughness characterized by SHARAD (~10–100 m) come more into play for factors such as rover trafficability and access to outcrops. From a SHARAD roughness standpoint, the ideal site for a roving spacecraft is one with a broad distribution of roughness that allows both a safe landing location with low roughness and features of scientific interest with high roughness. The Curiosity sitefits this criterion quite well, as do the Spirit site and the other proposed MSL sites (see Figure 8).
5. Conclusions
The SHARAD instrument has mapped an interface at depths of 9–66 m extending over a 2900 km2area that includes the landing site for the Phoenix Mars mission. This interface may represent the base of ground ice, although no similar detection has been found more broadly in the northern plains, and other geologic contacts cannot be ruled out. Relatively rapid aggradation or unique near-surface conditions in the Phoenix Green Valley area may have led to deeper ground ice and/or lower loss of the radar signal, perhaps explaining the geographically limited nature of this detection. Radar reflections from a shallower (~10–15 m) base of ground ice are likely to be obfuscated by sidelobes of the surface return.
Landing sites other than that of Phoenix have not yielded any clear evidence of radar returns from subsurface interfaces, likely due to some combination of feeble dielectric contrasts at volcanic and sedimentary layer interfaces and radar signal attenuation by absorption and/or scattering in the uppermost strata. A putative detection from beneath thefloor of Gale crater in and around the Curiosity landing site has proven to be attributable to an off-nadir return from a surface feature captured in a HRSC clutter simulation but not in one produced from MOLA data. This result provides an important cautionary example that should be considered when interpreting radar results in any area with substantial topographic relief.
Analysis of near-surface roughness from SHARAD returns at ~10–100 m scales complements both that from HiRISE imagery and surface-elevation models at ~1 m scales and that from MOLA data at ~1 km scales. The intermediate scales characterized by SHARAD provide unique information relevant to landing-site trafficability and to local geologic features and processes. While landing-safety considerations typically require smooth surfaces, features of scientific interest (i.e., rocks, outcrops, and geologic structures) are often associated with rougher surfaces, so optimal landing sites—especially for roving spacecraft—will generally exhibit a broad distribution of surface roughness. We found such distributions at many of the past landing sites, including those of the Curiosity and Spirit spacecraft. The global data set of roughness results from SHARAD could contribute toward improved assessments of sites being proposed for future landed missions.
References
Arvidson, R., et al. (2008), Mars Exploration Program 2007 Phoenix landing site selection and characteristics,J. Geophys. Res.,113, E00A03, doi:10.1029/2007JE003021.
Boyce, J. M. and P. J. Mouginis-Mark (2006), Martian craters viewed by the Thermal Emission Imaging System instrument: Double-layered ejecta craters,J. Geophys. Res.,111, E10005, doi:10.1029/2005JE002638.
Bramson, A. M., S. Byrne, N. E. Putzig, S. Mattson, J. J. Plaut, and J. W. Holt (2014), Thick, excess water ice in Arcadia Planitia, Mars, Lunar Planet.
Sci., XLV, Abstract 2120.
Campbell, B. A., N. E. Putzig, L. M. Carter, and R. J. Phillips (2011), Autofocus correction of phase distortion effects on SHARAD echoes,IEEE Geosci. Remote Sens. Lett.,8, 939–942, doi:10.1109/LGRS.2011.2143692.
Campbell, B. A., N. E. Putzig, L. M. Carter, G. A. Morgan, R. J. Phillips, and J. J. Plaut (2013), Roughness and near-surface density of Mars from SHARAD radar echoes,J. Geophys. Res. Planets,118, 436–450, doi:10.1002/jgre.20050.
Campbell, B. A., N. E. Putzig, F. J. IIFoss, and R. J. Phillips (2014), SHARAD signal attenuation and delay offsets due to the Martian ionosphere, IEEE Geosci. Remote Sens. Lett.,11, 632–635, doi:10.1109/LGRS.2013.2273396.
Campbell, B., L. Carter, R. Phillips, J. Plaut, N. Putzig, A. Safaeinili, R. Seu, D. Biccari, A. Egan, and R. Orosei (2008), SHARAD radar sounding of the Vastitas Borealis Formation in Amazonis Planitia,J. Geophys. Res.,113, E12010, doi:10.1029/2008JE003177.
Carr, M. H. (1996),Water on Mars, pp. 248, Oxford Univ. Press, New York.
Carter, L. M., B. A. Campbell, J. W. Holt, R. J. Phillips, N. E. Putzig, S. Mattei, R. Seu, C. H. Okubo, and A. F. Egan (2009a), Dielectric properties of lavaflows west of Ascraeus Mons, Mars,Geophys. Res. Lett.,36, L23204, doi:10.1029/2009GL041234.
Acknowledgments
The Shallow Radar (SHARAD) instrument was provided to NASA’s Mars Reconnaissance Orbiter (MRO) mission by the Italian Space Agency (ASI) and is operated from the Sapienza University of Rome. The MRO mission is managed by the Jet Propulsion Laboratory, California Institute of Technology, for the NASA Science Mission Directorate, Washington, DC. We are grateful to all the people at these and other institutions who have been instrumental in the success of the SHARAD experiment. We also thank SeisWare International Inc. for access to and support of the SeisWare interpreta- tion software, which serves an integral role in our mapping efforts. Data used in this work are available from the NASA Planetary Data System. Insightful comments from Misha Kreslavsky and an anonymous reviewer greatly improved the manuscript.
Carter, L. M., et al. (2009b), Shallow Radar (SHARAD) sounding observations of the Medusae Fossae Formation Mars,Icarus,199, 295–302, doi:10.1016/j.icarus.2008.10.007.
Christensen, P. R., N. S. Gorelick, G. L. Mehall, and K. C. Murray (2013), THEMIS Public Data Releases, Planetary Data System node, Arizona State University. [Available at http://themis-data.asu.edu.]
Clifford, S. M. (1993), A model for the hydrologic and climatic behavior of water on Mars,J. Geophys. Res.,98, 10,973–11,016, doi:10.1029/
93JE00225.
Ehlmann, B. L., J. F. Mustard, S. L. Murchie, J.-P. Bibring, A. Meunier, A. A. Fraeman, and Y. Langevin (2011), Subsurface water and clay mineral formation during the early history of Mars,Nature,479, 53–60.
Golombek, M. P., A. F. C. Haldemann, R. A. Simpson, R. L. Fergason, N. E. Putzig, R. E. Arvidson, J. F. Bell III, and M. T. Mellon (2008), Martian surface properties from joint analysis of orbital, Earth-based, and surface observations, inThe Martian Surface: Composition, Mineralogy, and Physical Properties, edited by J. F. Bell III, pp. 468–497, Cambridge Univ. Press, Cambridge, U. K.
Golombek, M., J. Grant, D. Kipp, A. Vasavada, R. Kirk, R. Fergason, P. Bellutta, F. Calef, K. Larsen, and Y. Katayama (2012), Selection of the Mars Science Laboratory landing site,Space Sci. Rev.,170, 641–737, doi:10.1007/s11214-012-9916-y.
Head, J. W., G. Neukum, R. Jaumann, H. Hiesinger, E. Hauber, M. Carr, P. Masson, B. Foing, H. Hoffmann, and M. Kreslavsky (2005), Tropical to mid-latitude snow and ice accumulation,flow and glaciation on Mars,Nature,434, 346–351.
Heet, T. L., R. E. Arvidson, S. C. Cull, M. T. Mellon, and K. D. Seelos (2009), Geomorphic and geologic settings of the Phoenix Lander mission landing site,J. Geophys. Res.,114, E00E04, doi:10.1029/2009JE003416.
Holt, J. W., M. E. Peters, S. D. Kempf, D. L. Morse, and D. D. Blankenship (2006), Echo source discrimination in single-pass airborne radar sounding data from the Dry Valleys, Antarctica: Implications for orbital sounding of Mars,J. Geophys. Res.,111, E06S24, doi:10.1029/
2005JE002525.
Holt, J. W., A. Safaeinili, J. J. Plaut, J. W. Head, R. J. Phillips, R. Seu, S. D. Kempf, P. Choudhary, D. A. Young, and N. E. Putzig (2008), Radar sounding evidence for buried glaciers in the southern mid-latitudes of Mars,Science,322, 1235–1238, doi:10.1126/science.1164246.
Hynek, B. M. (2004), Implications for hydrologic processes on Mars from extensive bedrock outcrops throughout Terra Meridiani,Nature,431, 156–159, doi:10.1038/nature02902.
Kreslavsky, M. A., and J. W. IIIHead (2000), Kilometer-scale roughness of Mars: Results from MOLA data analysis,J. Geophys. Res.,105, 26,695–26,711, doi:10.1029/2000JE001259.
Kreslavsky, M. A., and J. W. Head III (2002), Mars: Nature and evolution of young latitude-dependent water-ice-rich mantle,Geophys. Res. Lett., 29(15), 1719, 10.1029/2002GL015392.
Mellon, M. T., and B. M. Jakosky (1993), Geographic variations in the thermal and diffusive stability of ground ice on Mars,J. Geophys. Res.,98, 3345–3364, doi:10.1029/92JE02355.
Mellon, M. T., and B. M. Jakosky (1995), The distribution and behavior of Martian ground ice during past and present epochs,J. Geophys. Res., 100, 11,781–11,799, doi:10.1029/95JE01027.
Mellon, M. T., W. C. Feldman, and T. H. Prettyman (2004), The presence and stability of ground ice in the southern hemisphere of Mars,Icarus, 169, 324–340.
Mellon, M. T., W. V. Boynton, W. C. Feldman, R. E. Arvidson, T. N. Titus, J. L. Bandfield, N. E. Putzig, and H. G. Sizemore (2008a), A prelanding assessment of the ice table depth and ground ice characteristics in Martian permafrost at the Phoenix landing site,J. Geophys. Res.,113, E00A25, doi:10.1029/2007JE003067.
Mellon, M. T., R. E. Arvidson, J. J. Marlow, R. J. Phillips, and E. Asphaug (2008b), Periglacial Landforms at the Phoenix Landing Site and the Northern Plains of Mars,J. Geophys. Res.,113, E00A23, doi:10.1029/2007JE003039.
Mellon, M. T., et al. (2009), Ground ice at the Phoenix Landing Site: Stability state and origin,J. Geophys. Res.,114, E00E07, doi:10.1029/
2009JE003417.
Mustard, J. F., C. D. Cooper, and M. K. Rifkin (2001), Evidence for recent climate change on Mars from the identification of youthful near-surface ground ice,Nature,412, 411–414.
Neukum, G., R. Jaumann, and the HRSC Co-Investigator and Experiment Team (2004), HRSC: The High Resolution Stereo Camera of Mars Express, inMars Express: The Scientific Payload, vol. 1240, edited by A. Wilson and A. Chicarro, pp. 17–35, ESA Special Publication, Noordwijk, Netherlands.
Phillips, R. J., et al. (2008), Mars north polar deposits: Stratigraphy, age, and geodynamical response,Science,320, 1182–1185, doi:10.1126/
science.1157546.
Phillips, R. J., B. J. Davis, K. L. Tanaka, S. Byrne, M. T. Mellon, N. E. Putzig, R. M. Haberle, M. A. Kahre, B. A. Campbell, and L. M. Carter (2011), Massive CO2 ice deposits sequestered in the south polar layered deposits of Mars,Science,332, 838–841, doi:10.1126/science.1203091.
Plaut, J. J., A. Safaeinili, J. W. Holt, R. J. Phillips, J. W. Head III, R. Seu, N. E. Putzig, and A. Frigeri (2009), Radar evidence for ice in lobate debris aprons in the mid-northern latitudes of Mars,Geophys. Res. Lett.,36, L02203, doi:10.1029/2008GL036379.
Putzig, N. E., R. J. Phillips, J. W. Head, B. A. Campbell, A. F. Egan, J. J. Plaut, L. M. Carter, R. Seu, M. T. Mellon, and the SHARAD Team (2009a), Do shallow radar soundings reveal possible near-surface layering throughout the northern lowlands of Mars?, Lunar Planet. Sci., XL, Abstract 2477.
Putzig, N. E., R. J. Phillips, J. J. Plaut, M. T. Mellon, J. W. Head, B. A. Campbell, L. M. Carter, A. Egan, and R. Seu (2009b), Shallow Radar soundings of the southern highlands of Mars,Eos Trans. AGU,90(52), Fall Meet. Suppl., Abstract P13B-1280.
Putzig, N. E., B. A. Campbell, R. J. Phillips, and M. T. Mellon (2012), SHARAD sounding and surface roughness of once and future Mars landing sites, Lunar Planet. Sci., XLIII, Abstract.
Safaeinili, A., R. Phillips, N. Putzig, J. Plaut, R. Seu, D. Biccari, J. Holt, S. Smrekar, and D. Nunes (2008), Radar detection of a subsurface horizon at the Phoenix landing site,Eos Trans. AGU,89(53), Fall Meet. Suppl., Abstract P32B-06.
Seu, R., et al. (2007a), Accumulation and erosion of Mars’south polar layered deposits,Science,317, 1715–1718, doi:10.1126/science.1144120.
Seu, R., et al. (2007b), SHARAD sounding radar on the Mars Reconnaissance Orbiter,J. Geophys. Res.,112, E05S05, doi:10.1029/2006JE002745.
Smith, D. E., H. V. Frey, J. B. Carvin, H. J. Zwally, F. G. Lemoine, D. D. Rowlands, J. B. Abshire, R. S. Afzal, X. Sun, and M. T. Zuber (2001), Mars Orbiter Laser Altimeter: Experiment summary after thefirst year of global mapping of Mars,J. Geophys. Res.,106, 23,689–23,722, doi:10.1029/2000JE001364.
Stillman, D. E. and R. E. Grimm (2011), Radar penetrates only the youngest geological units on Mars,J. Geophys. Res.,116, E03001, doi:10.1029/
2010JE003661.
Watters, T. R., C. J. Leuschen, B. A. Campbell, J. A. Grant, M. Ga, N. E. Putzig, and R. J. Phillips (2014), MARSIS Subsurface radar sounding of Meridiani Planum, Mars: Implications for the properties of the plains deposits, Lunar Planet. Sci., XLV, Abstract 2521.
Wrobel, K., P. Schultz, and D. Crawford (2006), An atmospheric blast/thermal model for the formation of high-latitude pedestal craters, Meteorit. Planet. Sci.,41, 1539–1550, doi:10.1111/j.1945-5100.2006.tb00434.x.
