Formation and erosion of layered materials: Geologic and dust cycle history of eastern Arabia Terra, Mars

Authors


Abstract

[1] Eastern Arabia Terra is mantled in a layer of dust a few centimeters to a meter thick, yet contains morphologic features that suggest a history of multiple events of deposition and consolidation of fine-grained material and a significant amount of erosion. Early in Martian history, this region was affected by volcanic and fluvial activity but has since been dominated by aeolian processes. Five craters in this region contain interior mound material that ranges in height from 1600 to 2100 m above the crater floor. The fluted erosional pattern and the thermal inertia are suggestive of a weakly indurated material, and the extensive layering implies that these mounds were formed by a repeated process or processes. Although these materials primarily occur within craters, there are materials outside craters that have similar erosional features and fine laminations, suggesting a more extensive deposit. The most likely process to form this material is the deposition and cementation of air fall dust and is potentially related to Martian obliquity changes. The significant amount of erosion of the intracrater mounds unit indicates a dramatic change from a depositional environment to an erosional regime over the past 106–108 years. Currently dust is accumulating in this region in years with planet-encircling dust events, but global circulation model results indicate that dust devils may be removing slight amounts of dust from Arabia Terra. These observations suggest that the thickness of the dust mantle may not be currently increasing and may instead be in equilibrium.

1. Introduction

[2] Low thermal inertia regions on Mars, such as Tharsis Montes, Arabia Terra, and Elysium Planitia cover about a third of the Martian surface and the majority of the northern midlatitudes and are interpreted to be surfaces mantled in dust [e.g., Kieffer et al., 1973, 1977; Palluconi and Kieffer, 1981]. Eastern Arabia Terra (referred to as Arabia Terra in this work) contains morphologic features that can be viewed through this dust mantle, such as layers exposed within craters [e.g., Malin and Edgett, 2001; Edgett and Malin, 2002], dark intracrater material [e.g., Arvidson, 1974; Christensen, 1983; Thomas, 1984; Edgett and Malin, 2000; Edgett, 2002; Wyatt et al., 2003], and evidence for fluvial and volcanic processes [e.g., Zimbelman and Greeley, 1982; Greeley and Guest, 1987; Anguita et al., 1997]. The relationship between the processes that produced these features and those that formed the dust mantle offers insight into geologic environments and climates on Mars.

[3] Arabia Terra (Figure 1) is an ideal location to study the current and historical transport and deposition of dust because there are no major topographic structures, such as the Tharsis or Elysium Montes, that influence local wind circulation patterns, and this region has likely been a region with atmospheric conditions favorable for dust accumulation throughout much of Martian history [e.g., Haberle et al., 2003; Kahre et al., 2006; Haberle et al., 2006]. The objectives of this work are to (1) better understand the current sedimentary processes of dust accumulation and sand transport and its relationship to past sedimentary processes of deposition and erosion; and (2) expand these ideas to gain new insight into the cycles of dust circulation and deposition on Mars and relate this information to understanding past climates. These objectives have been addressed using multiple data sets to create a unit map of eastern Arabia Terra, Mars. This map is then used to facilitate the interpretation of the geologic history of this region, focusing on the present and past dust cycle and the formation of layered materials.

Figure 1.

Arabia Terra, Mars regional context. TES-derived bolometric thermal inertia (16 pixels per degree resolution) of Arabia Terra, Mars, overlaid onto MOLA shaded relief (32 pixels per degree). The study region is boxed in white.

1.1. Transport and Properties of Martian Dust

[4] Dust is believed to be primarily transported and deposited during and just after major dust events [e.g., Kahn et al., 1992]. In the present climate, planet-encircling dust events typically originate in the southern hemisphere in the regions of Hellespontus west of Hellas basin or Solis Planum [e.g., Briggs et al., 1979; Zurek, 1982; Martin, 1984; Strausberg et al., 2005]. After a planet-encircling dust event, dust settles out of the atmosphere uniformly over the planet. Areas of high wind shear stress are observed to have an increased atmospheric opacity, indicating the presence of airborne dust [Christensen, 1982], and experience post dust event darkening [e.g., Sagan and Pollack, 1969; Pleskot and Miner, 1981; Christensen, 1988]. These typically low-albedo surfaces become a local dust source as dust is removed from these regions and is redeposited on adjacent areas of low wind shear stress [Christensen, 1982]. Dust is also observed locally, and has been observed at all of the Martian landing sites [e.g., Moore et al., 1977, 1999; Greeley et al., 1999; Soderblom et al., 2004; Herkenhoff et al., 2004; Christensen et al., 2004b]. Thus, dust likely covers much of the planet but in thinner, more patchy deposits than the low thermal inertia regions. The thermal infrared spectral signature of dust is similar for all moderate to high-albedo regions [e.g., Bandfield and Smith, 2003], and the spectral and chemical signature of the dust component is similar at each landing site [e.g., McSween and Keil, 2000; Christensen et al., 2004b, 2004c, Yen et al., 2005]. Both observations indicate that bright air fall dust is not derived from local material, but instead suggests a global mixing mechanism for surface dust [e.g., McSween and Keil, 2000; Ruff and Christensen, 2002; Yen et al., 2005].

[5] The effective radius of atmospheric dust is ∼1.5 μm [Toon et al., 1977; Pollack et al., 1979; Clancy et al., 2003; Wolff and Clancy, 2003; Wolff et al., 2006]. This is lower than the 20–40-μm diameter implied by the thermal inertia data in regions inferred to be mantled in dust [Christensen, 1986a; Mellon et al., 2000; Putzig et al., 2005]. Jakosky [1986] suggests that gas conductivity may not vary significantly between particle diameters of 3 and 40 μm, and therefore the thermal inertia of 3- and 40-μm diameter particles may not be appreciably different [Christensen, 1986a]. Presley and Christensen [1997] measured the conductivity for 0–11 μm, 11–15.6 μm, 15.6–20 μm, and 25–30 μm size fractions. Although the 15.6–20 μm and 25–30 μm diameter particles have a similar conductivity, all other size fractions differ in conductivity under Martian atmospheric conditions. This laboratory result suggests that some additional process has occurred to increase the particle size of the surface dust mantle relative to atmospheric dust. This dust material could be a pyroclastic ashfall deposit in which air-filled vesicles lower the thermal inertia significantly [e.g., Kieffer et al., 1973; Zimbelman, 1986]. However, this material has a uniform thermal inertia and albedo, whereas a pyroclastic ash would likely have more variable thermophysical characteristics because of variations in erosion of the ash material [Zimbelman, 1986]. Thus, an unconsolidated air fall dust interpretation is a more likely explanation for the low thermal inertia deposits. The low thermal inertia regions may be composed of larger particle sizes than that of atmospheric dust, or more likely, some later process, such as electrostatic forces [e.g., White et al., 1997] or cementation [e.g., Jakosky and Christensen, 1986; Arvidson et al., 2006], may be causing smaller sized air fall dust particles to bond together.

1.2. Geologic Setting

[6] The study region in Arabia Terra is located from 20.0°E to 32.5°E longitude and 6.5°S to 13.5°N latitude and using primarily Viking data was mapped into dissected (Npld), ridged (Nplr), subdued cratered (Npl2), and ridged plains material (Hr) units [Greeley and Guest, 1987]. This region was mapped using visible images and although a dust mantle was known to be present, this dust layer was not considered and features observed through the dust mantle were mapped. The majority of this area is dissected unit material (Npld) interpreted to have formed during heavy bombardment during the Noachian Period. This surface is likely a mixture of volcanic materials, erosional products, and impact breccia, and is highly dissected by channels and channel networks. Ridged unit material (Nplr) is found in the western portion of the study area, is interpreted to be flood lava flows and ridges likely due to volcanic processes, and is also Noachian in age. In Henry crater subdued crater unit material (Npl2) fills the crater floor, and is interpreted to be either thin lava flows or sedimentary deposits [Greeley and Guest, 1987]; this is the only mapped occurrence of this material in the study region. In addition, there is a small region of ridged plains material (Hr) in the southern portion of the study area that is interpreted to be low-viscosity lava flows and Hesperian in age. Some craters contain smoother material filling the floors and occasionally central peaks, crater-rim, and crater-ejecta material are present [Greeley and Guest, 1987].

[7] Arabia Terra has a Thermal Emission Spectrometer (TES) [Christensen et al., 1992, 2001] bolometric albedo of 0.25 to 0.30, a TES-derived thermal inertia of 40–120 J m−2 K−1 s−1/2, and is interpreted to be a surface dominated by unconsolidated fines less than about 40 μm in diameter [e.g., Palluconi and Kieffer, 1981; Mellon et al., 2000; Putzig et al., 2005]. The regional rock abundance from Viking data is 5–7% [Christensen, 1982, 1986b; Zimbelman and Greeley, 1982] and less than ∼7% from TES data [Nowicki and Christensen, 2007], and is among the lowest observed on the planet. On the basis of Viking data, Arabia Terra is interpreted to be mantled in a dust layer a few centimeters to 1–2 m thick; the thickness is constrained by the low thermal inertia and morphologic features that can still be discriminated [Christensen, 1986a]. The dust layer has been interpreted to be less than ∼105 years old, on the basis of estimates of average annual dust deposition from planet-encircling dust events (10 μm/a, where a is years) and assuming a dust mantle thickness of 1 m [Christensen, 1986a]. Also, Bandfield and Edwards [2008] infer that Arabia Terra, Elysium Planitia, and Tharsis Montes are extraordinarily smooth surfaces at centimeter to meter scales, consistent with fine material mantling the surface.

[8] The study region has the highest concentrations of hydrogen outside the polar regions, as identified by the Mars Odyssey Gamma Ray Spectrometer/Neutron Spectrometer/High-Energy Neutron Detector (GRS/NS/HEND) instrument suite [e.g., Boynton et al., 2002; Feldman et al., 2002; Mitrofanov et al., 2002, Feldman et al., 2004; Mitrofanov et al., 2004; Ivanov et al., 2005; Jakosky et al., 2005], and is interpreted as having up to ∼10% water content in the uppermost surface layer [Feldman et al., 2004; Mitrofanov et al., 2004; Ivanov et al., 2005; Jakosky et al., 2005]. Because of a rough correlation with low-albedo terrains, including Arabia Terra, some hypotheses suggest that the enrichment in hydrogen is related to the surface dust mantle, likely in either adsorbed water from the atmosphere or a chemically bound form [e.g., Mitrofanov et al., 2004; Ivanov et al., 2005], with water-bearing minerals being the preferred explanation [Ivanov et al., 2005]. However, there is not a strong correlation between the low thermal inertia surfaces and the hydrogen elemental abundance, particularly on local scales and in the Tharsis region [Jakosky et al., 2005], and therefore atmospheric phenomenon may be playing a role [Jakosky et al., 2005].

[9] This dust layer masks the underlying surface from remote sensing observations, preventing the identification of the surface composition or the thermal inertia of this material. However, morphologic features, such as wrinkle ridges, channels, and bed forms, can be identified through this mantle [e.g., Zimbelman and Greeley, 1982; Anguita et al., 1997; Malin and Edgett, 2001]. In addition, craters containing dark albedo material in their floors, similar to intracrater deposits observed in Oxia Palus to the west [e.g., Arvidson, 1974; Christensen, 1983; Thomas, 1984; Presley and Arvidson, 1988; Edgett and Christensen, 1994; Edgett, 2002; Edgett and Malin, 2002; Wyatt et al., 2003] are common, and the Mars Orbiter Camera (MOC) [Malin et al., 1992] and High Resolution Imaging Science Experiment (HiRISE) [McEwen et al., 2007] have imaged intracrater materials, including sand dunes, sand sheets, layered materials, and mesa-forming units [Malin and Edgett, 2000; Edgett and Malin, 2000; Malin and Edgett, 2001].

2. Approach

2.1. Data Sets

[10] To help better understand the geologic and dust cycle history in Arabia Terra, many data sets were integrated to create a unit map of the study region. These data sets include Thermal Emission Imaging System (THEMIS) daytime and nighttime infrared images [Christensen et al., 2003, 2004a], THEMIS [Christensen et al., 2003, 2004a], MOC [Malin et al., 1992; Malin and Edgett, 2001], and HiRISE [McEwen et al., 2007] visible images, TES bolometric [Jakosky et al., 2000; Mellon et al., 2000] and THEMIS [Fergason et al., 2006] thermal inertia, TES albedo [Christensen et al., 1992, 2001], and Mars Orbiter Laser Altimeter (MOLA) data [Zuber et al., 1992; Smith et al., 1999] (Figure 2).

Figure 2.

Data sets used in analysis. (a) THEMIS daytime infrared mosaic (100 m per pixel resolution), (b) THEMIS nighttime infrared mosaic (100 m per pixel resolution), (c) THEMIS thermal inertia (100 m per pixel resolution), (d) TES bolometric thermal inertia (16 pixels per degree resolution), (e) TES albedo (16 pixels per degree resolution), (f) TES dust cover index (16 pixels per degree resolution), and (g) MOLA elevation (64 pixels per degree resolution). Black lines are due to a lack of data coverage.

2.1.1. Imaging Data

[11] The THEMIS infrared data were used to map regional-scale morphologic features and identify thermophysical differences between surface materials. These data are acquired using an uncooled multispectral microbolometer array with 320 cross-track pixels and 240 down-track pixels. It has an instantaneous field of view (IFOV) of ∼100 m per pixel and an image width of ∼32 km. THEMIS band 9 data, centered at 12.57 μm, was used to generate both the daytime and nighttime mosaics. Standard THEMIS data processing consisting of decompression, radiometric calibration, and systematic noise removal was applied to each image. A detailed description of the THEMIS instrument and calibration is provided by Christensen et al. [2004a].

[12] THEMIS, MOC, and HiRISE visible images were incorporated in this study to analyze small-scale features, such as crater fill material, layered units, and surface textures. The THEMIS visible imager has 1024 × 1024 μm pixels with a 10-m instantaneous field of view resulting in a spatial resolution of ∼18 m per pixel. All MOC images used in this work were acquired with the narrow angle camera. Images can range from 256 to 2048 elements across in increments of 16 pixels, and downtrack lengths can be an integer value in units of 128 lines. From an altitude of 400 km, this typically covers a region just over 3 km across at a spatial resolution of ∼1.5 m per pixel. A detailed description of the MOC camera and experiment details are given by Malin et al. [1992] and Malin and Edgett [2001]. The HiRISE instrument is the highest-resolution imager used in this work has an effective swath width of ∼20,048 pixels (RED images), and has a typical projected spatial resolution of ∼25 cm per pixel. For additional instrument detail, see McEwen et al. [2007].

2.1.2. Thermophysical Data

[13] The THEMIS data set provides the highest-resolution thermophysical data to date (100 m per pixel) and is used to derive the thermal inertia of the surface materials and assess their physical characteristics. In this work, we use the method of Fergason et al. [2006] to derive thermal inertia values from THEMIS nighttime infrared data. The brightness temperature of the surface is determined by fitting a Planck curve to band 9 (centered at 12.57 μm) calibrated radiance that has been corrected for instrumental effects. Nighttime temperatures only were used because the effects of albedo and sun-heated slopes have mostly dissipated throughout the night, and the thermal contrast due to differences in particle sizes are at a maximum [e.g., Kieffer et al., 1973, 1977; Jakosky, 1979; Palluconi and Kieffer, 1981]. The THEMIS band 9 temperatures are converted to a thermal inertia by interpolation within a seven-dimensional look-up table using latitude, season, local solar time, atmospheric dust opacity, thermal inertia, elevation (atmospheric pressure), and albedo as input parameters. Model parameters appropriate for the THEMIS image and the measured band 9 surface temperatures are then used to interpolate the thermal inertia between these calculated look-up table node values. Uncertainties in the THEMIS derived thermal inertia values are primarily due to instrument calibration, uncertainties in model input parameters at the resolution of the THEMIS instrument, and thermal model uncertainties. Considering all these uncertainties, the absolute accuracy of the THEMIS thermal inertia is ∼20%. For additional detail, see Fergason et al. [2006].

[14] TES thermal inertia is derived from a single nighttime temperature [Christensen et al., 1992, 2001], where a seven-dimentional look-up table (including albedo, thermal inertia, surface pressure, dust opacity, latitude, time of day, and season) is used to determine the model-derived thermal inertia that best fits the observed TES temperature under the appropriate physical conditions [Jakosky et al., 2000; Mellon et al., 2000]. The total model-based uncertainty of the bolometer-based and spectrometer-based thermal inertia is 6.0% and 16.9%, respectively [Mellon et al., 2000], and does not include uncertainties due to model input parameters. In this work, bolometeric thermal inertia data only was used, and constraints for the TES data include a longitude of 20.0–32.50°E, latitude of 6.5°S–13.5°N, emission angle less than 30°, nighttime data only, high-quality thermal inertia values, and avoiding seasons with a dusty atmosphere.

2.1.3. Additional Data Sets

[15] TES albedo, DCI, and MOLA elevation were also incorporated in this study. The TES instrument measured the bolometric albedo at 3 km per pixel (0.3–2.9 μm), and the absolute calibration is 1–2% [Christensen et al., 1992, 2001]. TES surface dust cover index (DCI) [Ruff and Christensen, 2002] is sensitive to volume scattering effects that commonly occur in silicates at particle sizes greater than ∼63 μm. It is used to distinguish dust-covered areas (DCI< 0.940) from less dusty areas (DCI > 0.962), and values that are intermediate represent surfaces that are partially dust covered. The DCI for a given spectrum is calculated by averaging TES effective emissivity between 1350 and 1400 cm−1 (∼7.1–7.4 μm). For a complete description of the DCI technique, see Ruff and Christensen [2002]. The MOLA elevation data was also utilized in this study. For additional information on the MOLA instrument and the derivation of surface elevation, see Zuber et al. [1992] and Smith et al. [1999].

2.2. Method

[16] Using multiple data sets, a unit map was constructed as a basis for determining the relative age of and environmental relationships between surfaces and to select areas for a more focused analysis. Units were first identified on the basis of thermophysical differences distinguished using THEMIS nighttime infrared mosaics. These units were then refined and further subdivided on the basis of surface texture variations and morphologic features observed in visible images and elevation differences observed in MOLA data. THEMIS daytime and nighttime infrared images [Christensen et al., 2003, 2004a] were used to identify channels, pedestal craters, low-albedo deposits, and the degree of crater degredgation. THEMIS, narrow angle MOC, and HiRISE visible images were then analyzed to confirm unit boundaries and to study morphologic features, such as layered materials and erosional characteristics of the intracrater mound material, in detail. THEMIS-derived thermal inertia (100 m per pixel) [Fergason et al., 2006] was used in conjunction with TES thermal inertia (3 × 6 km per pixel) [Jakosky et al., 2000; Mellon et al., 2000; Putzig et al., 2005] to differentiate thermophysical units and quantitatively interpret the physical nature of each surface. An effective particle size [Kieffer et al., 1973] was determined using the method described by Presley and Christensen [1997] for surfaces with a thermal inertia below 350 J m−2 K−1 s−1/2. TES albedo and the DCI [Ruff and Christensen, 2002] were used to identify the presence of surface dust, and MOLA topography was used to measure the thickness of layers and unit materials. The combination of data sets formed the basis for describing the surface materials, designating specific units, and interpreting the geologic history of this area.

3. Observations

[17] The study region was mapped into the following four units on the basis of differences in the thermophysical and morphologic characteristics of the surface: (1) plains material, (2) intracrater mounds, (3) intracrater material, and (4) wind streaks (Figure 3) and are summarized in Table 1.

Figure 3.

Arabia Terra unit map. Unit map of eastern Arabia Terra, Mars, based on thermophysical and morphological features. The unit map is overlaid on a THEMIS daytime infrared mosaic.

Table 1. Arabia Terra Unit Descriptions
Unit NameTHEMIS Thermal InertiaTES Thermal InertiaEffective Particle SizeaAlbedoDCISurface Characteristics
  • a

    Effective particles sizes are derived from THEMIS thermal inertia ranges.

Plains Material20–11525–80<15 μm; dust0.260.91–0.94morphologic features observed through dust mantle include fresh craters, lava flows, channels, and bed forms
Intracrater Moundstypically 40–14025–80<35 μm; dust or fine sand0.25–0.270.93–0.96height ranging from 300 to 2500 m, erodes in a fluted pattern
Intracrater Material140–460200–450>160 μm; sand0.18–0.200.95–0.98elliptically shaped with gradational boundaries suggestive of aeolian deposition
Wind Streaks115–350120–17015 μm–1.8 mm0.16–0.260.94–0.98majority oriented toward the south-southwest

3.1. Plains Material

[18] The plains material unit occurs over the majority of the study region. The TES bolometric and THEMIS-derived thermal inertia of the plains material unit is 25–80 J m−2 K−1 s−1/2 and 20–115 J m−2 K−1 s−1/2, respectively, and the TES albedo is ∼0.26. Both the thermal inertia and albedo are characteristic of unconsolidated to weakly consolidated surface dust, and the thermal inertia indicates a minimal dust thickness of several thermal skin depths (∼5–10 cm). The spectrally derived dust cover index (DCI) is 0.91–0.94, and is further evidence for the presence of fine-grained, unconsolidated material [Ruff and Christensen, 2002].

[19] This unit was originally mapped using Viking data into dissected (Npld) and ridged (Nplr) units [Greeley and Guest, 1987]. Using higher-resolution images and improved elevation data, these units were revised in this work and mapped into the revised dissected (revised Npld) and revised ridged (revised Nplr) subunits (Figure 3). The revised Nplr subunit was expanded to include topographically lower-lying areas that have a smoother surface, are less cratered, and have fewer channels than the revised Npld subunit. In addition, the revised Npld subunit has been eroded, leaving mesas or islands of remnant material that are ∼60 m in thickness (Figures 4 and 5) . Wrinkle ridges are often distinguishable throughout both subunits.

Figure 4.

Figure context map. THEMIS daytime infrared mosaic of the study region with white boxes indicating the locations of Figures 517.

Figure 5.

Layered material beneath the dust mantle. THEMIS visible image V05529007 (18 m per pixel resolution; centered at 25.62°E, 4.53°N) of layered material in Arabia Terra that is mantled in air fall dust.

[20] The majority of channels occur in the revised Npld subunit and are observed in clusters that are often separated by expanses of surface material devoid of channels. There are five clusters of channels in this region, including portions of the Naktong Vallis and Verde Vallis systems. These channels range from 35 to 150 km in length, and occur as either tributary systems (Figures 4 and 6a) or single channels (such as in crater walls) that do not connect to a larger system (Figures 4 and 6b). The channels occur in relative topographic lows, and assuming the current topography is representative of the landscape in which these channels formed, the channels sometimes terminate into craters. Within channels or topographic depressions and alongside the base of mesas, bed forms are observed (Figures 4 and 7) . These bed forms have wavelengths of a few 100 m, are similar in albedo to their surroundings, and are likely transverse aeolian ripples [Malin and Edgett, 2001; Wilson and Zimbelman, 2004].

Figure 6.

Variable channel patterns. (a) Portion of the THEMIS daytime infrared mosaic (100 m per pixel resolution), centered at 30.4°E, 7.5°N, illustrating a channel tributary system in the study region. The black line is due to a lack of data coverage. (b) Portion of the THEMIS daytime infrared mosaic (100 m per pixel resolution), centered at 22.5°E, 11.5°N, illustrating independent channels found in the wall of Henry Crater.

Figure 7.

Transverse aeolian ripples. Narrow angle MOC image R0401700, centered at ∼33.3°E, ∼4.8°N, of transverse aeolian ripples.

[21] In addition, three crater subunits (filled, unfilled, and pedestal) were mapped, for a total of five subunits in the plains material unit. Craters larger than ∼10 km in diameter were mapped according to the presence or absence of crater fill material and the degree of preservation of the crater rim (Figure 3). Craters with smooth floors or obvious infilling material were mapped as filled. Craters with raised and preserved rims and a lack of smooth surface material filling the crater were mapped as unfilled. Pedestal and rampart craters were mapped separately regardless of the presence of infilling material or crater rim appearance. The majority of craters have material that is infilling the interior. This fill material is often smooth with some wrinkle ridges present, but many craters have low-albedo deposits, or layered materials as well. The majority of crater ejecta in this region has been eroded and is no longer present; raised crater rims however, are common. Pedestal and rampart craters are primarily clustered in the central column of the study area, and the majority of these craters occur within 260 km of one another. These craters may be evidence that the subsurface contained volatiles during the crater's emplacement [e.g., Mouginis-Mark, 1979; Schultz and Lutz, 1988].

3.2. Intracrater Mounds

[22] Five craters in this region contain interior mound material that ranges in height from 1600 to 2100 m (Figure 3). This material does not rise above the crater rim and often fills the majority of the crater interior (e.g., Figures 4 and 8) . Most mounds are off center in the floor of the craters. Their position varies, but they are often skewed toward the western portion of the crater, and their position is likely an indication of the dominant wind direction in this region. In addition, the mounds are always associated with a low albedo, presumably unconsolidated deposit (part of the intracrater material unit). The TES albedo of the mounds ranges from 0.25 to 0.27, and this albedo is always consistent with the dust-mantled surface outside the crater. The TES DCI is 0.93 to 0.96 indicating that these materials are partially dust covered. The combined DCI and albedo values suggests that these surfaces are covered in a layer of dust that is thick enough to dominate the visible albedo data but not thick enough to dominate the infrared DCI data. The mounds have variable THEMIS thermal inertia values. The majority of mounds have a low thermal inertia (40–140 J m−2 K−1 s−1/2), suggesting that they are mantled by a minimum of a few centimeters of dust or fine sand. In one crater centered at 26.5°E, 1.5°N (Figures 4 and 9) , material on the crater floor erodes in a similar manner as the mound material. Because of the similar erosional nature, it is likely that this material is a remnant of the crater mound material. Relatively dark material (TES albedo of 0.18–0.20) is along the slopes and is probably unconsolidated sand that is scouring the surface or incorporating air fall dust in the deposit. This remnant material has a higher thermal inertia (400–435 J m−2 K−1 s−1/2) than the intracrater mound material. This higher thermal inertia value and the presence of dark material suggest that this surface is either free of dust or the dust layer is too thin (tens of microns) to measurably lower the thermal inertia. If this material is indeed a remnant of the intracrater mound unit, then a thermal inertia of 400–435 J m−2 K−1 s−1/2 is likely indicative of the intracrater mounds as well.

Figure 8.

Typical intracrater mound unit material. (a) Portion of the THEMIS daytime infrared mosaic (100 m per pixel resolution) of an example of the interior mound unit material, centered at 20.9°E, 8.5°N. (b) Portion of narrow angle MOC image M2000102 illustrating the erosional morphology of this material.

Figure 9.

Intracrater mound example. (a) Portion of THEMIS thermal inertia image I01229007 (100 m per pixel resolution), centered at 26.6°E, 1.3°N, illustrating the range in thermal inertia values of possibly dust-free intracrater mound material. (b) Portion of narrow angle MOC image M0310925 illustrating the texture of the mound material and the darker material traveling across the mound surface.

[23] In all cases, the interior mound material contains hundreds of individual layers (Figures 4 and 10). These layers are mantled in dust, so differences in the properties of layers based on composition or albedo are not identifiable. In HiRISE and MOC images, layers appear uniform in texture, similar in the degree of induration, and form terraces. Layered knobs and mesas are also observed. The layers appear similar in all crater mound materials, but the thickness of these layers and the frequency of their exposure vary between locations. The mounds also erode in a fluted or yardang pattern that occurs on the slopes and often on all sides of the mound, and are often associated with a scalloped erosional pattern (Figures 4 and 8).

Figure 10.

Layered material. Portion of HiRISE image PSP_003734_1905_RED (∼25 cm per pixel), centered at 24.3°E, 10.5°N, illustrating layered materials within Henry crater.

3.3. Intracrater Material

[24] Thermally distinct intracrater material is present in 14 craters in the study region (Figure 3). The TES bolometric thermal inertia and albedo of this material is 200–450 J m−2 K−1 s−1/2 and 0.18–0.20, respectively. The THEMIS thermal inertia values range from 140 to 460 J m−2 K−1 s−1/2, and are in good agreement with the thermal inertia derived from TES. The TES DCI is 0.95–0.98, indicating that the surfaces are partially dust covered in some areas and free of dust in others. These deposits only occur in craters with raised topography, such as intracrater mound unit material or central crater peaks, in the crater interior. The erosion of this topographically high material may be a potential source material. These deposits are typically elliptically shaped deposits, and the size of the deposit is unrelated to the crater size. In both THEMIS and MOC visible images, the deposits have a darker appearance in the center and then lighten toward the edges, suggesting a variable thickness of unconsolidated material or variably thick veneer of bright dust. This variability is also suggested by the DCI values. In some cases the deposit is thin enough that the underlying surface texture can be observed through it (Figures 4 and 11). These materials often, but not always, accumulate in the southern portion of craters between intracrater mounds and the crater wall. Exceptions include Henry crater (centered at 23.3°E, 10.9°N) where deposits are found in the eastern and western portions of the crater, and an unnamed crater centered at 28.2°E, 2.4°N, in which the deposit is found in the center of the crater near a small outcrop of the intracrater mound material.

Figure 11.

Intracrater material unit. Portion of THEMIS visible image V03020004 (18 m per pixel resolution), centered at 21.1°E, 8.3°N, illustrating an intrcrater material deposit of variable thickness. The underlying surface texture is observed through the intracrater material.

3.4. Wind Streaks

[25] Five craters have wind streaks that are identifiable in the THEMIS infrared daytime and nighttime data and occur in the southern portion of the study area only (Figure 3). The TES albedo of these streaks varies from 0.16 to 0.26, and the TES DCI ranges from 0.94 to 0.98. The THEMIS thermal inertia values range from 115 to 350 J m−2 K−1 s−1/2, which is more variable than TES bolometric thermal inertia values of 120–170 J m−2 K−1 s−1/2. The lower albedos are correlated with higher thermal inertia values. Higher thermal inertia values measured by THEMIS is likely a resolution effect, as THEMIS can resolve small crater rims, ridges, and additional features that may have elevated thermal inertia values, but are too small to be resolved by the TES data set. In this study area, the orientation of wind streaks ranges from 30° to 330°, and the wind streak orientation is often variable within the same streak. The two most common wind streak orientations are 180° to 220° and 300° to 330°, and all streaks originate at the intracrater material unit (Figures 4 and 12).

Figure 12.

Complex windstreak. THEMIS thermal inertia mosaic (100 m per pixel resolution) overlaid onto THEMIS daytime temperature mosaic (100 m per pixel resolution) of a wind streak. The primary wind streak orientation is toward the south-southwest, and white arrows indicate a secondary windstreak orientation toward the northwest.

4. Geologic History

[26] The mapping of units and the identification of age relationships between these units enables the interpretation of the geologic history of the study region. There is significant evidence that this region has not always been an area of dust deposition and accumulation, and this suggestion is also consistent with current global circulation model (GCM) results [Kahre et al., 2006; Haberle et al., 2006]. Instead, Arabia Terra has a complex history including volcanic, fluvial, and aeolian processes, which suggests that this region has undergone multiple environmental and climactic changes throughout its history. Following is a discussion of the proposed geologic history of this region, in chronological order when relative ages can be established, beginning with the oldest observed surface.

4.1. Ancient

4.1.1. Analysis of Plains Material Unit Crater Populations

[27] The age of the plains material surface was estimated by examining crater populations, and these results are summarized in Table 2. We matched geologic units within the study area with the crater database of Barlow [1988] for three crater populations, and use the crater density boundaries for Martian epochs as described by Tanaka [1986]. Interpretations of theses surfaces are based on surface textures, superposition relationships, and crater population statistics. The first population consists of all craters or eroded remnants ≥5 km in diameter in the study area. We interpret the total crater population to reflect surface ages of both the revised Nplr and revised Npld subunits. For the entire surface, these materials were emplaced during the middle to upper Noachian Period. The second population contains all craters ≥5 km in diameter in the revised Npld subunit. This surface has likely been reworked and the current surface materials were emplaced during the lower to middle Noachian period. The third population contains all craters ≥5 km in diameter in the revised Nplr subunit, and is interpreted as a topographically lower surface that has been infilled with material, possibly eroded from the revised Npld subunit or from volcanic infilling. On the basis of crater density statistics, this surface was emplaced during the middle to upper Noachian Period. Both the revised Npld and revised Nplr subunit surfaces are dated as approximately middle Noachian in age, and are consistent with crater densities derived by Greeley and Guest [1987].

Table 2. Cumulative Crater Counts and Interpreted Surface Ages
Unit NameArea (106 km2)N(5)aN(16)aAgeb
  • a

    N(5) and N(16) are the number of craters with a diameter ≥5 and ≥16 km, respectively, that occur in a 106 km2 area.

  • b

    Crater density boundaries for Martian epochs are from Tanaka [1986].

Total0.7671419 ± 23173 ± 15middle to upper Noachian
Revised Npld0.5097434 ± 25183 ± 19lower to middle Noachain
Revised Nplr0.2574389 ± 39155 ± 25middle to upper Noachian

4.1.2. Lava Flow Emplacement and Cratering Activity

[28] Likely in the Noachian Period, multiple volcanic lava flows were deposited [e.g., Tanaka, 1986; Greeley and Guest, 1987; this study]. On the basis of the lack of volcanic constructs in this region, this activity was probably fissure vent volcanism early in Martian history. Cratering also occurred during this time, and is consistent with the period of heavy bombardment [Tanaka, 1986; Greeley and Guest, 1987]. The frequent cratering activity homogenized the surface, and resulting craters and crater ejecta have masked evidence of volcanic features, such as lava flow fronts and source vents. Subsequent fluvial or aeolian processes may have also obscured source vents. Wrinkle ridges are observed throughout the study region and are likely tectonic in origin [e.g., Watters, 1988, 1993; Chicarro and Schultz, 1985].

4.1.3. Channel Formation

[29] The tributary organization of some channels (Figure 6a) suggests that fluvial processes, possibly surface runoff, also modified these surfaces [Quantin et al., 2005], and likely occurred during the Noachian Period [Greeley and Guest, 1987; this study]. Multiple channels generally occur in the same region, are often interconnected, and gaps are present between different channel systems (Figure 3). This clustering of channels suggests that the channeled and nonchanneled surfaces may be distinct layers that formed at different times in which the environmental conditions at the time of surface layer exposure were different. This variable preservation of channels may imply that there were distinct periods of time when fluvial activity was common, and periods when the surface was relatively dry. Alternatively, there may have been dissimilarities in the preservation of observed channels, and the evidence for channels in some regions may have been removed by erosion. Finally, it is also possible that channeled versus unchanneled surfaces reflect differing degrees of water accumulation, melting, or storage in surface snow or underground aquifers. In addition, the majority of channels occur in the revised Npld subunit. If channels were once pervasive in the revised Nplr subunit, any evidence may have been removed by erosion or buried, as this low-lying area was infilled.

[30] It is likely that volcanic activity and the tectonic compression that produced wrinkle ridges ceased before this fluvial activity occurred. Channels crosscut wrinkle ridges, yet there is little indication that compressional features transect channels (Figure 13). This superposition relationship and the lack of clear volcanic constructs or flow fronts suggests that Arabia Terra was dominated by volcanic activity, and then volcanism ceased before fluvial processes modified the surface. It is not clear over what time scales this transition occurred, or how long after volcanic and tectonic activity fluvial activity began. There is no evidence that these environments existed simultaneously, or that additional volcanic activity occurred after the period of fluvial activity.

Figure 13.

Cross-cutting relationships between wrinkle ridges and channels. (a) Portion of THEMIS visible image V18172017 (18 m per pixel resolution) centered at 19.8°E, 11.5°N. (b) Portion of THEMIS daytime infrared mosaic (100 m per pixel resolution) centered at 30.8°E, 7.0°N. White arrows indicate cross-cutting relationships between wrinkle ridges and channels that indicate a younger relative age for the channeling events.

4.2. Intermediate

[31] After volcanic and fluvial activity ceased, likely in the Late Noachian to Hesperian Period [Greeley and Guest, 1987; Fassett and Head, 2008], the formation of the intracrater mound unit, significant erosion of surface materials, and the formation of aeolian bed forms that are now inactive likely occurred. The absolute age of their formation is not well constrained; the units have few craters and are too small in area to provide reliable ages using crater density methods. This material formed some time after the volcanic and fluvial activity ceased, but there is little evidence that these processes are active under current Martian atmospheric conditions. The relative ages of these events is uncertain, but are discussed in the most likely order of occurrence, beginning with the oldest event. All geologic processes associated with this intermediate age suggest a transition from the previously discussed fluvial environment to a predominately aeolian environment, including the presence of strong winds resulting in the erosion of a considerable amount of material. In addition, there is evidence for repeated periods of moisture, possibly from surface frost or paleopolar processes, causing the induration and cementation of layered materials and bed forms.

4.2.1. Intracrater Mound

[32] Following the volcanic and channel-forming period, there was the deposition and consolidation of the intracrater mound unit. Although poor crater density statistics precludes the interpretation of a surface age, the interior mound material is likely younger than the plains material surface on the basis of observed superposition relationships. This unit consists of fine laminations at the resolution of MOC and HiRISE images (Figure 10), suggesting repeated cycles of deposition and cementation. All mound material erodes into fine flutes and yardangs, indicating significant modification by erosion (Figure 8), and is suggestive of a weakly indurated material. The presence of yardangs and the lack of boulders at the base of the mounds also suggest a fine-grained (e.g., dust, silt, or sand) precursor material and a sedimentary origin [Malin and Edgett, 2000]. The association with unconsolidated intracrater material (sand) suggests that the intracrater mound unit is sufficiently indurated to withstand abrasion by wind, and may be similar to other likely indurated deposits found on Mars, such as white rock [e.g., Ward, 1979; Williams and Zimbelman, 1994; Ruff et al., 2001] and the Medusae Fossae Formation [e.g., Scott and Tanaka, 1986; Hynek et al., 2003; Watters et al., 2007]. The high TES albedo (0.25–0.27) and relatively low DCI (0.93–0.96) is consistent with at least a thin layer of unconsolidated dust and suggests that this material is not currently eroding, but instead is accumulating a layer of air fall dust. The thermal inertia of the mounds may be 400 to 435 J m−2 K−1 s−1/2, much lower than bedrock, and suggests a less consolidated material, such as an ash flow tuff or weakly lithified dust.

[33] Although these materials primarily occur within craters, there is material north of Henry crater that has similar erosional features (Figure 14). These eroded materials are remnants of a previously more extensive layer, possibly the same layer as the intracrater mound materials. Material north of Henry crater has a jagged erosional boundary, smooth surface, and knobby outcrops are present. These materials do not occur in local topographic depressions and are evidence that the emplacement of this material occurred over a broader area than just inside craters, and may have occurred throughout the entire study region. If emplacement occurred in regions other than craters, then this observation constrains possible formation mechanisms and increases the amount of deposition and erosion that took place in this region in the Martian past. On the basis of superposition relationships, Malin and Edgett [2000] and Greeley and Guest [1987] suggest that the intracrater mound material in Henry crater is Noachian in age.

Figure 14.

Possible intracrater material remnants. Material inside and outside of craters that have a similar erosional morphology, suggesting a more extensive intracrater material unit. (a) Portion of THEMIS visible image V06041009 (18 m per pixel resolution) located northwest of Henry crater, centered at 21.4°E, 13.2°N. (b) Portion of THEMIS visible image V10422010 (18 m per pixel resolution) located along the northern crater wall of Henry crater, centered at 23.5°E, 12.0°N.

4.2.2. Inactive Aeolian Bed Forms

[34] Aeolian bed forms with an albedo similar to the surrounding surface are observed in many channels and topographic depressions and alongside the base of mesas (Figure 7). These bed forms are interpreted to be transverse aeolian ridges, and likely formed by winds channelized in topographic depressions [Malin and Edgett, 2001; Wilson and Zimbelman, 2004]. The albedo is similar to the surrounding surface, suggesting that these bed forms are no longer active and are possibly covered in a mantle of dust. In addition, these bed forms have grooves eroded into their flanks, suggesting that these bed forms may be cemented or indurated and are resistive to erosion (Figures 4 and 15). They may have ceased to be active because the wind regime has changed or because they have been cemented. These bed forms indicate a previous aeolian environment in which mobile material was transported and deposited, but the climate has since changed such that bed form material is no longer active. Because these bed forms occur in channels that were presumably dry, this aeolian environment likely existed after fluvial activity ceased. The erosion of the layers discussed above could be a source for this material.

Figure 15.

Eroded bed forms. Portion of narrow angle MOC image M0807684, centered at 32.89°E, 5.05°N, illustrating possible eroded grooves in bed forms.

4.3. Modern

[35] Following the formation and erosion of the intracrater mound unit, Mars transitioned into its current climate and atmospheric circulation pattern, which is dominated by aeolian activity. The intracrater material, wind streaks, and current dust mantle imply a dry environment, as there is little indication for the presence of moisture, induration, or cementation in these materials. It is likely that the upper surface material is currently active, but may have originally formed thousands to millions of years in the past.

4.3.1. Intracrater Material

[36] The low-albedo and gradational boundaries suggest that the intracrater material unit consists of unconsolidated sand that has been distributed by wind. In addition, the thermal inertia (140–460 J m−2 K−1 s−1/2) indicates a particle size of fine sand to granules (particle diameter of 200 μm and greater). Deposits with higher thermal inertia values have resistive outcrops present as well, and the subpixel mixing of these resistive outcrops and sand is likely resulting in the large range in thermal inertia values. Grains as large as 1–2 mm have been observed to move in the current Martian climate at the Gusev landing site [Greeley et al., 2005], and therefore it is possible that wind gusts capable of moving similar sized grains occur in Arabia Terra as well. The low albedo implies little to no dust on these surfaces [Wells et al., 1984], and indicates that these bed forms are currently or recently active. Any bed form activity would remobilize and remove dust that settles on them, or incorporate this dust into the sand sheet. The proximity of these unconsolidated materials to raised features that exhibit knobs or outlying structures indicative of erosion, such as intracrater mounds or central crater peaks, suggests that local material serves as a partial source for the sand or as a topographic trap. The unconsolidated intracrater material may have originated from the erosion of either the surface beneath the dust mantle or intracrater mound deposits, but it is equally plausible that this unconsolidated sand is not derived from a local source. Thus in Arabia Terra, aeolian transport is a currently active process, and the intracrater material is evidence for a current aeolian dominated environment and the presence of seasonal winds that are strong enough to mobilize sand-sized grains.

4.3.2. Wind Streaks

[37] The thermal inertia of the wind streaks (TES is 120–170 J m−2 K−1 s−1/2; THEMIS is 115–330 J m−2 K−1 s−1/2) corresponds to a particle diameter of 15 μm to 1.5 mm (silt to very coarse sand). This wide range of thermal inertia values and implied particle sizes is likely indicating a mixture of materials. Since the majority of TES and THEMIS thermal inertia values are similar and the wind streak material covers several TES pixels, the TES bolometric thermal inertia values are likely representative of the actual particle size of the bulk of this material (20 μm to 80 μm). While there may be course sand mixed at the subpixel level on these surfaces, the higher thermal inertia values derived from THEMIS are likely due to crater rims, ridges, and other higher inertia features that are below the resolution of TES raising the thermal inertia. The majority of TES and THEMIS thermal inertia values are similar suggesting that the material is fairly homogenous at 100-m scales, and thick enough (several centimeters to a meter) that many small, high thermal inertia features, such as crater rims and rocks, are not present on the surface. This may be a function of a currently active process that is mobilizing material several centimeters thick, masking any underlying structure. The broad inferred particle size range may also represent variations in the thickness of the aeolian material. The sand-sized particles, the origin of wind streaks at the intracrater material, and the implied wind direction are evidence that these wind streaks consist of materials that originated from the unconsolidated intracrater materials. This interpretation is consistent with wind streaks formed by the deflation of dark material inside craters that has been transported and then redeposited outside the craters (Type II wind streaks) [e.g., Sagan et al., 1972, 1973; Arvidson, 1974; Veverka et al., 1976; Thomas and Veverka, 1979; Pelkey et al., 2001].

[38] The majority of streaks have an orientation of 180° to 220° and 300° to 330°, although wind streak orientations range from 30° to 330° in this study region. GCM results indicate a seasonal variation in wind direction, with stronger winds from the northwest in the northern fall and northern winter and weaker winds from the southeast in northern spring and northern summer [Fenton and Richardson, 2001]. The stronger wind direction is not consistent with the majority of wind streak orientations, but there are streaks present that are the product of winds from the northwest. There is one wind streak with minor streaks oriented toward the northwest, which is perpendicular to the prevailing wind streak direction (Figure 12; white arrows indicate minor streaks), and is consistent with a seasonal variation in the prevailing wind direction [Fenton and Richardson, 2001]. This wind streak is the largest observed, and the wind streak material may be sufficiently thick to act as the source material for the minor streaks.

4.3.3. Dust Mantle

[39] The thermal inertia, albedo, and DCI all indicate the presence of unconsolidated fine-grained material. The homogeneous nature of thermal inertia and albedo values in this area at 100-m scales indicates that the dust layer must have a minimum thickness sufficient to mask the thermal signature from underlying material (a few 10s of centimeters), but is still thin enough to be able to distinguish small craters (10 m in diameter) and surface textures at the resolution of MOC and HiRISE visible imagery. In addition, ridges, lava flows, and other topographically high features are observed in THEMIS infrared images (100 m per pixel resolution), yet the thermal inertia of these features is that of dust. This observation implies that the dust mantle must be sufficiently thick to blanket this material on both the top and sides and obscure any signature from the underlying material. Assuming a depth:diameter ratio of 1:5 to 1:3 for simple craters [Melosh, 1989, p. 126], the smallest craters observed are 2–3 m deep, thus more than a few meters of dust would obscure these features. In MOC images, surfaces within a few 10s of meters of one another have dissimilar degrees of smoothing of their surface (Figures 4 and 16). This could be caused by differences in the degree of erosion affecting the surface, but it is more likely that this variation in smoothness represents variable thicknesses in the dust layer on spatial scales of 10s of meters. These observations using high-resolution imagery indicate that although unconsolidated dust is present on the surface, it cannot be more than 1–2 m thick, consistent with previous Viking era estimations [e.g., Christensen, 1986a], and is likely thinner in many localities.

Figure 16.

Variable dust thickness. Portion of narrow angle MOC image E1401210, centered at 24.15°E, 2.5°N, illustrating differing degrees of smoothness of adjacent surfaces suggestive of variable dust thickness.

5. Interpretation and Implications

5.1. Origin and Evolution of Intracrater Mound Material

5.1.1. Origin of Layered Intracrater Mound Material

[40] There are at least three possible scenarios for the origin of the intracrater mound material, which are as follows: (1) paleolake deposits, (2) deposition of volcanic air fall ash, and (3) lithification of air fall dust. Malin and Edgett [2000] propose that many of the layered materials found inside craters formed by lacustrine processes. A lacustrian or paleolake environment would explain the horizontal, thin layers observed in all the intracrater mound materials and the presumably fine-grained precursor material. The crater also provides an enclosed basin that could pond water [e.g., Cabrol and Grin, 1999, 2003; Malin and Edgett, 2000; Irwin et al., 2004]. However, many features expected for a paleolake environment are not observed in Arabia Terra [e.g., Cabrol and Grin, 1999, 2003; Malin and Edgett, 2000; Irwin et al., 2004]. There is no obvious source channel to deposit lacustrian sediments into the craters. In all cases, the crater rims are preserved with no evidence for a breach through the crater rim. In Henry crater and an unnamed crater, centered at 23.5°E, 3.2°S, channels are present along portions of the crater walls (Figure 6b), but the other three crater mound materials show no evidence for fluvial processes. Also in Henry crater and an unnamed crater, centered at 23.5°E, 3.2°S, the height of the mound material closely approaches that of the crater rim. These mounds may have once been higher than the crater rim, which cannot occur in a paleolake environment. Also none of the craters with intracrater mound material have evidence for shorelines, former lake levels, or deltaic deposits. In addition, to the northwest of Henry crater are materials that may be of the same origin as the intracrater mound material (Figure 14). The possible presence of mound material outside of craters suggests a more extensive deposit, and this material is not contained within a topographical enclosure capable of ponding water. Because of the lack of channels breaching through the crater rims and no landforms suggestive of paleolakes, lacustrine processes is not a favored mechanism for the formation of the intracrater mound unit.

[41] Deposition of volcanic ash from an explosive eruption is a possible formation mechanism for the intracrater mound unit. Tharsis Montes may have experienced explosive eruptions in its history [e.g., Edgett, 1997; Mouginis-Mark, 2002; Hynek et al., 2003], and is a possible source for volcanic ash in this region. Plume heights have been modeled as high as 100 km under current Martian conditions [e.g., Wilson and Head, 1994; Hort and Weitz, 2001], which would result in the lateral transportation of volcanic materials globally [e.g., Hynek et al., 2003] and suggests that volcanic material from the Tharsis Montes volcanoes was indeed deposited in eastern Arabia Terra. However, Glaze and Baloga [2002] report that previous work overestimated volcanic plume heights on Mars, and that a convective rise model is only valid to a height of ∼10 km, making global-scale distribution of ash nearly impossible. If one assumes that plume heights of 10 km were attained, then volcanic material from Tharsis could have been transported up to ∼3000 km [Hynek et al., 2003] and potentially only a few hundred meters [Glaze and Baloga, 2002], both of which are much less than the roughly 8700 km distance of this study region from Tharsis Montes. Thus it is not likely that significant amounts of ash from explosive volcanic eruptions from Tharsis were directly deposited in this study region, and is therefore not a preferred primary formation mechanism.

[42] Finally, the intracrater mound materials could consist of many layers of consolidated air fall dust. GCM results indicate that Arabia Terra is a low wind shear region under a variety of atmospheric conditions and obliquities [e.g., Haberle et al., 2003; Kahre et al., 2006; Haberle et al., 2006], and has likely been a region with atmospheric conditions favorable for dust accumulation throughout much of Martian history. Assuming (1) an annual dust deposition rate of 1–7 μm/a, on the basis of estimates of average dust deposition from planet-encircling dust events (∼7 μm during the 1977 dust event [Pollack et al., 1979] and dust mass loading a factor of 6–8 lower during the 1999 dust event [Cantor et al., 2001]); (2) an occurrence of planet-encircling dust events every 1 in 3 years [Cantor et al., 2001; Martin and Zurek, 1993; Zurek and Martin, 1993]; and (3) a compaction by half of air fall dust during induration, it would take 109 years to deposit 4200 m of dust (2100 m of the intracrater mound unit after induration).

[43] The mound material is layered at the resolution of MOC and HiRISE images, suggesting the presence of hundreds to thousands of layers, and thus a repeated process of dust deposition and compaction. It is possible that the repeated deposition and cementation of air fall dust is driven by changes in Martian obliquity. Under the current obliquity, conditions are favorable for dust deposition and accumulation in this region [Haberle et al., 2003]. Under higher obliquities, dust deposition would continue [e.g., Pollack et al., 1979; Haberle et al., 2003], CO2/H2O ice would sublimate onto (and into) the surface at these latitudes [Mellon and Jakosky, 1995], and could provide a means to cement the dust. The accumulation and cementation of air fall dust discussed here is similar to the proposed formation mechanism of the polar layered deposits, which are generally attributed to cyclic changes in climatic conditions and dust deposition rates due to changes in orbital obliquity parameters [Blasius et al., 1982; Cutts and Lewis, 1982]. This process would then repeat itself to produce the hundreds to thousands of layers observed in these materials. This proposed mechanism of repeated dust deposition and cementation of dust by ice under higher-obliquity conditions can explain the following: (1) the intracrater mound source material (dust), (2) the cementing agent for the unconsolidated dust, and (3) the layered nature of these mound deposits. Although repeated dust deposition and cementation may be the primary mechanism by which these layers were formed, it is also likely that volcanic ash has also contributed to their formation.

[44] This argument for the existence of ice sublimation as a means for dust cementation in Arabia Terra is strengthened by data collected by the GRS/NS/HEND instrument suite [e.g., Boynton et al., 2002; Feldman et al., 2002; Mitrofanov et al., 2002, Feldman et al., 2004; Mitrofanov et al., 2004; Ivanov et al., 2005; Jakosky et al., 2005]. GRS and NS data have shown hydrogen enrichments interpreted as having up to ∼10% water content in the uppermost surface layer in Arabia Terra [Feldman et al., 2004; Mitrofanov et al., 2004; Ivanov et al., 2005; Jakosky et al., 2005], and this presumed water may have been emplaced by a process similar to that described here. Tanaka [2000] suggested that portions of Arabia Terra are relic dust and ice deposits formed by a process similar to that proposed in this work. Although the area identified by Tanaka [2000] is northwest of this study region, the intracrater mound material may be outliers of this broader unit.

[45] Using Viking data, Schultz and Lutz [1988] proposed that the deposits in Arabia Terra were the result of paleopolar processes on the basis of their similarity to current polar layered deposits, which is a variation of the dust cementation hypothesis described in this work. Schultz and Lutz [1988] hypothesized that polar wandering occurred on Mars as a result of changes in the location of the planet's principal moment of inertia due to flood lava volcanism and the formation of Tharsis. They proposed that layered deposits in Arabia Terra are relics from a previous pole location. Tanaka [2000], however, argues against the polar wander hypothesis of Schultz and Lutz [1988]. In the polar wander scenario, the Medusae Fossae Formation and Arabia Terra deposits formed as antipodal, paleopolar deposits that predate the formation of Tharsis [Schultz and Lutz, 1988]. However, the Arabia deposits are generally cited as Noachian in age [Greeley and Guest, 1987], which is consistent with this work. The Medusae Fossae Formation is dated Middle to Late Amazonian in age [Scott and Tanaka, 1986; Greeley and Guest, 1987]. Thus the polar wandering model is not supported by the ages of these surfaces. In addition, the growth of Tharsis may have happened gradually over a larger geologic time span than suggested by Schultz and Lutz [1988], further complicating their hypothesis [Tanaka, 2000]. Thus, polar wander may not have been extreme enough to form the finely layered intracrater mound materials in Arabia Terra.

5.1.2. Erosion of Materials

5.1.2.1. Current Intracrater Mound Volume

[46] On the basis of the current exposures of intracrater mound material, ∼21,000 km3 of indurated material was deposited in this region. If this material was a uniform layer over the planet, it would be ∼0.6 m thick, and is also consistent with a ∼3 m thick layer of unconsolidated dust over the low thermal inertia regions located in Tharsis Montes, Arabia Terra, and Elysium Planitia. The similar morphology of materials in craters, the close proximity of craters containing intracrater mound materials, and the identification of similar materials outside craters suggest that the intracrater mound unit was once more extensive. It is also possible that more than one episode of erosion and deposition of intracrater mound material occurred, resulting in an unidentified unconformity and an underestimation of the amount of intracrater mound material. Thus, the estimated thickness is likely the minimum amount of dust deposited in these regions and then consolidated in the Martian past, and implies that dust deposition and transport may have played a significant role in the atmospheric circulation and climate in the late Noachian and Hesperian Period.

5.1.2.2. Intracrater Plains Erosion

[47] In the intercrater plains, there are locations where the overlying material has been eroded, exposing topographically, and likely stratigraphically, lower materials and leaving behind remnants of this eroded surface (Figure 5). This terrain has indentions and exposures along the erosional boundary forming a rough edge, and has a thickness of ∼60 m. Although the intracrater mound material has undergone erosion as well and potentially at the same time, it is not likely that the same process formed these two materials. The thickness of this single layer is only ∼60 m thick, whereas the intracrater mound material has hundreds to thousands of individual layers and the total exposure can be over 2000 m thick. In addition, in THEMIS visible images, this material shows no evidence of the fluted or yardang erosional pattern characteristic of the interior mound material, and therefore it is more likely that this terrain has a fundamentally different origin.

[48] This material is most likely either a lava flow or sedimentary material that has been partially eroded, and it is difficult to distinguish between these two hypotheses with the existing data available. In addition, isolated mesas and knobs 60 m in height are in close proximity to this terrain, and both the larger single layer exposure and the mesas and knobs have a similar erosional expression (Figure 5). The comparable thickness, erosional expression, and close proximity of these materials suggest that these knobs and mesas and the larger single layer outcrop were once a continuous unit, and also implies that this material may have been a more extensive deposit. Few remnants of this terrain are observed in the study area, and there are no superposition or crosscutting relationships between this terrain, channels, and the intracrater mound material to determine its relative age. However, there are no channels on this unit, and it is possible that the erosion that removed this terrain also removed the channels themselves. If this is the case, then the erosion of this material occurred after the fluvial/channeling activity. Because of the small number of remnants of this terrain, it is challenging to accurately estimate the original extent of this terrain. Incorporating the existing remnants and surrounding materials with a similar surface texture, at least 23,000 km3 of material has been eroded. The erosion of this terrain may be a partial source material for the low-albedo intracrater material and consolidated bed forms observed throughout this region. Alternatively, this material may have eroded to sand-sized grains that have since been transported out of the study region.

5.1.2.3. Erosional Agent

[49] Although the intracrater mound material and previously discussed volcanic and/or sedimentary terrain may have been eroded at different times, the potential process eroding these surfaces is similar. The erosion of the intracrater mound material and the volcanic and/or sedimentary terrain could have been produced by at least the following three different processes: (1) fluvial, (2) glacial, and (3) aeolian. Water as surface runoff tends to form channels, but the region surrounding the eroded volcanic and/or sedimentary terrain is devoid of any channels, and if some of the overlying material is preserved it is expected that channels would have been preserved on this material. In addition, the erosional boundary is relatively uniform in nature suggestive of a broad-scale erosional process rather than erosion by channelized water. There are some channels present near the intracrater mound materials, but these channels are not present in all instances and do not cut the mound materials. Thus, the erosion of the volcanic and/or sedimentary terrain and the intracrater mound materials was not likely caused by fluvial processes. Ice is a possible agent for erosion. However, the erosion produced by ice leaves morphologic features, such as moraines, kettles, and drumlins, which are not observed anywhere in this study region.

[50] Erosion by wind is the most likely scenario for the erosion of both the volcanic and/or sedimentary terrain and the intracrater mound material. The uniform degree of erosion across the volcanic and/or sedimentary terrain is characteristic of erosion by a regional process such as wind. In addition, yardangs commonly form in aeolian environments, and thus the erosional morphology of the intracrater mound materials is suggestive of erosion by wind. For wind to erode a resistive material such as lava or indurated sedimentary deposits, the winds must have been of sufficient velocity (∼50–100 m s−1 [White, 1979]) to saltate or suspend particles and have been sustained over long periods of time (105–106 years [Greeley et al., 1984, 1985]). These conditions are not predicted by GCM models, which indicate a low wind shear environment under both current conditions and different obliquities and do not predict winds strong enough to erode resistive rock units at any time in Arabia Terra history [e.g., Haberle et al., 2003]. However, the occurrence of eroded materials is strong evidence that significant erosion has occurred in this region and that Arabia Terra has not always been a region of active dust deposition or with low wind shear stresses as GCM models predict.

5.1.3. Relation to Other Intracrater Deposits

[51] West of this study area, many craters have intracrater materials and dark sand deposits [e.g., Arvidson, 1974; Christensen, 1983; Thomas, 1984; Edgett and Malin, 2000; Edgett, 2002; Wyatt et al., 2003]. These intracrater materials are often associated with knobby terrain and mesas suggesting the erosion of a previously more extensive deposit (Figures 4 and 17b). In addition there are massive layers in the floor of many craters that have been exposed by erosion. Some of the erosional features are similar to intracrater features in this study area, particularly along crater walls (Figures 4 and 17). This similarity suggests that these materials may have been part of the same or a related deposit, as the intracrater mound unit in this study region. In western Arabia Terra, mounds are rarely present, suggesting a more advanced erosional state of the same material as the intracrater mound unit. This is consistent with GCM results indicating stronger winds in western Arabia Terra than eastern Arabia Terra [e.g., Fenton and Richardson, 2001] that promote more efficient aeolian erosion. If these intracrater materials are related, then the intracrater mound materials could have extended 2,000,000 km2, which would include the intracrater mound material both in and west of the study region and assumes that the mound material was deposited uniformly over this area. The majority of this material has been eroded since its deposition.

Figure 17.

Similar erosional morphologies found in eastern and western Arabia Terra. (a) Portion of THEMIS visible image V06565021 (18 m per pixel resolution) in eastern Arabia Terra, centered at 28.2°E, 2.3°N. (b) Portion of THEMIS visible image V16201011 (18 m per pixel resolution) in western Arabia Terra, centered at 356°4 E, 5.4°N.

[52] The occurrence of similar materials to the intracrater mound unit in western Arabia Terra and other localities, such as white rock [e.g., Ward, 1979; Williams and Zimbelman, 1994; Ruff et al., 2001] and the Medusae Fossae Formation [e.g., Scott and Tanaka, 1986; Hynek et al., 2003; Watters et al., 2007], suggests a global formation of materials like the intracrater mound unit. The formation of fine-scale layers may be related to repeated events of dust deposition and cementation on Mars, possibly due to obliquity changes. Similarly, the formation mechanism of the polar-layered deposits is generally attributed to cyclic changes in climatic conditions and dust deposition rates due to changes in orbital obliquity parameters [Blasius et al., 1982; Cutts and Lewis, 1982]. Thus, the intracrater mound unit, crater deposits in western Arabia Terra, and layered materials found globally [e. g., Tanaka, 2000] may represent the equatorial equivalent of the polar layered deposits. This relationship implies that dust deposition has been an important process globally and likely throughout much of Martian history.

5.2. Martian Dust Cycle

[53] Recent GCM results suggest that dust devils play an important role in the removal and deposition of dust [Basu et al., 2004; Kahre et al., 2006; Haberle et al., 2006], and are likely occurring in low thermal inertia regions such as Arabia Terra. Because of the 10s of centimeters to a meter thick dust mantle, the upper few microns of material removed by a dust devil [Greeley et al., 2003] would not noticeably change the albedo of these surfaces. Therefore dust devil tracks are not expected, nor have they been observed in this study area. The global average annual nine-micron dust opacity measured by TES is 0.15–0.2 [e. g., Smith, 2004], indicating that a continual amount of dust is present in the atmosphere. To produce model results consistent with this background dust level, dust devils are a necessary component of the dust cycle [Kahre et al., 2006; Haberle et al., 2006]. The inclusion of dust devils in the GCM models also results in slight dust deflation rates in the low thermal inertia regions, including eastern Arabia Terra [Kahre et al., 2006]. However, atmospheric observations using Viking data indicate that dust is being deposited in regions of low thermal inertia after planet-encircling dust events [e.g., Pleskot and Miner, 1981; Christensen, 1982, 1988]. In addition, to produce the low thermal inertia values currently modeled, dust is either currently accumulating in these regions or has accumulated in the very recent past. Therefore, in years without major dust events, dust devils are occuring in low thermal inertia regions, and are an important process to maintain the background levels of dust present in the atmosphere [Kahre et al., 2006; Haberle et al., 2006]. It is also likely that dust is currently deposited in these regions, but only in years with planet-encircling dust events. These observations also suggest that the thickness of the dust mantle may not be currently increasing. Instead, deposition in years with dust events and dust removal in nondust event years may result in an overall net balance of dust thickness in the low thermal inertia regions.

[54] The observation of different surface morphologies and evidence for significant erosion, indicates that there has been a change in geologic environments over time in this region, and that this region has not been a continual region of dust accumulation throughout Martian history. The erosional morphologic features of the intracrater mound material implies that significant amounts of erosion must have occurred after the deposition of these layered deposits and before the deposition of unconsolidated dust characteristic of this region today. However, this amount of erosion is inconsistent with GCM results, which indicate a low wind shear environment under both current conditions and different obliquities and does not predict winds strong enough to erode consolidated material in Arabia Terra [e.g., Haberle et al., 2003]. Likely, atmospheric conditions not accounted for in the current GCM models are responsible for this discrepancy. On possibility is that a thicker Martian atmosphere caused by an increase in water content may be present at higher obliquities [e.g., Mellon and Jakosky, 1995; Jakosky et al., 2005], resulting in higher wind shears capable of eroding resistive materials.

6. Conclusions

[55] Eastern Arabia Terra, Mars is currently mantled by a few cm to a meter of dust, and has been interpreted as a location favorable for dust accumulation. However, analysis of multiple data sets has uncovered a complex geologic history that has broad implications for the global climate history of Mars. On the basis of the analysis of these data, we conclude the following:

[56] 1. In early Martian history, volcanic and fluvial processes modified the surface of Arabia Terra. Since then, this region has been dominated by aeolian activity, which has resulted in the deposition of layered materials as thick as 2100 m followed by significant amounts of erosion. Major changes in the Martian climate had to produce the environmental transitions implied by the morphologic features observed. This region has transitioned from a fluvial environment to an aeolian environment, from a primarily depositional environment to a erosional environment, and then to the low wind shear environment that dominates this region today.

[57] 2. The presence of morphologic features indicating that a significant amount of erosion has occurred is strong evidence that Arabia Terra has not always been a region of active dust deposition or with low wind shear stresses. Current GCM results do not predict significant amounts of erosion in the eastern Arabia Terra region at current or higher obliquities [Haberle et al., 2003]. A thicker Martian atmosphere in the past may be required to produce winds strong enough to erode indurated materials, such as lava flows or intracrater mound materials.

[58] 3. The eastern Arabia Terra region is likely an area currently experiencing dust deposition and accumulation in years with planet-encircling dust events [e.g., Pleskot and Miner, 1981; Christensen, 1982, 1988]. In addition, current GCM results predict that Arabia Terra undergoes slight dust deflation in years without major dust events [Kahre et al., 2006]. These observations suggest that the thickness of the dust mantle may not be currently increasing. Instead, deposition in years with dust events and dust removal in nondust event years may result in an overall net balance of dust thickness in the low thermal inertia regions.

[59] 4. The layered intracrater mound unit may have been formed by the repeated deposition and cementation of air fall dust. With increasing Martian obliquities, H2O and CO2 ice may have sublimated onto the surface at these latitudes, cementing any unconsolidated air fall dust present. This process would then repeat until hundreds of layers are formed, and it is possible that ice is still present in the interior of these mound materials.

[60] 5. The layered intracrater mound units observed in Arabia Terra are likely related to similar deposits found west of this study region. In addition, the formation mechanism for these deposits may be similar to many layered terrains found globally on Mars, and may represent the equatorial equivalent of the polar layered deposits. These deposits therefore may hold key information to the climate history on Mars.

Acknowledgments

[61] We would like to thank two anonymous reviewers whose comments greatly improved the clarity, overall scientific quality, and presentation of this work. We would like to thank Michael Kraft and Joshua Bandfield for reading early versions of this work; James Skinner for helping with crater counting; Hugh Kieffer for providing the thermal model used to derive THEMIS thermal inertia values; and Noel Gorelick, Christopher Edwards, Ryan Luk, Keith Nowicki, and the USGS Flagstaff ISIS software group for developing, and continuing to improve, the THEMIS mosaicing software used to create many of the figures in this manuscript. We also want to thank the THEMIS mission team, Kelly Bender, Jonathan Hill, Greg Mehall, and Kim Murray for acquiring, processing, and calibrating the THEMIS data set and the MOC and HiRISE teams for excellent images. This work was funded by the Mars Odyssey Science Office.