SEARCH

SEARCH BY CITATION

Keywords:

  • Mars;
  • crust formation;
  • spectroscopy

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[1] The earliest formed crust on a single plate planet such as Mars should be preserved, deeply buried under subsequent surface materials. Mars' extensive cratering history would have fractured and disrupted the upper layers of this ancient crust. Large impacts occurring late in Martian geologic history would have excavated and exposed this deeply buried material. We report the compositional analysis of unaltered mafic Martian crater central peaks with high-resolution spectral data that was used to characterize the presence, distribution and composition of mafic mineralogy. Reflectance spectra of mafic outcrops are modeled with the Modified Gaussian Model (MGM) to determine cation composition of olivine and pyroxene mineral deposits. Observations show that central peaks with unaltered mafic units are only observed in four general regions of Mars. Each mafic unit exhibits spectrally unmixed outcrops of olivine or pyroxene, indicating dunite and pyroxenite dominated compositions instead of basaltic composition common throughout much of the planet. Compositional analysis shows a wide range of olivine Fo# ranging from Fo60 to Fo5. This variation is best explained by a high degree of fractionation in a slowly cooling, differentiating magma body. Pyroxene analysis shows that all the sites in the Southern Highlands are consistent with moderately Fe-rich, low-Ca pyroxene. Mineral segregation in the ancient crust could be caused by cumulate crystallization and settling in a large, potentially global, lava lake or near surface plutons driven by a hypothesized early Martian mantle overturn.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[2] The initial formation of a planetary crust is a critical stage in a planet's evolution. It can contain evidence of the last stages of planetary formation and gives a starting point for all subsequent crustal modification. The nature of this early crust is difficult to address on Earth as plate tectonics and other active geologic processes have destroyed, recycled or deeply buried and altered most of the ancient rocks. The lunar crust offers a well-studied primary crust [Taylor, 1992], thought to be formed by a plagioclase floatation process from a magma ocean [Wood et al., 1970; Warren, 1990]. However, the small size and unique composition of the Moon makes it difficult to extrapolate this process to larger, wetter worlds. In this sense, Mars provides a valuable analog for the conditions of ancient terrestrial crusts with single-plate style of tectonics that would have preserved these crustal materials. Subsequently to ancient crust formation, Martian volcanism, alteration and impact cratering would obscure the ancient crust from surface observations. High-resolution observations from the Mars Reconnaissance Orbiter (MRO) can observe small, relatively compositionally pure outcrops of unaltered, igneous rocks that are exposed in crater central uplift structures as megablock fragments, which may represent this ancient crust. This ancient crust of Mars is the key to understanding the formation and early evolution of the planet.

[3] The igneous composition of the current Martian surface has been studied by a number of investigations that use data from orbital, landed and laboratory instrumentation. With the possible exception of the Martian meteorite ALH84001 [e.g., Mittlefehldt, 1994; Lapen et al., 2010] described below, all measurements of Martian lithologies are partially or entirely reprocessed after the planetary crustal formation. If the crustal blocks reported here are from this initial formational period, as we hypothesize, then we cannot directly compare them to previously reported results. However, the legacy of Martian igneous lithology observations detailed below illustrates the diversity of Martian volcanic processes and sets up the context for known Martian compositions.

[4] Global surface mineralogy was analyzed with the Thermal Emission Spectrometer (TES) [Christensen et al., 1992] which mapped the presence and distribution of mafic minerals and igneous lithologies [e.g., Bandfield et al., 2000; Wyatt and McSween, 2002; Rogers and Christensen, 2007; Koeppen and Hamilton, 2008]. Initial observations divided the primary crustal lithologies into a basaltic Surface Type 1 with 50% plagioclase feldspar, 25% clinopyroxene and 15% sheet silicates and an evolved Surface Type 2 with 35% plagioclase feldspar, 25% glass, 15% sheet silicates and 10% clinopyroxene with a hemispherical division between the types [Bandfield et al., 2000]. Surface Type 2 has since been suggested that it is the result of an alteration process [Wyatt and McSween, 2002; Ruff and Christensen, 2007]. Surface Type 1 has been further refined to show specific compositions unique to individual volcanoes and regions [Rogers and Christensen, 2007]. Thermal analysis of olivine exposures show that the mineral makes up approximately 10–20% of the aggregate surface abundance [Koeppen and Hamilton, 2008]. Olivine detections have seen a range of compositions with a concentration at intermediate to forsteritic compositions (Fo68-Fo53) [Koeppen and Hamilton, 2008; Hoefen et al., 2003] with some minor areas modeled with values as low as Fo39 and has high as Fo91 on a few basin rims. While TES provides the best available remote data set to determine global lithologies, the large, 3 × 6 km pixel resolution limits the ability to resolve crustal outcrops below the spatial resolution scale of this data set.

[5] Understanding of Martian surface mineralogy was further refined with visible and near infrared (VNIR) Observatoire pour l'Mineralogie, l'Eau, les Glaces, et l'Activite (OMEGA) observations (300 m to 5 km spatial resolution) mapping of iron-bearing igneous silicate mineralogy including high-calcium pyroxene (HCP), low-calcium pyroxene (LCP) and olivine [Mustard et al., 2005; Poulet et al., 2009]. The OMEGA observations confirmed and extended many of the TES results and greatly expand understanding of the aqueous alteration of the crust [Bibring et al., 2006]. Additional studies have shown compositional trends with time as younger terrains having a relative enrichment in high-calcium pyroxene (HCP) [Poulet et al., 2009; Skok et al., 2010] and a depletion in SiO2 [Baratoux et al., 2011] hinting at the thermal evolution of the planet. However, the resolutions still result in large spatial averages of surface compositions.

[6] Small scale surface compositions have been determined in select locations by in situ analysis of Mars igneous samples with instrumentation on the Spirit Mars Exploration Rover [Squyres et al., 2004]. An example is provided by the in situ analysis of Gusev basalts that have an estimated modal mineralogy of approximately 25% olivine, 30% pyroxene with the remaining as plagioclase and accessory minerals. In these basaltic rocks the olivine's cation composition was determined to be intermediate to mildly fayalitic (Fo52–45) from the alpha particle x-ray spectrometer (APXS) measurements and Fo60–35 based on MiniTES TIR observations [McSween et al., 2006]. Additional exploration of the Columbia Hills, located near the center of Gusev Crater [Squyres et al., 2006; Arvidson et al., 2008] yielded deposits interpreted as pyroclastics [Squyres et al., 2007] and materials consistent with an exposed hydrothermal system [Squyres et al., 2008; Ruff et al., 2011] suggesting a volcanic origin for the hills. An alternative explanation for the whole rock compositions is that they formed as a layered igneous intrusion with exposed harzburgite and olivine-norite and compositions ranging from ∼Fo65-Fo45 [Francis, 2011].

[7] The most detailed analysis of Martian crustal composition is provided by the study of Martian Meteorites [e.g., Mittlefehldt, 1994; Nyquist et al., 2001; McSween, 2002] These allow in-depth chemical analysis and a detailed petrographic history to be determined. Meteorite sampling has been skewed to only the most competent surface materials and therefore nearly all are relatively young, igneous rocks. The majority of the samples have been classified as basaltic Shergottites and have ages that range from 175 Ma to 475 Ma [Nyquist et al., 2001, and references therein], indicating relatively recent Martian volcanic activity, though other analyze techniques suggest these samples could be as old as ∼4.0 Ga [Bouvier et al., 2008]. However, the meteorites contain no contextual information, making it impossible to trace back the compositions to a particular region of Mars. Of particular interest is the meteorite, ALH84001, dated between 4.5 and 4.1 Gyr [Mittlefehldt, 1994; Nyquist et al., 2001; Lapen et al., 2010] depending on age determination method, it is the oldest sample of the Martian crust. The orthopyroxenite composition indicates the existence of early crustal processes capable of mineral segregation and may be a sample of a common early crustal component.

[8] Current understanding of the formation of Mars is determined from isotopic and petrologic observations from Martian meteorites [Bertka and Fei, 1997; Debaille et al., 2007, and references within] and geophysical/geochemical modeling of planetary evolution [Borg and Draper, 2003; Elkins-Tanton et al., 2005]. Isotopic studies show that the initial formation of Mars would progress rapidly with core formation having been largely completed 12.4 ± 4 Myr after solar system formation [Debaille et al., 2007] while mantle evolution of source regions would be mostly completed within 100 Myr [Debaille et al., 2007] (Figure 1). The details of this mantle evolution have been investigated with geophysical models. Magma ocean crystallization results have been predicted based a variety of magma ocean depths [Borg and Draper, 2003; Elkins-Tanton et al., 2005]. Due to high energy release from early planetary formation processes we focus on the deep ocean process. Elkins-Tanton et al. [2005]describe the results from an early magma ocean that would crystalize olivine and pyroxene cumulates, beginning with Mg-rich minerals and progressing to Fe-rich as the melt cools. The resulting density inversion would be corrected with a planetary scale mantle overturn. The volcanic activity resulting for the overturn event would be responsible for the formation of the bulk of the earliest crust of Mars. This overturn model predicts a global crustal dichotomy where shallow cumulate melting would lead to voluminous magma, low-Al, and high Fo#, while deep melts would lead to less magma, higher-Al and ∼Fo70 [Elkins-Tanton et al., 2005]. These modeled predictions can now be tested with high-resolution spectroscopic images of the Martian surface.

image

Figure 1. Planetary Formation Model. The best accepted model of Martian crustal formation combines our understanding of lunar formation, modeling of mineral crystallization and isotopic constraints from Martian meteorites. (a) Initial formation began with solar system condensation and accretion of meteoritic material. (b) Energy from impacts and differentiation would melt silicates forming a deep magma ocean and the segregation of metallic core. (c) Crystallization of the magma ocean would begin with Mg-enriched olivine and later pyroxenes that would be denser than the melt and forming a cumulate pile at the bottom of the magma ocean and result in an increasingly Fe-rich magma. (d) Crystallization would end with Fe-rich olivine and pyroxenes above a garnet layer stable at depths of 1050 km. The water content of the magma would suppress plagioclase formation. (e) The mineral crystallization sequence would create a density inversion that would be corrected with an overturn that is likely to be continuous with the crystallization of the Fe-enriched minerals. The displacement of the Mg-rich minerals toward spatially higher and lower pressure regions would result in decompression partial melting. The final stages of this process, occurring after the majority of the magma ocean has crystallized, would extrude on the ancient Martian surface and form the bulk of the Martian Crust. (Based onDebaille et al. [2009] and Elkins-Tanton et al. [2005]).

Download figure to PowerPoint

[9] We focus this study on compositional determinations of 23 central peak exposures with unaltered, mafic spectroscopic signatures to investigate the nature of the buried, ancient Martian crust and compositional diversity across the planet (Figure 2 and Table 1). Note that we define “unaltered” as spectral units that do not exhibit significant hydrous or hydroxylated phases at the spatial scale of the CRISM data. These observations are then integrated into a model for the formation of the ancient crust that can be used to refine and test future planetary formation models.

image

Figure 2. Map of Analyzed Sites. Twenty-three sites have unaltered mafic spectra that can be modeled with the MGM. The result of this selection process is a natural grouping in four general regions, North Argyre (NA), North Hellas (NH) Nili Fossae (NF) and Northern Plains (NP). Red marks have only olivine outcrops, Green marks have only pyroxene outcrops, Black marks have both olivine and pyroxene outcrops. (MOLA color DEM background).

Download figure to PowerPoint

Table 1. Details From Each of the Central Peak Structures Reported in This Studya
CodeNameCRISM IDLatitudeLongitudeDiameterOlivinePyroxene
  • a

    Column Key: Code: Central peak designation for this project. Letters refer to region, NA: North Argyre, NH: North Hellas, NF: Nili Fossae, NP: Northern Plains. Number is based on analysis order. Name: Name of crater where available, bold name indicates that the examined cater is an unnamed crater within the larger named crater in bold. CRISM_ID: Is a concatenation of CRISM FRT information as: Obs Year_Obs Day_Obs Hexidecimal ID. Longitude has positive values going east. Diameter is the average diameter of the observed crater. Olivine and Pyroxene are marked if that mineral has been modeled at that site.

NA1Alga Crater2007_167_6415−24.34−26.6518OP
NA2Ostrov Crater2010_099_17D02−26.51−28.0970OP
NA3Hale Crater2009_066_117BC−35.58−36.43130OP
NA4Ladon Basin2008_018_97FF−16.71−28.1422OP
NA5Kasimov Crater2008_186_B5BE−24.93−22.787O-
NA6 2009_042_10FE8−31.41−16.6637-P
NH1 2009_042_C2A5−25.7834.4239-P
NH2 2008_032_9BCF−24.2643.4722-P
NH3 2008_248_C554−18.7462.6350OP
NH4 2009_231_147FB−21.5344.7542OP
NH5 2011_173_1EB32−19.3264.0331-P
NH6 2011_176_1EBA0−31.4381.6320OP
NH7 2010_086_177F9−6.2393.6326-P
NH8 2007_286_82E8−15.7596.8735OP
NH9 2008_245_C4D0−31.29108.6773-P
NF1Hargraves Crater2008_185_B57320.7675.8162OP
NF2 2009_044_110B720.2269.4250OP
NP1Kunowsky Crater2008_013_961056.86−9.2362OP
NP2 2010_127_18A0652.6415.2425OP
NP3 2008_242_C41759.84135.8030O-
NP4 2008_232_C16F55.58139.6526O-
NP5Stokes Crater2008_155_ADA455.65171.3563OP
NP6 2010_093_17AA166.42144.0629O-

2. Analysis Techniques

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[10] To develop the database of exposed, unaltered crust on Mars for analysis, we considered all potential central peak exposure observations. The survey initially considered ∼150 sites included in the Crater Exposed Database compiled by Tornabene et al. [2010], which characterized central features in craters on the basis of High Resolution Imaging Science Experiment (HiRISE) [McEwen et al., 2007] and Context Imager (CTX) [Malin et al., 2007] morphological data. Of these 105 had full resolution targeted (FRT) observations from the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) instrument [Murchie et al., 2007]. FRTs of central peaks were first examined with spectral summary parameter maps [Pelkey et al., 2007] that quickly summarize where likely exposures of unaltered crust exist. To confirm that in fact alteration minerals (identified on the basis of 1.9 μm H2O combination tone) are absent and that crystalline igneous mafic minerals, olivine and/or pyroxene are present in bedrock exposures the areas highlighted in spectral summary parameters were examined in detail with the full resolution spatial and spectral CRISM data. Observations with strong spectral signatures were then used to guide the selection of multipixel regions of interest (ROI). ROIs are selected based on regions highlighted in the mafic summary spectral parameters with associated ratio denominators selected in the same columns in a region of bland mafic parameters, typically in the crater floor. The result was the 23 crater central peaks have CRISM observation, mafic spectral signatures and negligible alteration.

[11] On Mars, crustal exposures occur in three major types of environments; outflow channels [Rogers et al., 2005; Loizeau et al., 2010], tectonic rifting such as in Valles Marineris [Flahaut et al., 2011], and impact craters [Melosh, 1989; Baratoux et al., 2007; Cahill et al., 2009; Tornabene et al., 2010]. Of these, impact craters are the most globally pervasive with the deepest excavated materials typically exposed in the central peak structures [Melosh, 1989]. The vertical relief of central peaks is the key to preserving the strong spectral signatures by allowing surface refreshing and preventing crater fill from obscuring the observations. Since we require compositional analysis capable of distinguishing multiple lithologies within these peaks, we will use spectral data from the CRISM instrument to identify, map and analyze the central peak units. Multispectral, 100 m/pixel resolution data from the Thermal Emission Imaging System (THEMIS) instrument [Christensen et al., 2004] are also used to compliment and interpret the CRISM-based spectral mapping of craters peaks in Alga and Ostrov Crater. Morphology of both central peaks was analyzed with HiRISE 25 cm/pixel imagery. Stereo HiRISE [McEwen et al., 2007] observations of the Alga central peak were processed into a digital elevation model [Kirk et al., 2008] that allowed determination of precise 3D perspectives.

[12] CRISM. The primary data set used in this study was observations from the CRISM VNIR hyperspectral imaging instrument [Murchie et al., 2007]. CRISM acquires full resolution targeted (FRT) images at 18 m per pixel spatial resolution and 544 spectral bands ranging from 0.32 – 4.0 μm. CRISM observations were corrected for instrumental artifacts and were converted into I/F [Murchie et al., 2007]. To use the observations on surface mineral measurements, a simple multiplicative correction was used to remove the atmospheric contribution to the spectra. This was determined using a volcano scan method developed and tested on ISM [Bibring et al., 1989] and OMEGA [Mustard et al., 2005]. A detailed description of the volcano scan method is in McGuire et al. [2009]. The CRISM data are collected on a short wavelength S-detector (VNIR: 362–1053 nm) and a long wavelength L-detector (IR: 1002–3920 nm). Full wavelength modeling requires detector coordination. Ratioed spectra were calculated for each detector independently, then joined by removing the overlapping spectral data (1000 nm – 1053 nm) from the S-detector data and including an additive scaling factor on the S-detector data to match the value of the minimum wavelength of the L-detector [Murchie et al., 2007]. The absolute value of the ratioed value is dependent on the relative albedo of the region of interest and denominator area and does not affect modeling as long as the values are consistent between detectors. Mafic mineral deposits are identified with mafic parameter maps that highlight spatial regions with characteristic spectral absorptions [Pelkey et al., 2007; Salvatore et al. 2010]. The parameter equations used to identify olivine, low-Calcium pyroxene (LCP) and high-Calcium pyroxene (HCP), respectively, are given below:OLINDEX2:

  • display math

LCPINDEX:

  • display math

HCPINDEX:

  • display math

Where R#### is the reflectance value at #### nm and RC#### denotes the value of a point at a wavelength of #### nm along a modeled line that follows the average slope of the spectrum. Average spectra were collected from regions of interest in the mineral deposits and were ratioed to a spectrally bland region in the same column of pixels, typically in the crater floor fill.

[13] MGM Method. The modified Gaussian method (MGM) [Sunshine et al., 1990] was used to estimate the cation compositions of the minerals olivine and pyroxene (Figure 3). The MGM deconvolves overlapping absorptions of mafic mineral spectra into their fundamental absorption components. Individual absorption components are modeled as modified Gaussian distributions that mathematically describe the specific shape of electronic transition absorptions and are parameterized by a band center, band width, and band strength. The MGM models a spectrum's absorption features with Gaussian distributions and relates them to known absorption features. Gaussians are defined as:

  • display math

where s is band strength, σ is the band width, and μ is the band center. The MGM superimposes several Gaussians onto a continuum to model the spectrum with known absorption features at fixed wavelengths due to specific electronics transitions, discussed further below. MGM computations are carried out in energy and natural log reflectance space and thus overlapping absorptions are additive and can be modeled using linear inverse theory [Sunshine et al., 1990]. The inversion method applied in Sunshine et al. [1990] is the stochastic inversion of Tarantola and Valette [1982] that allows for the inclusion of a priori information as constraints on the solutions. Sunshine and Pieters [1993] showed that the constraints helped to stabilize the inversion process and prevent physically unrealistic solutions. The inversion is an iterative process and all absorption band parameters and the continuum are free to move until the residual errors between the log of the actual spectrum and the log of the modeled spectrum are less that 10−5 between consecutive model runs [Sunshine et al., 1990].

image

Figure 3. MGM Example. (a) MGM run on laboratory Olivine with Fo # 96 (Relab sample: c3po53). (top left) RMS error between the observed spectra and modeled fit. (middle left) Three olivine absorptions with fitted positions, widths, and strengths. Initial positions, depths and widths are listed in Table 1. (bottom left) Olivine spectra (cross hatched points) with fit (black line) and curved continuum (thin black line). (b) MGM run on laboratory Pyroxene (Relab Sample: c1xp23). (top right) RMS error between the observed spectra and modeled fit. (middle right) Five absorptions with fitted positions, widths, and strengths. (bottom right) Pyroxene spectra (cross hatched points) with fit (black line) and curved continuum (thin black line).

Download figure to PowerPoint

[14] Initial MGM development was performed on laboratory measurements of pyroxenes [Sunshine et al., 1990; Sunshine and Pieters, 1993] with later development using laboratory measurements with a range of olivine compositions [Sunshine and Pieters, 1998]. MGM has since been used on laboratory measurements of lunar olivine [Isaacson and Pieters, 2010], remotely sensed lunar olivine from the M3 instrument [Isaacson et al., 2011], laboratory and remotely Martian pyroxenes [Kanner et al., 2007; Skok et al., 2010], remote terrestrial observations [Combe et al., 2006] and Martian mafic mixtures from OMEGA observations [Clénet, 2011]. This investigation is the first full treatment of CRISM data using the MGM for compositional determination. Initial selection of Gaussian absorption properties (band center, band width, and band strength) was determined based on knowledge of the spectral properties of mafic minerals (Table 2) [Burns, 1993; Sunshine and Pieters, 1993, 1998], and the model was then allowed to automatically modify these values to produce the best fit to the observed spectrum. In cases where the automatic fitting routine failed to provide a satisfactory fit, attempts were made to refine the fit by first fitting a continuum curved in energy space [Hiroi and Sasaki, 2001] before allowing freedom of the individual band absorptions.

Table 2. List of Initial MGM Modeled Absorption Propertiesa
Absorption FeatureBand Center (μm)Band Width (μm)Band Depth
  • a

    Band strength in natural log reflectance.

Olivine M1–10.90.228−0.08
Olivine M21.080.176−0.08
Olivine M1–21.280.424−0.08
Pyroxene 1 μm1.00.288−0.08
Pyroxene 2 μm2.00.424−0.08
Pyroxene 1 μm LCP0.90.228−0.08
Pyroxene 1 μm HCP1.080.228−0.08
Pyroxene 1.2 μm1.280.176−0.08
Pyroxene 2 μm LCP1.90.424−0.08
Pyroxene 2 μm HCP2.30.424−0.08

[15] Olivine Analysis. Olivine ((Mg, Fe)2SiO4) is a magnesium-iron silicate with a nesosilicate structure. It is spectrally identified in the near-infrared by a broad absorption centered near 1μm that is the superposition of three overlapping absorptions caused by electronic transitions in Fe2+ ions located in distorted octahedral crystal lattice sites [Burns, 1970, 1974, 1993]. The exact band center of each of these absorptions is related to the relative proportion of Fe and Mg in the M1 and M2 sites with increasing Mg causing shorter wavelength absorptions and increasing Fe causing longer wavelength absorptions [Burns, 1970; King and Ridley, 1987; Sunshine and Pieters, 1998]. To avoid spectral effects outside the diagnostic 1 μm band, we only model from 0.8 μm to 1.8 μm but visually check that there are no 2 μm spectral features related to pyroxene. Three bands are used to model this absorption (Table 2) with the resulting fitted band centers determining the composition [Sunshine and Pieters, 1998; Isaacson and Pieters, 2010]. To determine the composition of a given spectrum, the three band centers are fit to a laboratory derived compositional trend line to fit for least error using the following relationship [Isaacson and Pieters, 2010].

  • display math
  • display math
  • display math

where A and B are constants that constrain the slope of the experimentally determined compositional relationship that relates the composition of the olivine to the band center for each of the three olivine absorptions. M1–1, M2, and M1–2 represent the band centers (in nm) for the three olivine absorptions. Fo is the cation composition ratio of Mg wt.% /(Mg wt.% + Fe wt.%) ranging from 0 (100% Fe) to 100 (100% Mg).

[16] The formula can be modified to be fit to any two band inputs if needed. When modeling CRISM data we found that limiting analysis to the L detector data produces most reliable results because of the L + S detector alignment. The natural properties of the long wavelength M1–2 feature gives it the most wavelength variability per change in composition, meaning that small changes in bandcenters will have the least effect on the modeled composition making it the most stable feature band to fit. For these reasons, we include the modeled fits both using all three olivine bands and alternatively only using the M2 and M1–2 bands.

[17] The MGM analysis of olivine was initially shown to be independent of grain size [Sunshine and Pieters, 1998], but subsequent work has detected a dependence in laboratory spectra of particulates [Clénet et al., 2011]. The precise nature of this dependence is still an active source of research and of questionable applicability for spectra measurements of crystalline material with interlocking crystals with no space between crystal faces, creating a much different reflection path than in particulate material.

[18] Pyroxene Analysis. Pyroxene ((Ca, Mg, Fe)Si2O6) is spectrally identified by absorptions at 1 and 2 μm with the band center slightly dependent on the Mg-Fe cation content and strongly dependent on the Ca content [Adams, 1974; Cloutis et al., 1986; Cloutis and Gaffey, 1991; Klima et al., 2007] and a weak 1.2 μm absorption. The pyroxene absorptions are caused by crystal field transitions of iron in octahedral coordination with the 1- and 2-μm absorptions due to the composition of the M2 crystallographic site. This causes distortion in the crystal structure based on the size of the cation. The 1.2 μm absorption is due to molecular distortion in the M1 crystallographic site [Burns, 1993]. Pyroxene Ca content can be determined by modeling each absorption as a combination of end-member low-calcium and high-calcium pyroxenes.Kanner et al. [2007]has shown that the ratio of modeled end-member absorption strengths is proportional to the relative composition of each end-member. Since pyroxene composition is spectrally distinguished by the Ca content, we refer to high-Ca (HCP) and low-Ca (LCP) compositions. This spectral determination does not directly distinguish structural variants of orthropyroxene (OPX) and clinopyroxene (CPX). However, in practice LCP strongly coincides with OPX and may indicate enstatite or pigeonite while HCP coincides with CPX, typically augite or diopside. Pyroxene is modeled in two ways. The first is with 5 absorptions (Table 2); the LCP and HCP end-members of both the 1 and 2μm absorptions and an absorption at 1.2 μm are used to provide a reliable fit but not used to determine composition. Relative Ca content is determined by ratioing the modeled LCP end-member band strength by the combination of the LCP and HCP strength, LCP/(LCP + HCP) determining the normalized band strength ratio (NBSR) [Kanner et al., 2007]. This is done for both the 1 and 2 μm absorptions. A value of 1.0 indicates there is only LCP while a value of 0.0 indicates only HCP, with intermediate values proportional to the relative Ca content. The second way pyroxene is modeled is with a single 1 and 2 μm absorption to calculate the overall band centers for classification against terrestrial pyroxenes.

[19] MGM Error Consideration. MGM was developed and tested by analyzing laboratory spectral measurements of well-characterized samples. Olivine compositional error occurs at two separate steps. The first is during the MGM spectral fitting process that models the CRISM spectra with the Gaussian absorptions. The model is running until the RMS increases less than 10−5 with a successive iteration and is always lower than 10−2. However, errors from slight variations in initial spectral parameters lead to an error of ∼±5 Fo# for olivine [Isaacson and Pieters, 2010]. Additional errors are expected when applying the MGM to remote sensing data with lower signal-to-noise and additional spectral components. The second step is the calculation of Fo# from the modeled bandcenters.Isaacson and Pieters [2010] determined a composite error for the MGM of olivine with the M3 data set to be ∼±20 relative Fo# units. The composition is calculated as a least squares fitting to either 3 or 2 slopes. The greater the difference between the laboratory determined slopes and actual data the larger the error in the compositional determination. This error in the determination of the precise composition of a single unit spectral average encourages the use of the modeled results to determine general ranges and regional trends instead of the exact composition of a single exposure. A separate study applying the MGM to pyroxenes with laboratory and OMEGA data determined that the calcium content can be constrained to ±10% [Kanner et al., 2007].

[20] Thermal Infrared. We use the multispectral thermal infrared observations from the Thermal Emission Imaging System (THEMIS) instrument [Christensen et al., 2004] to provide an independent check on the composition. THEMIS operates both a TIR multispectral mode with nine bands between from 6.8 to 14.9 μm with 100-m per pixel resolution and a visible mode with five bands and 18-m per pixel resolution. Hyperspectral observations from the TES instrument were examined for both the Alga and Ostrov central peaks. Alga's central peak did not have high quality data coverage. A few TES spectra were available for Ostrov's central peak; however these were determined to have artifacts invalidating mineral determinations.

[21] THEMIS daytime multispectral images with warm (>265 K) surface temperatures were selected for detailed analysis of the two central peaks. The images were calibrated and atmospherically corrected using the methods described by Bandfield et al. [2004]. Spectrally distinct units within each scene were determined by examining CRISM mafic parameters and THEMIS decorrelation stretch (DCS) [Gillespie et al., 1986] images. To quantitatively map the spectral unit distributions, spectra were extracted from distinct units identified in the DCS images and averaged. These spectra were then used to model the scene emissivity with a linear least squares minimization routine. For both Alga and Ostrov, the most olivine-rich spectral units could be modeled as a combination of another unit in the scene plus a laboratory spectrum of olivine. Because of this, the olivine-rich spectral unit does not represent a true end-member and the olivine-rich unit spectrum was replaced with a pure olivine spectrum for mapping the spectral unit distributions [Rogers and Bandfield, 2009]. Root-mean square (RMS) error images calculated from the least squares minimization were examined to ensure that all surfaces were relatively well-modeled by the spectral library.

3. Type Region Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[22] The spectral analysis of CRISM observed unaltered, well-exposed crater central peaks have yielded 23 sites (Table 1). The observed craters range in diameter from ∼130 km to 7 km. The observed sites can be divided into four concentrated regions; North Argyre (NA), Northern Hellas (NH), Nili Fossae (NF) and the Northern Plains (NP) (Figure 2 and Table 1) that we will describe to look for regional relationships. While each crater has its own unique attributes, we include in-depth unit analysis of the two best exposed central peaks in Alga and Ostrov Craters (Figure 4) and general descriptions of one example from each designated region to highlight some of the observed diversity and then focus on the aggregate compositions to analyze global trends.

image

Figure 4. Alga and Ostrov Crater Context Map. (a) Alga and Ostrov craters are located just east of the Margaritifer-Uzboi fluvial system, south of Ladon and east of Holden Crater. (MOLA DEM). (b) Alga is a 18 km crater just off center of the 86 km Chekalin Crater. (CTX: P18_007929_1555_XI_24S026W). (c) Ostrov is a modified 65 km crater (CTX: P17_007639_1514_XN_28S027W).

Download figure to PowerPoint

[23] Alga Crater. Alga crater is 19 km wide and is located just east of the center of the 85 km diameter degraded Chekalin crater. The rim of Alga crater exposes layered light-toned olivine-bearing outcrops best exposed toward the southeast. The western and northern crater wall and terraces show scattered exposures of olivine- and pyroxene-bearing outcrops. The central peak of Alga is ∼200 m high and exposes a complex relationship among local mafic units. The terrain was divided into four units based on morphologic and spectroscopic analysis of the central peak (Figure 5): a distinct light-toned olivine-bearing unit, a pyroxene-bearing bedrock unit, a light-toned pyroxene unit found on the floor of the crater and a fine-grained deposit with a varying clast content that are interpreted to be an impact melt unit on the flanks of the central peak (Figure 6). Representative regions of each unit were analyzed with the methods described above to determine the cation composition and lithology.

image

Figure 5. Map of Alga crater central peak units. Olivine-bearing unit (area A). Pyroxene-bearing unit (area B). Light-toned pyroxene (area C). Possible melt unit (area D). CRISM FRT00006415 mafic parameters on HiRISE PSP_007573_1555. R:Olindex2, G:LCPindex, B:HCPindex.

Download figure to PowerPoint

image

Figure 6. Oblique view of Alga Crater from HiRISE draped on HiRISE DEM (5x vertical exaggeration). (a) View from south showing main peak structure dominated by Pyroxene-bearing bedrock. Light-toned Pyroxene unit seen to south and east of main peak. (b) View of peak from west. Light toned units to left and bottom are Olivine-bearing. Potential melt unit on peak flank following central couloir. (c) View from east showing tabular nature of Pyroxene-bearing bedrock unit.

Download figure to PowerPoint

[24] Alga Crater Olivine Unit. The olivine-bearing unit correlates with partially to well-exposed relatively light-toned massive outcrops and relatively dark deposits bearing abundant light-toned clasts, which drape and appear to flow off the bedrock of the central uplift on the northern slopes. The olivine-rich outcrops are located primarily on the northern and western slopes of the central peak with a few small megablocks observed on the eastern side. This unit typically features a sharp boundary with the adjacent units that are easily observed by the relative contrast in tone, but the unit is also characterized by a dense network of relatively dark-toned fractures and mantles of dark-toned materials, interpreted to be impact melts generated by the Alga-forming event. Spectrally the ratioed olivine exhibits the characteristic olivine absorptions and no other detectable spectral features (Figure 7) in the VNIR range. Regions of interest were selected from ten locations within the unit for MGM analysis. MGM modeling and composition determination yields modeled result range of Fo9–43 with a mean of Fo18 (Table 3).

image

Figure 7. Alga Olivine Results. (a–d) Olivine deposits shown in dotted lines. Truncated spectra used in MGM modeling. (e) MGM modeled olivine band center fits, results in Table 3. (f) Full spectral range of Alga crater olivine.

Download figure to PowerPoint

Table 3. List of Modeled Olivine Regionsa
 M1–1M2M1–23-Band Comp (Fo)Comp Fitting Error2-Band Comp (Fo)Comp Difference
  • a

    Bandcenters and Compositions. Column 1:Region ID Region: Region#:ROI #, 2: Olivine M1–1 band, 3: Olivine M2 band, 4:Olivine M1–2 band, 5: Olivine least squares fit composition calculated from M1–1, M2, and M1–2 bands, 6: Least squares fitting error for three band fit, 7: Olivine least squares fit composition for M2 and M1–2 bands, 8: Absolute difference between 3 band and 2 band compositions.

North Argyre Olivine Unit
NA1_18771046125745160461
NA1_289910721282163151
NA1_38891063129018270117
NA1_490110691298717225
NA1_59021069129398763
NA1_6897106712891491103
NA1_7902106712851443140
NA1_8900107912811444131
NA1_9896106412762226231
NA2_1843103512645914084415
NA2_2836103512546811725315
NA2_3838103112517011235613
NA2_4837103412636316534518
NA4_19001080126820190233
NA4_2897108012791686142
NA5_1835104012586513404718
NA5_28911053126334135384
NA5_3874104812475162532
 
North Hellas Olivine Unit
NH3_1895107412941015147
NH3_2899107512851320112
NH3_3898108012811482123
NH3_4899108012811474122
NH4_19031077126920144244
NH4_2899108012801571131
NH4_3899108112791582132
NH4_4899108112791584131
NH4_5901107812791447140
NH4_68971088127616292133
NH4_78911078126925123232
NH6_1907107312898580
NH8_1888106512523897424
NH8_29051075126024341328
NH8_3917108312432516534318
 
Nili Fossae Olivine Unit
NF1_1900108012801467131
NF2_19041075129372643
NF2_2900108212955124−16
NF2_3899108012801476132
NF2_4901107912801452131
NF2_5898108312966183−17
 
Northern Plains Olivine Unit
NP1_18611039126351661438
NP1_28731045126344285413
NP1_3887105512663478351
NP1_48551039125757587488
NP1_58561040125458466507
NP1_68671044124656106560
NP1_7879105112445082555
NP1_88661044126049355445
NP1_98701041125252246520
NP1_108871052126236122393
NP1_118671034125156436560
NP1_128831050126139121411
NP1_138811048126041173421
NP1_148631037125157360552
NP1_158661039124857252560
NP1_168741044124056141615
NP1_17899106112732378273
NP2_1900106812861434121
NP2_28891062128521198155
NP2_3896107612911111456
NP3_187410591298211107616
NP3_28851062129119419109
NP3_390510711282139152
NP3_489810701295914846
NP3_590310741284121110
NP3_6898107412871331103
NP3_7899107912831357113
NP3_8895107712921014247
NP3_987110611301211378318
NP3_108881069130310607−212
NP4_18741051127834511268
NP4_286210491279399532514
NP4_385810431270478613612
NP4_486310471273437093211
NP4_588610641263331330
NP4_68711047125847181453
NP4_78871074129315337410
NP5_1896106312722423262
NP5_2892105812692973312
NP5_388810621268309301
NP5_48701043125053164530
NP5_590110651249344414511
NP5_68941067125633139386
NP5_78941055126034150406
NP5_8897108012791693142
NP5_9896106112623082355
NP5_10901106912702153254
NP5_11899107912801556132
NP5_12898108012801581132
NP5_139011063126228187358
NP5_1489810711276192201
NP5_15898107612771734171
NP5_1689610641243404355011
NP5_178791043123756253648
NP5_1890110581246375475013
NP5_19890105512613590394
NP5_20899106512682564295
NP5_219041064126524169327
NP6_1857104912903617701719
NP6_28781076131471478−1421
NP6_386410581299261691520
NP6_487310651308151512−520

[25] Alga Crater Pyroxene-Bearing Bedrock Unit. The pyroxene-bearing bedrock unit comprises the remaining bulk of the central peak bedrock units, dominating the peak's southern portion. Fractures within this unit display a variety of textures ranging from semi-angular on the west side of the peak to tabular fractures on certain parts of the eastern slope of the peak (Figure 6). Spectral analyses of eight regions of interest all confirm low-calcium pyroxene with similar band centers (Figure 8). A slightly longer 2 μm band for the observed 1 μm band position indicates that this unit is relatively Fe-rich [Adams, 1974; Cloutis and Gaffey, 1991]. NBSR values of each of the pyroxene absorption feature show reassuring agreement between the 1 and 2 μm absorptions (Table 4), supporting the absence of olivine spectral effects, which typically expresses itself as a modification of the 1 μm absorption features. The elevated NBSR values (2 μm: 0.59–0.61) support a LCP enrichment of this pyroxene unit.

image

Figure 8. Alga pyroxene results. (a) Southwest portion of Alga central peak containing most of the Pyroxene Bedrock Unit. (b) Ratioed Alga pyroxene spectra used in MGM analysis. (c) Alga crater Pyroxene Bedrock 1-, 2μm band centers plotted against terrestrial pyroxene band centers [Adams, 1974; Cloutis and Gaffey, 1991]. From bottom left to upper right, pyroxene composition goes from Mg-rich to Fe-rich to Ca-rich. Alga pyroxene plot as low-Ca, moderate-Fe. (d) Pyroxene NBSR demonstrating a consistent LCP enrichment.

Download figure to PowerPoint

Table 4. Pyroxene Observation Tablea
 1 μm LCP Band Center1 μm LCP Band Depth1 μm HCP Band Center1 μm HCP Band Depth1 μm NBSR2 μm LCP Band Center2 μm LCP Band Depth2 μm HCP Band Center2 μm HCP Band Depth2 μm NBSR1 μm Band Center2 μm Band Center
  • a

    MGM fits of pyroxene spectra. LCP and HCP bandcenters and NBSR were determined by 5 Gaussian modeling. NBSR is calculated from bandstrengths (LCP/(LCP + HCP). One and 2 μm bandcenters determined by 3 Gaussian modeling.

North Argyre Pyroxene
NA1_1889−0.111075−0.050.671760−0.122145−0.080.599331893
NA1_2886−0.121064−0.070.631746−0.142138−0.100.599381884
NA1_3893−0.091075−0.050.661757−0.102152−0.070.599411892
NA1_4890−0.111075−0.060.641751−0.132133−0.090.599411886
NA1_5889−0.101078−0.040.721773−0.122135−0.080.609221896
NA1_6896−0.081068−0.060.591759−0.142131−0.090.609571890
NA1_7892−0.101080−0.040.711769−0.122132−0.080.619271891
NA1_8893−0.081065−0.030.741782−0.102148−0.070.599161909
NA1_9894−0.111082−0.060.631743−0.132129−0.080.619501867
NA2_1901−0.231099−0.060.791801−0.212142−0.110.659181892
NA2_2899−0.161086−0.050.761850−0.172199−0.080.688981912
NA2_3902−0.151110−0.050.741912−0.262528−0.160.628271892
NA2_4903−0.161103−0.040.801839−0.172183−0.070.709111904
NA2_5896−0.201083−0.050.801841−0.162162−0.070.719161911
NA2_6898−0.201078−0.070.741775−0.192168−0.110.639191877
NA2_7901−0.101151−0.040.711936−0.222517−0.180.569162141
NA2_8891−0.221078−0.030.861834−0.162142−0.050.779021887
NA2_9920−0.151077−0.020.881894−0.1222990.001.009811979
NA2_10900−0.151141−0.020.891908−0.252563−0.210.548051875
NA3_1905−0.0210820.000.841779−0.062164−0.030.649211882
NA4_1920−0.081075−0.030.701809−0.092200−0.070.569571954
NA4_2913−0.071084−0.030.671792−0.072192−0.060.569611941
NA4_3906−0.061088−0.050.541783−0.082181−0.070.569971932
NA6_1904−0.021080−0.030.461872−0.042285−0.030.589941991
NA6_2921−0.021074−0.030.461850−0.052236−0.030.6110051964
NA6_3919−0.021075−0.020.471850−0.042232−0.030.5910051987
NA6_4923−0.061120−0.070.491744−0.102165−0.070.599991876
NA6_5916−0.051112−0.050.501762−0.082158−0.060.609901890
NA6_6899−0.021081−0.020.411906−0.032340−0.040.479942100
NA6_7920−0.051115−0.050.511792−0.072180−0.040.629941901
NA6_8906−0.071105−0.090.461735−0.132150−0.100.5710211919
NA6_9915−0.031088−0.040.471749−0.072155−0.050.5710011892
 
North Hellas Pyroxene
NH1_1957−0.071183−0.060.541761−0.102222−0.080.5710141939
NH1_2948−0.071192−0.090.441749−0.112177−0.070.6110981955
NH1_3930−0.051109−0.070.421756−0.112193−0.100.5210161955
NH1_4921−0.011060−0.020.351837−0.062208−0.040.5910101957
NH2_1920−0.081075−0.030.701809−0.092200−0.070.569571954
NH2_2913−0.071084−0.030.671792−0.072192−0.060.569611941
NH2_3906−0.061088−0.050.541783−0.082181−0.070.569971932
NH3_1918−0.041073−0.030.541827−0.082216−0.070.549831984
NH3_2905−0.051082−0.050.541800−0.112180−0.090.569861952
NH3_3900−0.061085−0.050.571782−0.102165−0.080.569781935
NH3_4909−0.061083−0.050.561803−0.112184−0.080.569831953
NH3_5907−0.061083−0.050.551796−0.112192−0.090.549891958
NH3_6907−0.011084−0.020.381811−0.032239−0.020.5810151977
NH3_7906−0.021079−0.040.401804−0.072215−0.060.5610251950
NH3_8894−0.031086−0.030.471845−0.092242−0.070.569891984
NH3_9898−0.021060−0.020.541859−0.072235−0.060.559652004
NH3_10905−0.061085−0.050.531793−0.102182−0.080.569911945
NH3_11900−0.071080−0.060.521781−0.092180−0.080.559911941
NH3_12895−0.061081−0.050.541784−0.092183−0.080.559801945
NH3_13913−0.031083−0.010.711869−0.062246−0.050.559322008
NH3_14910−0.031069−0.010.681879−0.052255−0.040.569352013
NH3_15901−0.061084−0.050.531782−0.102176−0.080.569881933
NH4_1912−0.041080−0.040.521814−0.062219−0.040.609931927
NH4_2904−0.061097−0.050.511784−0.062199−0.040.619781893
NH4_3893−0.051089−0.050.511784−0.082195−0.050.629941892
NH4_4925−0.051069−0.020.701868−0.082219−0.060.609561982
NH4_5777−0.071023−0.070.471813−0.092256−0.070.559551997
NH4_6854−0.081203−0.060.571961−0.112467−0.040.749571984
NH4_7901−0.011082−0.010.571859−0.032283−0.020.568471993
NH5_1922−0.071128−0.040.641851−0.082190−0.030.719601917
NH5_2921−0.061121−0.030.641844−0.062192−0.030.699601919
NH5_3917−0.061101−0.030.691861−0.062180−0.030.699521937
NH5_4903−0.021079−0.020.561900−0.0323010.000.869921961
NH5_5911−0.041080−0.010.731866−0.042275−0.020.719371924
NH5_6921−0.071129−0.050.561809−0.062183−0.050.589821936
NH5_7915−0.041088−0.050.471811−0.072191−0.050.569961956
NH5_8920−0.041095−0.030.621896−0.052236−0.020.759671950
NH5_9923−0.051104−0.030.611850−0.042207−0.020.659761947
NH6_1900−0.031087−0.030.481872−0.022291−0.010.739702011
NH7_1901−0.011080−0.010.601900−0.042300−0.020.669711977
NH8_1910−0.051086−0.030.601937−0.032279−0.020.609372010
NH8_2873−0.041075−0.040.491949−0.032329−0.040.469962171
NH9_1853−0.081010−0.090.471753−0.242224−0.190.569231335
NH9_2857−0.081062−0.030.701817−0.132344−0.110.568821341
NH9_3923−0.081022−0.030.751724−0.142159−0.110.559381340
NH9_4912−0.091031−0.050.621812−0.152274−0.120.559521335
 
Nili Fossae Pyroxene
NF1_1854−0.081019−0.050.611750−0.112104−0.080.578941886
NF1_2891−0.061064−0.040.611755−0.092121−0.060.589471897
NF1_3795−0.071038−0.060.541787−0.092304−0.060.638661886
NF1_4856−0.081044−0.060.561769−0.092190−0.070.589281922
NF1_5896−0.091095−0.040.691754−0.092168−0.070.569431920
NF1_6879−0.081071−0.050.621769−0.102166−0.070.589481918
NF2_1922−0.031110−0.040.421794−0.062229−0.040.6010631898
NF2_2922−0.081141−0.070.521734−0.102137−0.070.6010391818
NF2_3909−0.041078−0.030.591781−0.072179−0.060.569691934
NF2_4920−0.031064−0.020.561771−0.062194−0.050.579691911
NF2_5940−0.071168−0.090.441769−0.092203−0.060.6111561828
NF2_6906−0.031097−0.040.431816−0.042236−0.030.6210451946
 
Northern Plains Pyroxene
NP1_1918−0.041118−0.050.461762−0.062192−0.040.5910121880
NP1_2917−0.061131−0.060.481736−0.082173−0.060.5910171882
NP1_3911−0.021075−0.040.401789−0.062201−0.050.5510121946
NP1_4920−0.021063−0.020.481814−0.062201−0.040.589881945
NP1_5913−0.021083−0.030.411767−0.052206−0.040.5910171882
NP1_6907−0.011087−0.020.361774−0.042215−0.030.6210321878
NP1_7913−0.031103−0.050.371745−0.092170−0.060.6010351876
NP1_8913−0.021084−0.040.381786−0.072181−0.050.5810261921
NP1_9911−0.021091−0.030.421782−0.062176−0.040.5710201923
NP2_1877−0.101068−0.110.481939−0.062302−0.110.369672194
NP2_2900−0.041079−0.030.531904−0.052279−0.060.429722141
NP2_3908−0.041076−0.050.461901−0.072318−0.120.379842208
NP2_4874−0.081084−0.090.462004−0.062377−0.130.349582287
NP2_5906−0.031077−0.020.591897−0.032321−0.080.299202267
NP5_1918−0.041118−0.050.461762−0.062192−0.040.5910121880
NP5_2917−0.061131−0.060.481736−0.082173−0.060.5910171882
NP5_3911−0.021075−0.040.401789−0.062201−0.050.5510121946
NP5_4920−0.021063−0.020.481814−0.062201−0.040.589881945
NP5_5913−0.021083−0.030.411767−0.052206−0.040.5910171882
NP5_6907−0.011087−0.020.361774−0.042215−0.030.6210321878
NP5_7913−0.031103−0.050.371745−0.092170−0.060.6010351876
NP5_8913−0.021084−0.040.381786−0.072181−0.050.5810261921
NP5_9911−0.021091−0.030.421782−0.062176−0.040.5710201923
NP5_10920−0.031112−0.040.481758−0.072160−0.050.5910121886
NP5_11914−0.031097−0.040.431756−0.072167−0.050.6110171877
NP5_12909−0.011073−0.010.531824−0.062241−0.050.559801969
NP5_13902−0.011079−0.010.521899−0.032301−0.020.539872052
NP5_14905−0.111097−0.080.561745−0.162142−0.120.579401907
NP5_15917−0.061088−0.050.531807−0.122206−0.100.5510021969
NP5_16912−0.031082−0.020.601784−0.062184−0.040.579671922

[26] Alga Crater Light-Toned Pyroxene Unit.The second pyroxene-bearing unit is located on the crater floor in two distinct occurrences just south and east of the central peak (Figure 9). This unit is lighter than the pyroxene-bearing bedrock and is significantly brecciated with a wide range of clast sizes. The lighter-tone of the bedrock makes this unit easy to detect and distinguish from the darker background material of the crater floor. MGM spectral analysis of 7 regions show well-formed pyroxene spectral features with band centers similar to the pyroxene bedrock unit, again consistent with a relatively high-Fe, low calcium pyroxene. However, this unit displays less VNIR spectral contrast, creating more error in the spectral modeling. The lower spectral contrast may be caused by several contributing factors, the most likely being smaller grain sizes or the intimate mixing of a spectrally neutral but relativity high albedo material. Both of these options would lead to the albedo contrast and observed spectral properties, but the rocky appearance of the outcrop would support the mixing hypothesis with a material similar to the glassy melt unit described below.

image

Figure 9. Alga Light-toned pyroxene results. (a) HiRISE view of southern Light-toned Pyroxene Unit. (b) HiRISE view of eastern Light-toned Pyroxene Unit. (c) Ratioed Alga Light-toned Pyroxene spectra used in MGM analysis. (d) Alga crater Light-toned Pyroxene 1-, 2μm band centers plotted against terrestrial pyroxene band centers [Adams, 1974; Cloutis and Gaffey, 1991]. From bottom left to top right, pyroxene composition goes from Mg-rich to Fe-rich to Ca-rich. Alga Light-toned pyroxene plot as low-Ca, moderate to high Fe. (e) Pyroxene NBSR demonstrating a consistent LCP enrichment. Values listed inTable 4.

Download figure to PowerPoint

[27] Three of the regions of interest (ROI: 1, 2, 5) display a secondary absorption feature centered at 2.2 μm superimposed on the pyroxene absorption (Figure 10). This absorption is similar in width and position to a metal-OH feature, particularly Si-OH [Anderson and Wickersheim, 1964]. These spectral features may indicate that this unit is a mixture of pyroxene and a silica-rich impact melt or potentially a product of an impact induced hydrothermal activity. The presence of these features in only a few of the examined regions may be due to a heterogeneous molecular hydration throughout this unit. This feature is unique to this unit and not observed in any other pyroxene unit examined in this study.

image

Figure 10. Selected Light-toned Pyroxene Spectra (ROI: 1, 2, 5). Spectra highlights 2.2μm feature consistent with Si-OH feature with line at 2.2μm. Potentially the result of hydration of a silica rich impact melt or an impact induced hydrothermal system.

Download figure to PowerPoint

[28] Alga Crater Impact Melt.The final unit that figures prominently in the central peak is a fine-grained deposit that displays weak pyroxene features and makes up a large part of the western portion of the central peak (Figure 11). This unit is characterized by potential flow textures going downslope and contains meter-sized clasts of the pyroxene-bearing bedrock at low elevations. Based on these characteristics and geologic context we interpret this unit as an impact melt-bearing deposit. The melt is located on the central peak apex and fills the peak's northern, southern and eastern couloirs. The high elevation regions of this unit near the peak summit appear clast free at HiRISE's 25 cm/pixel resolution. Near the base of the peak the melt contains meter sized clasts of the pyroxene-bearing material. All examined regions of the melt unit on the central peak show detectable pyroxene spectral signatures. This unit has similar band centers to the other pyroxene units but with much weaker band depths.

image

Figure 11. Impact melt. (a) HiRISE view of northern section of Alga central peak with impact melt unit centered in dotted line. (b) Ratioed Alga impact melt spectra.

Download figure to PowerPoint

[29] Pyroxene Unit Comparison. The Pyroxene Bedrock, Light-Toned Pyroxene, and Impact Melt units all display clear pyroxene spectral absorptions with similar band centers but have a significant difference in band depth (Figure 12). The Pyroxene Bedrock has a low calcium signature with a well-developed 1.2μm shoulder caused by Fe2+ in the M1 site [Klima et al., 2008]. Strong spectral absorptions make this unit the best developed pyroxene member in the scene. Conversely, the impact melt unit has a weak pyroxene band depth and is thought to be comprised of an intimate mixture of spectrally neutral glass with clasts of the pyroxene units providing the relatively weak spectral absorptions. The remaining Light-Toned Pyroxene unit has an intermediate band depth but similar band positions as the Pyroxene Bedrock signifying similar composition but with the possible addition of a brightening spectrally neutral component. Under this model, the three pyroxene units are accessing the same lithologic material but are including varying degrees of spectrally neutral materials. The Impact Melt unit would have a quenched glass component to the matrix that would provide the spectral modification, while the modifying material in the Light-Toned Unit is less clear. Impact glass produced in this impact is a possibility as is melt from the preexisting Chekalin crater or other nearby crater basins, but also melt can be injected into fractures during uplift formation. Alternative materials such as plagioclase and quartz could also provide the desired spectral effect but are difficult to constrain with current observations. An alternate possibility is that shock effects on pyroxene can affect spectral contrast, but not the bandcenters [Adams, 1979].

image

Figure 12. Pyroxene comparison. Here we plot representative spectral from the pyroxene bedrock, light-toned pyroxene, and impact melt units. We see that the band centers are similar but a strong difference in absorption strength. We suggest that all three units are sampling the same composition units but are distinguished by differing quantities of a post-impact spectrally modifying component such as an impact glass.

Download figure to PowerPoint

[30] Ostrov Crater. The nearby Ostrov crater provides another well exposed central peak to apply the MGM compositional analysis. Ostrov is a 65 km diameter crater located 150 km to the southwest of Alga. Ostrov has had significantly more modification from erosional modification and deposition than Alga, exemplified by a deeply dissected rim and multiple fans filling the crater interior. The central peak structure stands above these deposits and display strong mafic absorption features (Figure 13a). While Ostrov displays several important similarities with Alga, it is simplified with only three discernible spectral units; an olivine-bearing and a pyroxene-bearing unit and a third spectrally bland region that may be analogous to the glassy melt unit. Much like what we observe in Alga, these units are strongly dominated by a specific mineral phase with little evidence of spectral mafic mixing.

image

Figure 13. Ostrov crater. (a) Ostrov crater with CRISM parameter overlain on HiRISE and CTX. (R:Ol2index, G:LCPindex, B:HCPindex). CRISM:FRT00017D02, HiRISE: PSP_017357_1530, CTX: P17_007639_1514_XN_28S027W. (b) HiRISE of northern olivine deposit. (c) HiRISE of eastern olivine ridge. Shows olivine-bearing light-toned clasts. (d) Ratioed olivine spectra used in analysis. (e) Degraded S detector data (700 nm–1000 nm) caused bad band fitting results. Composition is determined with M2 and M1–2 expressed in L detector (1000 nm–1800 nm). Results inTable 3.

Download figure to PowerPoint

[31] Ostrov Crater Olivine Unit. The olivine-bearing unit in Ostrov (Figure 13) is characterized by the inclusion of many small light-toned outcrops with the olivine dominated region but otherwise is morphologically similar to the pyroxene-bearing unit. The unit has a very sharp boundary with the pyroxene-bearing units and shows no spectral signature of a pyroxene component, again indicating a minimal pyroxene component. A spectral feature near 0.9μm affects the fit of the M1 absorption in each of the ratioed spectra, requiring the use of the M2 and M1–2 bands for determining the cation composition. The MGM analysis of four regions of interest within the olivine unit result in slightly elevated Fo# compared to the Alga olivine unit. The measured olivine spectra had atypical response in the S-detector data. This response is unsuitable for full MGM modeling due to distortion of the M1–1 spectral band. The long wavelength shoulder of the olivine spectra is well formed and suitable for modeling, making the calculation of the Fo# that was done with only the M2 and M1–2 absorptions more reliable for this unit. The high-quality of the near-infrared data and well-developed olivine absorption in this region makes the modeled fit reliable even with the two bands. The longest wavelength M1–2 absorption is the best constrained by the spectra and least dependent on small band center shifts. This unit displays values ranging Fo44 to Fo56 with a mean of Fo50.

[32] Ostrov Crater Pyroxene Unit. The Ostrov crater pyroxene unit spatially dominates the central peak structure. Nine regions of interest have been selected for MGM processing based the extent of pyroxene exposure. The resulting band center determination is consistent with a very low Ca, moderate Fe pyroxene (Figure 14). NBSR values of these pyroxenes show close agreement between the 1 and 2 μm values, supporting a relatively pure pyroxene composition. NBSR values fall in the rage from 0.6 to 0.8, indicating enrichment in LCP. Compared to the Alga pyroxene these are even more enriched in LCP and have slightly less Fe content (Table 4).

image

Figure 14. Ostrov Pyroxene. (a) Ratioed Ostrov pyroxene spectra. (b) Ostrov Pyroxene NBSR results consistent with LCP-enrichment. (c) Ostrov pyroxene 1-, 2μm band centers plotted against terrestrial pyroxene band centers [Adams, 1974; Cloutis and Gaffey, 1991]. From bottom left to upper right, pyroxene composition goes from Mg-rich to Fe-rich to Ca-rich. Ostrov pyroxene plot as low-Ca, moderate Fe with some outliers. Results inTable 3.

Download figure to PowerPoint

[33] THEMIS Spectral Analysis. Analysis of THEMIS observations of Alga's central peak identifies three main spectral units (Figure 15). The first type is dominated by olivine, but is not a spectral match to pure olivine. The THEMIS spectrum is consistent with olivine plus one or more components; a good spectral match can be achieved with a 60:40 mixture of olivine and a palagonite-like high-silica poorly crystalline phase. It can also be well modeled as a mixture of olivine plus the surrounding central peak surface. The second spectral type maps to the majority of central peak and is spectrally consistent with TES Surface Type 1 [Bandfield et al., 2000], within the THEMIS spectral range, though the spectral bands of THEMIS limit the determination of a precise mineralogy. The third type is spectrally similar to an intermediate between Surface Type 1 and Surface Type 2 [Bandfield et al., 2000] and maps to the crater floor and rim. Surface Type 1 is considered to be basaltic in nature. Surface Type 2 is consistent with altered basalt [Wyatt and McSween, 2002] or some elevated silica content and may represent the mixture of the central peak mafics with a glassy impact melt component.

image

Figure 15. THEMIS thermal results. (a) Alga THEMIS spectra from end-member regions. (b) Reference end-member library spectra. Laboratory measured olivine spectra, Mars derived Surface Type 1 and 2 [Bandfield et al., 2000]. (c) Ostrov THEMIS spectra from end-member regions. (d) Alga DCS (9,6,4) highlighting the mafic mineralogy. (e) Alga Peak Unit 1 end-member map. The end-member has the distinct feature at 11μm consistent with the olivine lab spectra. (f) Alga Peak Unit 2 end-member map. (g) Alga Floor Units end-member map. (h) Ostrov DCS (9, 6, 4) highlighting the mafic mineralogy. (i) Ostrov Peak Unit 1 end-member map. (j) Ostrov Peak Unit 2 end-member map. K) Ostrov Peak Unit 3 end-member map.

Download figure to PowerPoint

[34] THEMIS analysis of Ostrov central peak (Figure 15) can distinguish three spectra units. The first maps to the CRISM olivine-bearing units and is dominated by olivine, but is not a spectral match to pure olivine, similar to the Alga olivine unit. The second type maps to the pyroxene-bearing units and is spectrally consistent with high concentrations of LCP; however the coarse resolution prohibits precise and unique mineralogic assignments. The third type has low emissivity in shorter wavelengths (∼8.5–9.5μm) and higher emissivity in longer wavelengths (11–12 μm), consistent with a low abundance of mafic minerals.

4. Regional Analysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[35] North Argyre.The Northern Argyre (NA) region has contains six central peaks in this study that have well-defined, unaltered mafic spectra, including Alga Crater (NA1) and Ostrov Crater (NA2). Those two examples, detailed above, are among the best exposed examples in this study and display a clear lithologic distinction between olivine-bearing and pyroxene-bearing units, with olivine ranging from fayalitic compositions in Alga to intermediate compositions in Ostrov. The pyroxenes for both regions were consistent with a high-Fe, low-Ca composition.

[36] An additional example from this region, NA4, is a 22 km diameter unnamed crater in the Northeast part of Ladon Basin (16.71°S, 28.14°W, Figure 16) that is contained within CRISM observation FRT000097FF. This example highlights the difficulty in determining the original context of excavated peak material. The crater impacts into the Ladon Basin that was strongly modified and filled. The Ladon Basin itself sits on crust strongly modified by previous impact and volcanic processes. This makes it difficult to determine the original context and origin of the crustal materials exposed in the crater's central peak. The central pit of NA4 is filled with aeolian materials, but the pit rim, particularly the northeastern portion, is comprised of mafic-bearing blocks. Light-toned olivine-bearing units form much of the northern and eastern pit rim, while pyroxene units dominate the western and southwest sides of the central pit. Two olivine-bearing regions were successfully modeled with compositions of Fo20 and Fo16 (Table 3). These values are among the most fayalitic in this region but are comparable to the range from Alga Crater (NA1). Three pyroxene regions were modeled with 2 μm NBSR values consistent at 0.56 (Table 4) with absorption bandcenters just above the LCP end-member values supporting the LCP enrichment of the pyroxene-bearing units.

image

Figure 16. North Argyre example region. Ladon Basin context (Figure 1a, inset) with arrow pointing to crater that contains this central pit structure. (a) NA4 is one of the few sites that have a central pit. (CTX). (b) Mineral map showing mafic deposits on central pit structure. (CRISM FRT00097FF Parameter Map R:Olindex2 G:LCPindex B:HCPindex on CTX). (c) Ratioed Olivine Spectra. (d) Ratioed Pyroxene Spectra.

Download figure to PowerPoint

[37] North Hellas. The North Hellas (NH) region contains nine well-defined and exposed central peaks covering a wide spatial distribution. Here we describe one example, NH6, a relatively small (20 km) crater on the northern edge of Hellas Basin (31.43°S, 81.63°W,Figure 17). The upper layers of the central peak, just below peak's apex are distinguished by a light-toned olivine-bearing lithology that is preferentially eroded from the darker surrounding material. The apex of the peak retains this darker material with a distinct pyroxene signature. One olivine region was successfully modeled yielding a three band fit with a composition of Fo8 (Table 3). This observation is the most Fe-rich olivine-bearing site from the NH region, which are all consistently fayalitic within the fifteen olivine-bearing NH regions and show a full range from Fo8 to Fo38 (Table 5). One pyroxene-bearing region from NH6 was modeled with an intermediate 1μm NBSR (0.48) and LCP-rich 2μm NBSR (0.73) (Table 4). While some variation does exist the pyroxenes from the Northern Hellas region consistently show a moderate to high LCP enrichment.

image

Figure 17. North Hellas example region. (a) NH6 is a classic central peak structure with a ring of light-toned olivine and a peak of pyroxene-bearing material. (b) CRISM Projected RGB. Observation is 10 km across at center. R:2.5μm G:1.5 μm B:1 μm (FRT0001EBA0). (c) CRISM projected mineral map showing mafic deposits on central pit structure. CRISM Parameter Map R:Olindex2 G:LCPindex B:HCPindex. From CRISM: FRT0001EBA0. (d) Ratioed Olivine Spectra. (e) Ratioed Pyroxene Spectra. From CRISM: FRT0001EBA0.

Download figure to PowerPoint

Table 5. Summary of Modeled Olivine Composition Resultsa
 RegionsFo RangeFo MeanFo SD
  • a

    Region; Number of modeled spectra in the region; Range of modeled values; Mean modeled composition of all regions; Standard deviation of modeled compositions. Table summarizes full results in Table 3.

NA Olivine187–703423
NH Olivine158–38187
NF Olivine65–14104
NP Olivine877–583215

[38] Nili Fossae. The third region examined is the smallest current sample set with only two sites due to the high degree of alteration in this area [Ehlmann et al., 2009]. The Nili Fossae (NF) region is classified as Southern Highland terrain but is located on the western edge of the Isidis basin and just south of the dichotomy border. The Nili Fossae region has been well examined due to well-exposed bedrock and a diverse geologic history [e.g.,Hoefen et al., 2003; Mangold et al., 2008; Fassett and Head, 2005; Mustard et al., 2009; Ehlmann et al., 2009]. Both craters are similar in size and each contains exposures of olivine and pyroxenes. Here we detail Hargraves (NF1) (20.76°N, 75.81°E, Figure 18) with a central pit structure that is dominated by pyroxene-bearing bedrock with a few small olivine outcrops around the pit. Unlike most of the other examples, the olivine-bearing material does not correlate to a clear difference in tonality. The modeled olivine region in this central peak had a composition of Fo14, while the range for the six modeled regions in NF have a range of Fo5–14, again showing a strong fayalite enrichment. Pyroxene modeling of six regions in NF1 results in range of a 2 μm NBSR of 0.56–0.63 with a similar range for the 1 μm, again consistent with a slight enrichment in LCP.

image

Figure 18. Nili Fossae example region. (a) NF1 CTX overview of Hargraves crater. (b) CRISM projected mineral map showing mafic deposits on central pit structure. Observation is 10 km across at center. R:Olindex2 G:LCPindex B:HCPindex. (c) CRISM Projected RGB. R:2.5 μm G:1.5 μm B:1 μm (FRT0000B573). (d) Ratioed Olivine Spectra. (e) Ratioed Pyroxene Spectra. From CRISM: FRT0000B573.

Download figure to PowerPoint

[39] Northern Plains. The North Plains (NP) region includes six craters with concentrations in Acidalia Planitia (NP1, NP2) and Utoipia Planitia (NP3-NP6). Two sites, Kunowsky Crater (NP1) and Stokes Crater (NP5), are relatively large at 60 km, while the other four are 30 km or less, all of which are large enough to penetrate the Hesperian basalts that resurfaced the lowlands and that are estimated to be hundreds of meters thick [Head et al., 2002]. The four smaller craters all contain well-formed and preserved central peak structures surrounded in part by dark, olivine bearing dune materials. These are identified by the low albedo compared to the noticeably brighter olivine bedrock. In selecting regions of interest we choose spatially continuous mafic units based on the spectral parameter maps. With this method, central peaks in the Southern Highlands contained at most nine modeled olivine regions and in some cases only one or two regions. In contrast, the large NP craters, Kunowsky (NP1) and Stokes (NP5) have 17 and 21 olivine-bearing outcrops respectively.

[40] Stokes Crater (NP5) (Figure 19) contains one of the more complex central peak structures examined in this study. The peak itself is offset to the north of center of the crater with depressions to the south and east of center. Large diffuse light-toned olivine-bearing units outcrop from the peak and edge of the southern basin. In between the olivine exposures on the peak are small pyroxene-bearing units. The eastern depression contains dark olivine-bearing dunes which overlie light-toned, highly fractured megaclasts with strong olivine signatures. The 21 modeled olivine regions yield a diverse compositional range of Fo15–65 ranging from mildly forsteritic to strongly fayalitic. The 2 μm pyroxene NBSR values for the 16 modeled regions range from 0.53 to 0.62 with values for the 1 μm absorption about 0.1 less (Table 4). Stokes Crater fits into the pattern we are observing across the planet with light-toned olivine breccia interspersed with pyroxene bedrock units.

image

Figure 19. Northern Plains example region. (a) NP5 Context image of Stokes Crater (CTX). (b) CTX view of central peak. Light-toned small clasts in upper right are classic olivine-bearing breccias. Lower right light material are olivine bearing particulate material not used for olivine analysis. (c) CRISM projected mineral map showing mafic deposits on central pit structure. Observation is 10 km across at center. R:Olindex2 G:LCPindex B:HCPindex (CRISM FRT0000ADA4). (d) CRISM projected RGB. R:2.5μm G:1.5 μm B:1 μm (CRISM FRT0000ADA4). (e) Ratioed Olivine Spectra. Poor registration from the S detector are unable to resolve the 1 μm pyroxene absorption, causing increased error with three band olivine fitting. This is a case when the two band fits would provide more confident results. (f) Ratioed Pyroxene Spectra. From CRISM:FRT0000ADA4. NP5. (g) Light-toned olivine-bearing breccias exposed in the depression just east of the Stokes Crater center. (h) Pyroxene-bearing bedrock material exposed north of the Stokes Crater center. HiRISE: ESP_016980_2360.

Download figure to PowerPoint

[41] These regional examples show strong similarities among the observed central peak mafic units. In all the locations we observed megaclasts with no evidence of spectral mixing. The observed olivine range from moderately forsteritic to strongly fayalitic and the pyroxene units are consistently enriched in LCP. Here we present the composition results of all the regions to determine any regional or global trends.

[42] Nature of Olivine Compositional Trends. MGM modeling of the olivine regions from all of the sites show a wide range of values from Fo5 to Fo70 (Figure 20a and Table 5) with a mean value of Fo28for all regions. This reinforces the Fe-rich fayalitic nature of the observed olivine units but highlights the diversity in the compositions. NA olivine-bearing units have a mean of Fo34±23 (Figure 20b). This region has the highest Fo of the four regions as well as the largest range. NH has a mean of Fo18±7. The NH sites (Figure 20c) are systematically fayalitic with only a small variation indicating a clear difference from the composition distribution in NA. NF has only 6 modeled regions and contains the most fayalitic values with a mean of Fo10 and a narrow 4 Fo unit range. The NP region (Figure 20d) has a composition distribution similar to NA with a mean of Fo32±15, and can be subdivided into the Acidalia and Utopia populations. The Acidalia region (NP1, NP2) has a range of Fo11–58 with a mean of Fo43, while Utopia (NP3, NP4, NP5, NP6) has a range of Fo7–56 with a mean of Fo26. The olivine in the Acidalia region is significantly more Mg-rich than Utopia highlighting the regional differences that are emerging from the observations (Table 4).

image

Figure 20. (a) All Outcrop Olivine Distribution. Plot shows the modeled composition for all examined olivine regions. (b) Argyre Olivine Outcrop Distribution. (c) Hellas Olivine Outcrop Distribution. (d) Northern Plains Olivine Outcrop Distribution. All plotted data is in Table 3.

Download figure to PowerPoint

[43] Pyroxene Bandcenter Analysis. Modeling the pyroxene absorptions with 2 absorptions to get bandcenter allow the comparison of regional compositions to terrestrial pyroxene values. Figure 21 plots the modeled bandcenters against a set of laboratory measured CPX and OPX [Adams, 1974; Cloutis and Gaffey, 1991]. The 2 μm bandcenters for the NF (Mean:1897 nm) and NA (Mean:1922 nm) regions are lowest and consistent with low-Ca, moderate-Mg pyroxene end-member (pigeonite). The NH area pyroxenes have higher 2μm bandcenters (Mean:1962 nm) indicating higher Fe content but still low-Ca (ferrosilite). The 2μm bandcenters for the NP (Mean:1965 nm) have a wide spread but are generally elevated compared to the Southern Highland pyroxenes with many comparable to moderate Ca compositions (pigeonite).

image

Figure 21. Pyroxene Bandcenters: 1-, 2μm bandcenters of all pyroxene regions plotted against terrestrial pyroxene bandcenters [Adams, 1974; Cloutis and Gaffey, 1991] by region. From bottom left to upper right, pyroxene composition goes from Mg-rich to Fe-rich to Ca-rich. Plot of modeled pyroxene bandcenters plotted against terrestrial laboratory measured bandcenters. Significant variation exists on the 1μm axis while the 2 μm axis is well modeled. While variation exists in each region, averages vary from low-Fe, low-Ca toward high-Fe, high-Ca in the order of North Argyre, North Hellas, Nili Fossae, and Northern Plains. Variation in 1μm modeled could be a contribution of olivine in the spectrum or L-S detector alignment problems. The addition of glass could also shift the bandcenters to longer wavelengths. All plotted data is inTable 4.

Download figure to PowerPoint

[44] Pyroxene NBSR Analysis.MGM modeling of the pyroxene unit spectra with five absorptions allow for the determination of the relative proportions of low and high-calcium pyroxene components by calculating the NBSR value for both the 1 and 2μm absorptions (Figure 22 and Table 6). This provides an additional composition consistency check as a well measured and modeled pyroxene should have consistent NBSR values for both absorptions. Here we observe an agreement between the 1 and 2 μm was within 0.04 for the Southern Highland sites and 0.08 for the Northern Lowland sites (Table 6). Comparing between regions highlights additional regional differences in composition. The three Southern Highland regions show a consistent LCP enrichment with a slight difference between regions while the NP sites have a noticeable HCP enrichment.

image

Figure 22. Pyroxene NBSR. LCP/(LCP+HCP) Band depths. NBSR for all modeled pyroxene observations, colored by region. Significant variance exists on the 1 μm axis while 2 μm NBSR shows regional differences. In order of decreasing LCP, the regions rank North Argyre, North Hellas, Nili Fossae and Northern Plains. Data summarized Table 2 and detailed in Table 4.

Download figure to PowerPoint

Table 6. Summary of Modeled Pyroxene Composition Resultsa
 Regions1 μm NBSR Mean1 μm NBSR SD2 μm NBSR2 μm NBSR SD
  • a

    Region; Number of modeled spectra in the region; Mean 1 μm NBSR values; Standard deviation of 1 μm NBSR values; Mean 2 μm NBSR values; Standard deviation of 2 μm NBSR values. Table summarizes full results in Table 4.

NA Pyroxene320.650.140.620.09
NH Pyroxene460.560.090.600.07
NF Pyroxene120.550.080.590.02
NP Pyroxene300.460.070.540.09

[45] A final analysis was performed to look for compositional trends as a function of central peak excavation depth. No detectable trends were determined as either a function of excavation or crustal thickness. The degree of vertical mobility of the central peak materials from the past cratering processes seems to have swamped any remnant compositional stratigraphy that may have existed.

[46] TES Thermal Infrared Constraints. A dedicated analysis of Martian central peaks with the TES thermal IR instrument is currently underway [Pan and Rogers, 2011]. While many of the craters and all of the spectral units detailed in this study are too small for TES analysis, two sites (NA3 and NH3) have been analyzed by Pan and Rogers [2011]. While the results are still preliminary, these two sites were grouped with several others with a similar thermal spectral response and classified as relatively feldspar rich with a collective compositional abundance of 32% Feldspar, 19% LCP, 17% HCP, 0.1% Olivine and 14% High-silica phases with the remaining 18% as accessory phases and blackbody. The spatial resolution differences between instruments and the several central peak averages in this study make a direct comparison difficult but point out the significant feldspar component that is not characterized by CRISM observations and has potential for future development.

5. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[47] Considering the single plate nature of Mars, crater central peaks are the best source of exposed buried crustal material, the history and context of that material is dependent on the complex history of the local area. The basement rocks of Alga crater could include ejecta from Argyre basin to the south and is within the outer rings of Ladon basin and Chekalin crater. This history would have resulted in intense brecciation of the crust, redistribution and burial of impact products (e.g., melts), and potentially basin related volcanism [Schultz and Glicken, 1979]. This local complexity makes constraining the ancient crust from these exposures a difficult task.

[48] Implications of Global Distribution. The selection of these 23 central peak structures was the result of considering CRISM observed unaltered, mafic bearing central peaks. After identification and successful modeling of the mafic spectra extracted from the central peaks, the locations were plotted and the four regional groupings were identified (Figure 2). These resulting groupings are possibly the result of several properties of the Martian surface or may provide deeper understanding of the nature of these exposed units. A similar distribution is observed in the overall distribution of the Crater Exposed Bedrock Database [Tornabene et al., 2010] that considers only the HiRISE morphology and not the mafic spectroscopy, indicating that is it more than just an effect of surface spectral quality. Several surface properties that may contribute to the observed distribution, including dust cover [Ruff and Christensen, 2002], crater density, bedrock exposures and outcrop surface refreshing rates, and the presence and pervasiveness of alteration. For example, the volcanically dominated areas of Tharsis and Elysium lack observations in this study due to the lack of preserved craters and the high dust cover [Ruff and Christensen, 2002] of these elevated regions. While conclusive evidence for this regional discrepancy is elusive, we offer several possible explanations assuming that the observation is real and not solely an artifact of the current CRISM FRT coverage. One hypothesis is that the concentration around the main Martian basins (Argyre, Hellas, Isidis, and Utopia) is not coincidental and that the reported central peaks are excavating ejecta materials that were created and/or redistributed by these basins. The effect of the basin excavation could potentially allow a source from the lower crust or even the Martian mantle. Another possibility is that the regions outside of our listed sites are regions of heavy crustal alteration, so that minimally altered central peaks are no longer observed there. This option is unlikely due to the correlation of the observations in the Crater Exposed Bedrock Database [Tornabene et al., 2010], which are based on morphology and not composition. A third possibility is that the regions without reported observations have lower abundance of well-exposed crustal bedrock. For instance, the lack of exposed mafics at southern midlatitudes (30°S–60°S) may be due to mantling deposits covering the bedrock [Mustard et al., 2001]. This would explain the lack of observations to the south of Argyre and Hellas Basin and toward the polar regions. One final hypothesis is that the bedrock exposures we are describing are not globally distributed. It could be that the textures and morphologies we are documenting are plutonic in origin and that we are observing the excavated remnants of large but localized volcanic plumes occurring below places like the North Argyre and North Hellas areas where the crust has been compromised by the formation of these large basins. Future mapping and additional understanding of the observed lithologies may help decide between these hypotheses.

[49] Lithology Determination. VNIR spectroscopy of mafic minerals is sensitive to cation composition of the dominant mineralogy but may not detect minerals without a distinct near-infrared signature or sufficient abundance. Rock compositional results must then include assumptions about the target materials and an understanding of the VNIR and TIR spectroscopic ranges. One notable limitation with the VNIR spectral range is that the plagioclase absorption feature can only be observed if it contains trace amounts of Fe2+ and is present in extremely high abundances (>∼85%) [Bell and Mao, 1973; Nash and Conel, 1974; Crown and Pieters, 1985; Cheek et al., 2011]. The strong mafic signatures of the units reported here overwhelm any potential signature from plagioclase, limiting our ability to reliably report on its presence, abundance and composition with the VNIR observations. The application of THEMIS IR observations are able to help constrain the silicate compositions though they have spatial and spectral resolution limits. Near-infrared observations are well suited for the detection of Fe-bearing olivine and pyroxene minerals. The compositional determination of the Alga olivine-bearing lithology focuses on the complete absence of a 2μm pyroxene absorption indicating pyroxene content less than ∼5–10% and THEMIS observations show spectra consistent with an olivine end-member. These observations indicate that this unit has a mafic only composition in the range of a dunite (>90% olivine) composition with a fayalite composition ranging from Fo45–7. The coupled THEMIS observations however indicate some degree of feldspar mixing, suggesting a trOctolite-dunite composition.

[50] The pyroxene-bearing lithologies are interpreted using similar reasoning. The 1μm region can be fully modeled with the expected pyroxene absorptions without requiring an olivine component at slightly longer wavelengths. VNIR observations indicate nearly pure pyroxene with olivine content less than 5%, suggesting a pyroxenite lithology (although spectral analysis cannot rule out a borderline noritic composition). This is consistent with but not uniquely confirmed by the THEMIS observations. The band centers of the observed pyroxenes are both in the low range on the pyroxene continuum indicating low-Ca content but the relativity long-wavelength band center of the 1μm absorption, while still in the LCP range, suggests a Fe-enrichment. While there is some spread to the observations they are generally consistent with an enrichment of low-calcium and relatively high-Fe cation content. These results would signify an average composition in the range of a Fe-rich pigeonite. However, large scale spatial mixing would have difficulty discriminating between pure pigeonite and an intimate mixture of end-member minerals such as an augite-ferrosilite mixture.

[51] The light-toned pyroxene unit also shows no evidence of VNIR spectral mixture with other minerals and is similar in cation composition to the pyroxene-bearing bedrock with weaker band depth. We suggest that this is most easily explained as a mixture with the pyroxene-bearing bedrock and a VNIR spectrally featureless material, possibly the impact melt deposited on or injected into the exposed bedrock of central peak, or potentially, a plagioclase phase that would increase reflectance but not affect the VNIR observation. This would have the effect of attenuating the absorption feature band depth but keeping the band centers fixed as observed [Tompkins and Pieters, 2010]. Current thermal observations are too coarse to determine the pyroxene proportion of the unit or the bright phase.

[52] The impact melt unit is dominated by a bland spectral signature with a small proportion of pyroxene causing weak absorptions. The position of the fine grained impact melt unit in the overall unit stratigraphy is still uncertain, but we suggest two possible scenarios. The first is that the impact melt formed from the pyroxene bedrock unit during the formation of Alga crater and it covered a portion of the central peak, entraining clasts of pyroxene-bearing material. This scenario is supported by the presence of possible flow like features (seen inFigures 6 and 11) that could represent the remnants from the early flow of these hot materials, prior to solidification. However, Alga crater over time has been eroded and may be expected to wear away a surficial veneer of impact melt. The second scenario is that an impact melt component was produced in a previous basin scale impact event (e.g., Ladon or Argyre basin formation) and was emplaced as part of deep crustal stratigraphy and subsequently excavated by Alga. The relatively weak nature of the melt would cause preferential erosion and the observed recession of this unit.

[53] The central peak of Ostrov crater has only two discernible units and, unlike Alga, these are only easily distinguished with spectral mapping. The units are an olivine-bearing unit seen in three locations and a pyroxene-bearing unit comprising the rest of the central peak (Figure 13). Similar to the Alga units, we observe no mafic mixing in the VNIR or TIR, again indicating that the units are dominated by a single mafic phase and are interpreted to be dunite (or trOctolite) and orthopyroxenite (or norite) respectively. The olivine-bearing unit has significantly higher Fo# than the Alga olivine units. While it is difficult to draw conclusions about the exposed crust from two examples, we note compositional diversity and relatively high Fe contents compared to most of the reported values from other Martian olivine samples, typically ranging from Fo30–80 [Hoefen et al., 2003; Koeppen and Hamilton, 2008; McSween et al., 2006; Edwards et al., 2008; Tornabene et al., 2008]. The Ostrov pyroxene-bearing unit is compositionally similar to the Alga pyroxene unit characterized by high-Fe, low-Ca pyroxene.

[54] These two crater central peaks offer a look into the composition of the deep crust of Mars. While not representative of the whole crust, we begin to see some important trends. The first is that in both examples we observe cumulate outcrops, dunites and pyroxenites, with no evidence of mixtures of these minerals. This may indicate that the ancient crust was largely formed in large, slow cooling magma bodies similar to terrestrial large mafic intrusions [Cawthorn, 1996] rather than a series of quick cooling lava flows. The second is that the olivine displays a range of compositions from intermediate to fayalitic, with a significant Fe-enrichment compared to other measured Martian olivine. Finally, the observed pyroxenes seem to be dominated by a high-Fe, low-Ca calcium composition. This is consistent with the high Fe values seen in the olivine-bearing units and with the LCP pyroxenes seen in Noachian terrains throughout Mars [Mustard et al., 2007].

[55] The observations that the ancient Martian crustal material has been mineralogically segregated with a range of Fe-rich compositions indicate that it had a crystallization history fundamentally different than the basalts that cover most of the surface [Rogers and Christensen, 2007; Poulet et al., 2009]. This ancient crust would have had to experience crystal segregation into the observed dunites and orthopyroxenites. We note the similarity between these hypothesized cumulates and the units described by Francis [2011] for the Columbia Hills in Gusev Crater. These layers would have to be thick enough to become brecciated clasts several hundreds to thousands of meters across and must occur close enough in proximity that a single central peak can excavate both units. Only a few known methods exist for segregating minerals on large scales and we examine the possibilities.

[56] Crustal Formation. The complex cratering history of Mars makes it difficult to reconstruct the original stratigraphy of the exposed units, however only a few mechanisms are known to segregate mafic lithologies at kilometer scales. These include dunite channels in harzburgite as seen in ophiolites [Kelemen et al., 1995] and olivine deposition from komatiite surface flows [Walter, 1998] or melt fractionation of basin induced impact melts [Warren et al., 1996]. The first example has only been observed as a product of terrestrial plate tectonics in mid-ocean ridges, a feature not observed or expected on Mars. The second example would be unlikely to produce mineralogically segregated units thousands of meters thick as required by the spatial area of the observed blocks. The third option is viable and would potentially be difficult to distinguish from an internally erupted lava lake. The most commonly observed method for mineral segregation on Earth is a cumulate settling processes forming a layered mafic sequence. These features, well studied in variety of terrestrial locations from the Bushveld complex in South Africa to the Skaergaard intrusion in Greenland [Cawthorn, 1996] can produce the observed Martian lithologies at the observed scales. One well studied example is the Bushveld Complex in South Africa [Cawthorn, 1996]. The Bushveld complex has been divided into four zone from top to bottom, Upper, Main, Critical and Lower. We will specifically focus on the Upper Zone, the one that would be the most likely to be excavated by impacts. This unit contains olivine that range from Fo63–5, a similar range observed in the Mars observations presented here. Orthopyroxenes in the Upper Zone range from a mg# 60–Mg#30 with most areas having values in the mid 50s mg#. The entire Bushveld complex is depleted in clinopyroxenes, again similar to our presented Mars observations. However, the Bushveld is only illustrative of the process that could have occurred on Mars. The Bushveld is the product of multiple igneous injections and a complex history that would be impossible to resolve on Mars with current observations.

[57] Several alternative possibilities could explain the observed compositions. These include that the basin forming impacts (Argyre, Hellas, Isidis, Utopia or Borealis [Andrews-Hanna et al., 2008]) excavated the mafic cumulates of the lower crust and possibly the upper mantle. The post-overturn upper mantle predicted byElkins-Tanton et al. [2005]would be Mg-rich bearing olivine and pyroxene. Our measurements instead show a range of intermediate to low Mg#. Excavation of an un-overturned or partially overturned mantle unit could produce consistent results but would be unlikely given the gravitational drive of the overturn for the former and the unlikely chance of only sampling the overturned remnants in the latter. Another option that we will explore in detail is the formation of an igneous secondary crust that would fit the modeled constraints of the overturn and the crustal observations detailed here. The overturn would result in diapirs of Mg-rich cumulates buoyantly rising through the early Martian mantle. These diapirs would have formed from the earliest Martian cumulates, dominated by forsterite (Mg2SiO4) and enstatite (MgSiO3). The rising cumulates would experience decompressional melting as the diapir passes the early Martian adiabat forming a partial melt. The initial melting would concentrate on minerals with incompatible elements and the most Fe-rich minerals. Rising diapirs would potentially encounter a proposed garnet-bearing mantle layer near 1050 km depth, providing a source of Al that would facilitate the eventual crystallization of a plagioclase component commonly observed in younger crusts on Mars [Bandfield et al., 2000]. Early crustal observations are consistent with HCP depletion and no garnet detections implying the partial melting of the diapirs past the CPX-out phase of 10% at low pressures [Walter, 1998]. The common observation of OPX and olivine suggest a partial melting in the 10–23% range [Parman and Grove, 2004]. The resulting melt would be less dense than the rising cumulate diapir and begin its own buoyant accent to the surface. Current observations cannot distinguish between surface eruptions forming lava lakes or near surface intrusions (Figure 23) forming the crust with serial magmatism [Longhi, 2003]. Observations only require that the magma bodies are large enough to fractionally crystalize into cumulate layers hundreds to thousands of meters thick and be close enough to the surface to be brecciated and excavated by impact craters. This model of crustal formation explains the observed mineralogically segregated lithologies, the Fe-rich but highly fractionated olivine compositions and the regional differences in composition values that are dependent on minor mantle chemical heterogeneities and varying diapir ascension histories.

image

Figure 23. Proposed model of crustal formation. (bottom) Mantle density instabilities would lead to overturn with rising diapirs of Mg-rich minerals. Decompression melting would preferentially melt the Fe-rich faction which would buoyantly rise faster than the diapir. Large volume eruptions would cool slowly and stratify into cumulate layers. Figure 23 (bottom left) illustrates magma eruptions as lava lakes. Figure 23 (bottom right) illustrates pluton cooling/serial magmatism. (top) Extensive cratering would brecciate cumulate layers, mixing and excavating mafic lithologies.

Download figure to PowerPoint

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[58] A planet wide survey of Martian central peaks with well exposed mafic outcrops as determined from CRISM results in 23 examples of deeply excavated Martian crust. The exposures are grouped into three regions in the Southern Highlands; Northern Argyre, Northern Hellas, Nili Fossae and several sites in the Northern Plains. In each of the exposures we observe olivine and/or pyroxene-bearing outcrops, including the impact melt unit, with little spectral evidence of mineral mixing in the VNIR, and the possible occurrence of plagioclase from TIR observations. This suggests that the deep crust is comprised of mineralogically segregated units. Olivine outcrops show a range of compositions from strongly fayalitic to moderately forsteritic. Pyroxene outcrops in the Southern Highlands show a relative enrichment in LCP with the Northern Argyre and Nili Fossae regions being more Mg-rich than the North Hellas sites and the Northern Plains.

[59] While this crust sampling method could be excavating younger igneous rocks emplaced in plutons or dikes, it is unlikely that in each of the examined cases we are only excavating younger material. The general consistency in the observations does not indicate a clear signal that a single exposure is of unique origins. While excavating a younger unit is possible the most likely result on a single plate planet is the excavation of a voluminous ancient crust. The deep crust compositional observations are best explained by a crustal formation model that segregates mafic minerals into cumulate bodies. The global distribution of these materials indicates a planetary scale crustal formation process. One such reasonable driver of such an event would be the decompression partial melting of rising diapirs of Mg-rich mantle cumulates in an early planetary overturn process. This crustal formation process would result in a compositionally stratified ancient crust that would be brecciated, incorporated with subsequent volcanics and aqueous alteration to create the crust we observe today.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[60] We are very grateful for the fine work of the NASA MRO project team and the fine job by the CRISM Science Operations Center (SOC). This work was supported by NASA through a subcontract with the Applied Physics Lab at Johns Hopkins University. The themes and techniques presented here were greatly improved by conversations with Peter Isaacson and Carle Pieters. The manuscript benefited by suggestions from Sandra Wiseman, Leah Cheek, and Steve Ruff. We thank Ed Cloutis and an anonymous reviewer for substantially improving the manuscript.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Analysis Techniques
  5. 3. Type Region Results
  6. 4. Regional Analysis
  7. 5. Discussion
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
jgre3121-sup-0001-t01.txtplain text document2KTab-delimited Table 1.
jgre3121-sup-0002-t02.txtplain text document0KTab-delimited Table 2.
jgre3121-sup-0003-t03.txtplain text document4KTab-delimited Table 3.
jgre3121-sup-0004-t04.txtplain text document9KTab-delimited Table 4.
jgre3121-sup-0005-t05.txtplain text document0KTab-delimited Table 5.
jgre3121-sup-0006-t06.txtplain text document0KTab-delimited Table 6.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.