Determining olivine composition of basaltic dunes in Gale Crater, Mars, from orbit: Awaiting ground truth from Curiosity



[1] The successful landing of the Mars Science Laboratory Curiosity rover in Gale Crater, Mars, presents a rare opportunity for validation of a spectral index developed for determining olivine chemistry from orbital midinfrared remote-sensing data. Here, a spectral index is developed using laboratory emissivity data of 13 synthetic Mg-Fe olivines. Utilizing this spectral index, a prediction of olivine composition (~Fo55 ± 5) is made from orbital data for a NE-SW trending dune field near the Curiosity rover. This dune field will be crossed during the mission as the rover travels toward a ~5 km-high sediment stack (Mount Sharp) that contains orbitally detected clays and sulfates. Curiosity can use its instrument suite (ChemMin, Alpha Particle X-ray Spectrometer, ChemCam) when it reaches the dunes to verify or refute the olivine-chemistry prediction presented here. The ability to validate the developed spectral index using the rover's ground-truth instruments will strengthen olivine-chemistry mapping across the Martian surface using this spectral index.

1 Introduction

[2] On 6 August 2012, the Mars Science Laboratory (MSL) rover Curiosity landed in Gale Crater (4.59°S, 137.44°E) to begin a planned nominal 2 year mission to survey the rocks, sediments, and atmosphere of Mars. Images from orbit (Figures 1a and 1b) and from the surface (Figure 1c) show dunes near the rover's landing site. Curiosity will pass through this dark dune field that extends from a large sand sea west of the central mound to get to the ~5 km high stack of sediments in the center of the crater that will be investigated during the course of the mission.

Figure 1.

(a) Oblique view of Gale crater, derived from a combination of data from the Mars Express High Resolution Stereo Camera (HRSC) and the MRO Context Camera. White dot indicates the landing site of the MSL rover. Image credit: NASA/JPL-Caltech/ESA/DLR/FU Berlin/MSSS. (b) A section of High-Resolution Imaging Science Experiment (HiRISE) image ESP_18854_1755_RGB of the dark basaltic sand dunes that Curiosity would cross on the way to the crater's central mound. Image strip is just over 1 km wide. Image credit: NASA/JPL-Caltech/University of Arizona. (c) Section of a Mast Camera image mosaic showing the view from the landing site, including dark purported olivine-enriched basaltic dunes and Mount Sharp in the background. Image credit: NASA/JPL-Caltech/MSS.

[3] The dune field appears to be a constant lithology because many of its properties are approximately uniform across it, including the dark tone, high thermal inertia [530–740 J m−2 K−1 s−1/2], and general elevation (i.e., all are trapped at the lowest part of the crater [Pelkey et al., 2004]). These dunes are active [Silvestro et al., 2013], having a low amount of fine dust cover in an otherwise high-dust area [Rogers and Bandfield, 2009]. A Mars Odyssey Thermal Emission Imaging System (THEMIS) decorrelation-stretched (DCS, Gillespie et al. [1986]) image mosaic also shows the dune field as generally similar false-color, indicating mineralogic uniformity, and as different from the surrounding nondune lithologies shown as different colors (Figure 2). These characteristics and the ample size of the dune field through which Curiosity will cross (~1–2 km wide and 35 km long) enable the dune field's spectral nature to be studied from orbit using the Mars Global Surveyor (MGS) Thermal Emission Spectrometer (TES) data.

Figure 2.

(a) THEMIS DCS image mosaic, where red, green, and blue bands are assigned to bands 8, 7, and 5, respectively. The band centers are at 11.79, 11.04, and 9.35 microns (848, 905.8, and 1069.5 cm−1), respectively. Basaltic dunes in the crater appear as pink/magenta and are distinct from the surrounding terrains. THEMIS spatial resolution is approximately 100 m/pixel. White arrow points to the MSL landing site. White lines indicate 2 possible dune-crossing strategies. Image credits: NASA/JPL-Caltech/Arizona State University. (b) Spectral index maps of olivine for Fo40, Fo55/(50), and Fo65. Scale bar represents olivine index value (not actual percentage). Blue represents the lowest index values; oranges and reds represent the highest olivine index values. Comparison among the panels shows that higher values occur for the Fo55/(50) index, suggesting the olivine in the Gale Crater basalt dunes is more similar to Fo55/(50) than to either of the other olivine compositions. TES spatial resolution is approximately 3 × 6 km/pixel. Index maps were constructed using JMARS software [Christensen et al., 2009].

[4] Visible/near-infrared data from the Mars Reconnaissance Orbiter (MRO) Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) instrument identified the dunes as mafic, containing olivine and high-Ca pyroxene [Milliken et al., 2010]. Rogers and Bandfield [2009] assessed the lithologies of the Gale Crater units using Mars Odyssey THEMIS and MGS TES data and found the dune field to be similar to “Surface Type 1” (ST1, Bandfield et al. [2000]), which is basaltic; however, the THEMIS band at ~11 µm (909 cm−1) was deeper, related to an additional olivine component.

[5] Using our previously acquired synthetic olivine emissivity spectra [Lane et al., 2011], a spectral index is developed in this work for identifying the Mg-Fe chemistry of olivine. The spectral index is then used to map the extent and composition of olivine in the dunes of Gale Crater through the analysis of TES data.

2 Methodology

2.1 Olivine Spectral Index

[6] Synthetic olivine pellets (from pressed powders) ranging in composition from the Mg2SiO4 end-member (forsterite, Fo100) to the Fe2SiO4 end-member (fayalite, Fo0) (including Fo89.5, Fo80, Fo75, Fo70, Fo65, Fo60, Fo55, Fo50, Fo40, Fo30, Fo20, and Fo10) were analyzed using thermal emission spectroscopy to study the effects of Mg-Fe solid solution [Lane et al., 2011]. The Fo60 pellet fell apart and was too small to obtain a good spectrum. In that study, it was observed that olivine fundamental spectral bands gradually change in position (and strength) as a function of Fo#, from Mg2SiO4 at larger wave numbers to Fe2SiO4 at smaller wave numbers. Using shifted fundamental band positions for compositional analysis has been widely practiced in the past [e.g., Pieters, 1982; Gaffey et al., 1993; Lucey et al., 1998; Christensen et al., 2000; Hoefen et al., 2003; Hamilton and Christensen, 2005; Rogers and Christensen, 2007; Sunshine et al., 2007; Koeppen and Hamilton, 2008; Edwards et al., 2008; Bandfield and Rogers, 2008]; however, Lane et al. [2011] discuss the systematic shifting of an additional feature in olivine spectra due to compositional changes. This additional feature is a local emissivity maximum (a convex-upward bend) that occurs between two fundamental bands (Figure 3). It is the shifting of this flection position (farther shifting than any of the fundamental bands) that is utilized here for the development of a spectral index for remote identification of olivine composition.

Figure 3.

Thermal emissivity spectra of synthetic Mg-Fe olivines across the solid solution series. Bands 9 and 12 (equation ((1))) and the flection position (dot) are identified and are shown to shift with composition and vary in strength, and are described in Lane et al. [2011]. Arrows indicate the general concavity of the spectra at the flection position. Spectra are offset for clarity. Some water vapor appears as noise (jaggedness) in the spectra at long wavelengths.

[7] Spectral bands for each olivine composition are defined, and their positions related to composition are presented in Lane et al. [2011] (Table 1 and Figure 1 therein). The spectral index based on the flection position [Lane et al., 2011, Table 7 and Figure 11] follows the general formula of

display math(1)

where εfp is the measured emissivity value at the flection position, εBand# is the emissivity value at the Band 9 and Band 12 positions (described at length in Lane et al. [2011], Band 9 ranges from 472 to 507 cm−1, Band 12 ranges from 363 to 420 cm−1, and the flection position ranges from 398 to 486 cm−1, as shown in Figure 3). Because the flection position and the positions of Bands 9 and 12 shift for each olivine composition (Figure 3), there is a different specific formula for each Fo#, but they all follow the general formula. Higher index values represent higher abundances of olivine for a given Fo#. In order to apply the developed spectral index to interpreting the TES data from Mars, the laboratory spectra (at ~2 cm−1 spectral resolution) were degraded to the ~10 cm−1 spectral resolution (using TES filter functions, Christensen et al. [2001]) to identify correct TES bands for the index formulae. Because of the coarseness of the TES spectral resolution, some Fo#s are represented by an equation identical to a neighboring Fo#. This was the case for Fo50&55 as well as Fo70&75, indicating a minimum error in compositional determination of ±5 Fo#.

Table 1. TES-Derived Abundances of Minerals/Surface Types for the Low-Albedo Dunes in Gale Crater
This Study (NE-SW Trending Dune)Rogers and Bandfield [2009]
  1. aWhere the feldspar (20%) and pyroxene (30%) values are combined into an assumed Surface Type 1 (ST1) composition. ST1 is from Bandfield et al. [2000].
51% ST150% ST1a
22% Olivine (Fo55)15% Olivine
7% High silica19% High silica
8% Sulfate12% Sulfate
12% Other (including 10% Fo70)4% Other

2.2 Mapping the Olivine-Rich Dunes in Gale Crater Using the Spectral Index

[8] Using Java Mission-planning and Analysis for Remote Sensing (JMARS) software [Christensen et al., 2009], the olivine index was applied to TES data, and compositional maps of Gale Crater were produced. The TES data were constrained by surface temperatures of ≥255 K, emission angles of ≤30°, and wave number resolution of 10 cm−1. The cooler colors (blues) represent less olivine of the given composition, whereas areas of warmer colors (oranges and reds) indicate more olivine due to higher olivine index values. Figure 2b shows the results of the applied olivine index.

3 Results

[9] Figure 2b shows that for Fo55/(50), the olivine index values are the highest overall and have the best correspondence to the dark sands as shown in Figure 1a. For comparison, the basaltic sands in the flanking Fo40 and Fo65 index maps are not nearly as well defined. Hence, the prediction is made that the enhanced olivine of the basaltic dunes in Gale Crater is dominated by Fo55 ± 5 composition.

[10] To test this composition another way, linear spectral unmixing [Ramsey and Christensen, 1998] of an average of 3 select dune-crossing TES spectra (constrained by warm surface temperature, low emission angle, low ice clouds, etc.) was performed using an end-member spectral library (similar to that used in Rogers and Bandfield [2009], with the inclusion of our full synthetic olivine spectral suite) consisting of an array of silicates, glasses, carbonates, sulfates, and other alteration minerals, as well as spectral end-members derived for Mars, including atmospheric components (CO2 vapor, dust, and water ice clouds) required for deriving the surface spectrum [Bandfield et al., 2000]. This low-albedo unit was found to be high in plagioclase, pyroxene, and olivine, consistent with an olivine-rich basaltic sand. In order to highlight the enhanced olivine signature of the dunes, spectral unmixing of the surface unit also was performed by including Mars' widely mapped ST1 and Surface Type 2 (ST2) spectral end-members [Bandfield et al., 2000], keeping the variety of secondary phases, and restricting the primary silicate phases presumed to be part of ST1. The results (Table 1) show the NE-SW trending dune field to be dominated by ~51% ST1 and an additional ~22% olivine of Fo55 composition. Additional components were indicated, including ~10% Fo70 olivine, ~8% sulfate, ~6% Si glass, ~2% smectite, and <1% carbonate; however, mineral species identified at or below ~10% [Rogers and Christensen, 2007] (and possibly up to ~15%, Christensen et al. [2000]) are questionable. The retrieved surface spectrum and the model fit are shown in Figure 4, as compared to the spectral of ST1, ST2, and Fo55. Our retrieved surface spectrum is similar to the retrieved surface spectrum of the basaltic unit in Rogers and Bandfield [2009]; however, our spectrum (Figure 4) shows a deeper olivine feature at ~900 cm−1. Our results are consistent with the conclusion of Milliken et al. [2010] who identified olivine and high-Ca pyroxene (likely in ST1) using CRISM data and of Rogers and Bandfield [2009] (Table 1) who found the dark dunes to be similar to ST1 with an additional olivine component.

Figure 4.

Thermal emissivity spectra of ST1 and ST2 [Bandfield et al., 2000] and Fo55 olivine [Lane et al., 2011] (offset by −0.1) for comparison to the retrieved surface spectrum and model fit (offset by −0.04) of the NE-SW trending dune field in NW Gale crater. The spectral gap occurs where atmospheric CO2 disallows viewing of the Martian surface. Spectral unmixing results indicate the linear dune field is dominated by ~53% ST1 and an additional ~21% olivine of Fo55 composition.

[11] Table 1 shows that the composition estimates derived from this work of the NE-SW trending dune field are slightly higher in olivine and lower in high-Si phases than that of Rogers and Bandfield [2009], but other preliminary unmixing analyses we conducted suggest that the NE-SW trending dune field is enriched in olivine relative to the large dune sea west of the central mound, from which the NE-SW trending dune extends, and that the large dune sea may be higher in glassy components. Differences in composition (Table 1) between this study and the results of Rogers and Bandfield [2009], although slight, could be explained by different study targets of TES data, because it is likely that sand sea spectra were analyzed by Rogers and Bandfield [2009], who stated only that the spectra were “from the darker unit.” The possibility of an elevated olivine component in the dune field, relative to the sand sea, may be supported by the THEMIS DCS mosaic (Figure 2a) that shows more intense magenta coloring of the NE-SW trending dune extension. Sullivan et al. [2008] and Mangold et al. [2011] suggest that some dunes on Earth and Mars have a higher apparent olivine content than their basaltic source rocks, indicating olivine enrichment likely due to aeolian grain sorting, which may be the case for these olivine-enriched NE-SW trending dunes near the Curiosity landing site in Gale Crater.

4 Discussion

4.1 Assumptions and Limitations

[12] Several assumptions have been made for this study. First, the olivine spectral index equation and subsequent mapping assumes Mg-Fe olivine and has not accounted for other possible cations (e.g., Ca) in the olivine. Second, this olivine index has not accounted for the spectral effects other minerals in the basalt per se; however, this index strategy has successfully predicted the correct Fo# (±5 at lab resolution) of some olivine-bearing meteorite samples whose whole-rock emissivity spectra were measured (e.g., Yamato 984028 [Dyar et al., 2011], North West Africa 2737 [Pieters et al., 2008], La Paz 04840 and Goalpara (M. D. Lane, unpublished data)). Those correct predictions suggest that in the ~500–390 cm−1 spectral range, the olivine flection spectral characteristic is dominant in olivine-enriched rock spectra or is not significantly altered by the presence of other minerals, for the studied chemistries. Further work will be conducted to test the rigor of this analysis strategy with basalt/meteorite samples containing less olivine, because other rock-forming silicates that are common in basalts display spectral features as well. The MSL rover payload also can help to address this issue.

[13] Lastly, TES channels are coarse enough (~10 cm−1 spacing) [Christensen et al., 2001] that some laboratory spectra of chemistries differing by only 5 Fo# will, when degraded to TES resolution, translate to an identical olivine index equation (i.e., for Fo55/50 and for Fo75/70) or to only a one-TES-channel difference. Thus, there is some loss of sensitivity in determining composition using TES-resolution data as compared to the full spectral resolution of laboratory data. Nonetheless, both the olivine spectral index (using the long-wavelength spectral range of the flection position migration) and the spectral unmixing strategy (using the much broader 200–1300 cm−1 spectral range) have corroborated the dune-extension olivine composition as ~Fo55 at Gale Crater. This unanimity suggests that the coarser TES data can still be used for determining olivine composition, despite the 10 cm−1 spectral resolution.

4.2 Ground Truth With the MSL Curiosity Payload

[14] The MSL Curiosity rover is equipped with several instruments that are capable of identifying the mineralogy on Mars either by direct structural or indirect elemental analyses. These instruments include ChemMin [Blake et al., 2012], Alpha Particle X-ray Spectrometer [Campbell et al., 2012], and ChemCam [Wiens et al., 2012]. As such, this payload also can be used to identify the mineralogy of the basaltic dunes as they are crossed on the way to the central mound in Gale Crater. More specifically, the Fo# of the olivine in the dunes can be determined. When the olivine composition is determined for the NE-SW trending dune field, the results can be used either to support or refute the olivine compositional prediction made here using our developed olivine spectral index strategy in association with the TES data.

5 Conclusions

[15] A general olivine spectral index equation was developed from a suite of synthetic Mg-Fe olivine spectra [Lane et al., 2011] to identify olivine composition from midinfrared spectral data. The general equation was modified for each of the 13 unique Mg-Fe olivine compositions and applied to orbital TES data of Gale Crater to identify the olivine composition therein. The highest olivine index values overlie and closely map the dark basaltic dunes that surround the central mound in Gale Crater, highlighting the presence of olivine within the basaltic dunes. The olivine index maps indicate the olivine composition to be ~Fo55/(50). Spectral unmixing analyses of TES data show the NE-SW trending dune field to be dominated (~51%) by a Surface Type 1 basalt component with an additional olivine enrichment (22%) also indicated to be a Fo55 composition. The enrichment in olivine may be due to aeolian grain sorting. This olivine-composition prediction can be tested by the instrument suite on Curiosity when it reaches the dune field as the rover heads to the 5 km high central mound later in the mission.

[16] The presence of the MSL rover in Gale Crater presents a rare opportunity to ground truth these orbitally derived mineral compositions. If the instrument package on Curiosity also identifies ~Fo55 olivine in the NE-SW trending dune field, then the olivine spectral index developed here for determining the compositions of Mg-Fe olivines may be more robustly applied to the orbitally acquired global TES data set.


[17] Thanks are extended to the JMARS software developers and to the reviewers of the manuscript for providing valuable feedback. This research was funded through NASA's Mars Odyssey Participating Scientist Program. This work is PSI contribution 604.

[18] The Editor thanks John Mustard and Kimberly Seelos for their assistance in evaluating this paper.