The igneous evolution of Mars is well represented in stratigraphic settings that transition across major time stratigraphic boundaries. Here we analyze in detail the morphology and composition, determined through visible–near-infrared spectroscopy, of igneous volcanic terrains in Ares Vallis, Mars. Upland plateau units with crater-filling volcanic plains are of Noachian age; smooth units to the west and north of the mouth of Ares Vallis have been mapped as Hesperian in age. The age and origin of the units that comprise the floor of Ares Vallis, connecting the upland plateau units and the smooth units to the west and north of the mouth of Ares Vallis, is less certain due to the small area available for crater counting. The mafic mineral compositions of these units are well resolved with OMEGA (Observatoire pour la Mineralogie, l'Eau, les Glaces et l'Activité) and CRISM (Compact Reconnaissance Imaging Spectrometer for Mars) data. The data show that the Noachian volcanic units are distinctly olivine-rich, while the plains units to the west and the floor of Ares Vallis volcanics show diagnostic absorptions of olivine-pyroxene, typical of Hesperian volcanics elsewhere on Mars. Based on these findings, we propose there has been a change in the temperature and/or degree of partial melting in the mantle over time for this region of Mars.
 Understanding the history of surface volcanism on Mars is important in piecing together the igneous evolution of the planet and the role that volcanism may have had in the evolution of the atmosphere and subsurface [Craddock and Greeley, 2009; Carr and Head, 2010]. Volcanism is well recognized as the dominant form of resurfacing in the Hesperian era where the northern lowlands were resurfaced [Head et al., 2002], extensive ridged plains units of the southern highlands were formed (e.g., Syrtis Major) [Hiesinger and Head, 2004], and the ridged plains surrounding and forming the upper sections of Valles Marineris were emplaced [e.g., McEwen et al., 1999]. Hesperian volcanic units have been extensively analyzed with remotely sensed data [e.g., Mustard et al., 1997; Christensen et al., 2001; Rogers and Christensen, 2007; Rogers et al., 2007; Hahn et al., 2007; Baratoux et al., 2013], which shows these rocks are dominantly basaltic in composition. Mineralogy determined from visible near-infrared (VNIR) and thermal infrared (TIR) data shows an olivine basalt composition dominated by high calcium pyroxene (HCP) and plagioclase [Poulet et al., 2009b; Rogers and Christensen, 2007; Rogers et al., 2007]. In contrast, although Noachian volcanism is widely understood to be an important process [e.g., Phillips et al., 2001; Jakosky and Phillips, 2011; Spohn et al., 2001; Werner, 2009], clearly defined Noachian-aged volcanic plains that can be characterized compositionally with remotely sensed data are comparatively rare.
 Nearly half of the Martian surface is thought to be of definitive volcanic origin [e.g., Greeley and Spudis, 1978; Mouginis-Mark et al., 1992; Greeley et al., 2000; Head et al., 2001]. In general, the Noachian era saw a peak in volcanism, which gradually decreased over time [Head et al., 2001] and is consistent with the idea that Tharsis was mainly in place by the end of the Noachian [Phillips et al., 2001; Jakosky and Phillips, 2011; Spohn et al., 2001]. Volcanism in the Noachian era was likely concentrated in the Tharsis region [e.g., Phillips et al., 2001; Carr and Head, 2010], where some of the volcanics are exposed in the layered walls of Valles Marineris [e.g., McEwen et al., 1999; Hamilton et al., 2003; Christensen et al., 2003, 2005; Edwards et al, 2008; Koeppen and Hamilton, 2008; Flahaut et al., 2012]. It has been demonstrated with remotely sensed data [Bandfield et al., 2000; Bandfield, 2002; Mustard et al., 2005; Rogers and Christensen, 2007; Rogers et al., 2007; Koeppen and Hamilton, 2008; Poulet et al., 2009a, 2009b; Rogers and Fergason, 2011] that the mineralogy of the oldest Martian surfaces and studies of Martian meteorites [McSween, 1994; Nyquist et al., 2001] are suggestive of a crust that is enriched in mafic minerals. Results from the thermal emission spectrometer (TES) data [Bandfield et al., 2000; Bandfield, 2002] showed that the surface of Mars can be broadly characterized into two surface types of either basalt or basaltic andesite to andesite, where the mineralogy consists mostly of high calcium pyroxene (HCP) and plagioclase feldspar with no low calcium pyroxene (LCP) above the detection limits (5%–10%). Additionally, olivine has been detected globally with various instruments including TES [e.g., Rogers et al., 2005; Hoefen et al., 2003; Bandfield, 2002; Hamilton and Christensen, 2005; Koeppen and Hamilton, 2008], THEMIS [e.g., Rogers et al., 2005; Hamilton and Christensen, 2005], the Observatoire pour la Mineralogie, l'Eau, les Glaces et l'Activité (OMEGA) [e.g., Mustard et al., 2005; Bibring et al., 2005; Poulet et al., 2007], and the MER rovers in situ [e.g., Christensen et al., 2004; McSween et al., 2004; Morris et al., 2004]. From the VNIR analyses of later missions, although high-calcium pyroxene is present in all age terrains, olivine and low-calcium pyroxene are the most commonly detected minerals in older terrains and low calcium pyroxene content decreases with surface age [e.g., Mustard et al., 2005; Bibring et al., 2005, 2006; Poulet et al., 2007, 2009b; Flahaut et al., 2012]. Skok et al.  agree that, in Noachian terrains, olivine and low-calcium pyroxene tend to occur in distinct, highly segregated concentrations of either pyroxene or olivine, which is different than the nature of Hesperian material that shows more of a lithology of olivine and pyroxene; however, studies using TES and THEMIS data support crustal heterogeneity [e.g., Rogers et al., 2005; Rogers and Fergason, 2011].
 The continuation of the volcanic record that occurs near Valles Marineris into the Hesperian era is also preserved at the top-most section of the walls [e.g., Greeley and Guest, 1987; McEwen et al., 1999; Flahaut et al., 2012]. This type of Hesperian volcanism, also known as the ridged plains [Greeley and Spudis, 1981; Tanaka, 1986], has been estimated to have resurfaced about 30% of the Martian surface [Head et al., 2002; Carr and Head, 2010]. The majority of the surface expression of Hesperian volcanics occurs in the northern lowlands [Greeley and Spudis, 1978; Tanaka et al., 1988; Mouginis-Mark et al., 1992; Greeley et al., 2000] is no more than 1 km thick and is underlain by a cratered, Noachian-aged surface [Frey et al., 2001; Head et al., 2001] that exhibits the typical ancient-altered crustal mineralogical signatures [Carter et al., 2010]. The northern lowland volcanics have been shown to be continuous with ridged plains farther south [Head et al., 2002; Carr and Head, 2010]. Additionally, it has been noted that the Hesperian ridged plains within the heavily cratered terrain are likely volcanic [Head et al., 2006; Carr and Head, 2010]. Hesperian volcanics, in general, have orbital signatures consistent with a mixture of olivine, low-calcium pyroxene, and high-calcium pyroxene with the latter being the more abundant of the pyroxenes [Mustard et al., 2005; Bibring et al., 2005, 2006; Poulet et al., 2007, 2009b; Salvatore et al., 2010; Skok et al., 2012].
 While determining the precise timing of volcanism in ancient terrains is often troublesome for areas that are not near volcanic centers, the composition of Noachian-aged volcanic terrains warrants further study. The cratered Noachian terrain just to the east of Aram Chaos [Tanaka et al., 2005] is of interest for three main reasons: (1) a significant amount of volcanism has been proposed to have taken place on Mars in the Noachian [cf. Greeley and Spudis, 1981], some of which is preserved throughout the Noachian-aged plains studied herein; (2) change in igneous composition can be assessed since well-exposed Noachian-aged volcanics are in close proximity to igneous terrain that is mapped as Hesperian; and (3) exposure in this region is good as compared to other dusty Noachian terrains. The study area, in particular, exhibits resurfacing in the form of filled or ghost craters bearing some of the most olivine-rich compositional signatures detected by the TES [e.g., Koeppen and Hamilton, 2008]. Further characterization of these resurfaced plains units may give us more insight into the manner and location of Noachian-aged volcanism. Moreover, units beyond the mouth of Ares Vallis and into the northern plains are clearly Hesperian in age; thus, the stratigraphy from the Noachian highlands to the Hesperian lowlands stretches across a transitional temporal boundary, providing insight into the igneous evolution of the region.
 In this study, we characterize the composition and morphologies of units in the study area, present strong evidence for a compositional change from more to less olivine-rich in the Ares Vallis region on Mars, and hypothesize why this change occurred. Areas with igneous signatures in and around Ares Vallis can be detected with the high-resolution Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) [Murchie et al., 2007] and the OMEGA instrument on Mars Express [Bibring et al., 2004]. OMEGA is especially key in allowing the broad characterization of the widespread Noachian plains units that stretch between Ares Vallis and Meridiani Planum. Their morphological and geological context are assessed with the Context Camera (CTX) [Malin et al., 2007], the High-Resolution Imaging Science Experiment (HiRISE) [McEwen et al., 2007], as well as other imaging data, allowing for a more detailed geomorphic characterization of the deposits than has been performed previously. In addition, Mars Orbiter Laser Altimeter (MOLA) topography data [Smith et al., 2001] aids in the characterization of stratigraphy of units and excavation depth calculations.
 Ares Vallis is an approximately 1,500 km long, 2 km deep channel that originates in Iani Chaos near 17.5°W longitude, 5°S latitude (Figure 1). The channel is up to 300 km wide, near the mouth, and 21 km wide at its narrowest. The entire surface of the region, including the shape of Ares Vallis, has been structurally controlled by the Chryse and Aram Chaos Basins (see section 5.1. and Figure 1a) [Schultz et al., 1982]. The channel is surrounded by older highland plains to the east that appear to be resurfaced. Only one detailed morphological and mineralogical study has been performed in this area to date and is discussed further in section 2.3 [Rogers et al., 2005].
2.2 Ages of Units
 Crater counting of valleys and channels is particularly difficult because the area for counting is small and water and wind can modify a cratered channel floor, causing it to appear pitted (pits can easily be confused as small craters). The eroded craters appear to be smaller than their noneroded counterparts and, therefore, return a younger age than that of the actual formation. In addition, any subsequent deposits lain in a channel floor can also cover up craters, having the same effect. Tanaka et al.  dated these surfaces and determined the age of the main channel of Ares Vallis to be middle of the Late Hesperian (Figure 1b), whereas the date of last fluvial activity for the large tributary to the west, which flowed north eastward around Aram Chaos, has been dated by Warner et al.  to be Amazonian in age. Some even propose [e.g., Marchenko et al., 1998] that Ares Vallis is Early Hesperian in age, whereas the Tiu Valles system that crosscuts Ares Vallis at its mouth is Late Hesperian to Early Amazonian in age. Regardless, it is evident that Ares Vallis has experienced the last major episodes of fluvial and/or volcanic resurfacing between the Hesperian (main channel) and Early Amazonian (tributary).
Tanaka et al.  map four units of interest to this study in two key regions (Figures 1b and 2). The first region, the Noachian plains surrounding Ares Vallis, contains units Nl and Nn. Nl is early to middle Noachian in age and described as a combination of reworked, volcanic, sedimentary, and primordial crustal materials. Nn is middle to late Noachian in age and described as a combination of reworked, volcanic, and sedimentary resurfaced materials. The second region, the Hesperian units in the floor of Ares Vallis, is composed of units HCa and HCc4. HCa is late Hesperian in age and interpreted to be an indurated cap of possible igneous origin, while HCc4, also late Hesperian, is interpreted as a debris flow or compacted sedimentary material. A time-related context is clearly present for separating spectral detections on the floor of Ares Vallis from the plains to the east. This framework, on the basis of the mapping of Tanaka et al. , establishes that the plateaus to the east of Ares Vallis are Noachian in age, while the floor of Ares Vallis and the outflow channel region to the north and west are Hesperian. While this provides the relative age of the surfaces exposed in the region, it is not clear if the flat floor of Ares Vallis is a depositional or erosional unit. It may be a postfluvial, volcanic deposit, or it may be a consistent erosional unit from outflow channel formation. In addition, the map scale of Tanaka et al.  (1:15 M) is relatively coarse compared to some of the features of interest presented: spectrally unique, small-scale exposures on the channel floor may differ in age from the larger map unit that contains them.
2.3 Previous Studies
 Ares Vallis is one of many outflow channels on Mars generally thought to have formed by the catastrophic outflow of subsurface water [e.g., Carr, 1979; Baker, 1982; Baker et al., 1992; Clifford et al., 1993]. A number of other processes for channel formation including lava flow, aeolian processes, flow of water ice, flow of carbon dioxide, emplacement of debris flows, and crustal extension have also been proposed [Leverington, 2011, and references therein]. Catastrophic outflow of subsurface water is the most accepted view, yet the general observations show that alteration minerals are lacking in and around most outflow channels [e.g., Bibring et al., 2006; Mangold et al., 2007, 2008]. As mentioned previously, olivine is one of the more readily detected mineral signatures on Mars and particularly in the Ares Vallis region. The presence of diagnostic signatures of igneous compositions in Ares Vallis (and other channels), as opposed to alteration minerals may indicate a history where aqueous activity was not the sole process at work.
 Generally, a few studies have interpreted the significance of olivine on large scales, which included detections in the Ares Vallis region. Koeppen and Hamilton  mapped olivine globally using TES data and grouped the deposits by their magnesium content, expressed in Fo number, noting that the most common Fo numbers of their end-members were Fo68 and Fo53; the most common deposits were observed to occur in the floors of outflow channels and smooth-floored craters. Furthermore, Edwards et al.  imply that regional olivine-rich deposits associated with outflow channels may reflect extensive flooding by primitive lavas during the Noachian, which were later exposed by outflow channel erosional events.
 Volcanism has been thought to have caused resurfacing in the northern portion of Ares Vallis, consistent with the volcanic Hesperian-aged northern plains units [e.g., Robinson et al., 1996]. Salvatore et al.  and Ody et al. [2011, 2012] verify with CRISM and OMEGA, respectively, that olivine-rich areas exist in the Hesperian-aged northern plains, at the mouth of Ares Vallis and beyond. This igneous signature could be attributable to the Hesperian ridged plains volcanism outlined by Head et al. . However, it is not clear whether this potential volcanism is related to the olivine seen in the southern portions of the channel.
 In particular, Rogers et al.  performed a detailed study of a larger portion of the region studied here using TES and THEMIS data, revealing that there are many olivine-rich areas, including solid rock outcrops, which they attribute to one or more regional bedrock units that could be intrusive or extrusive in nature. Rogers et al.  point out three main units in the region. The unit on the floor of Ares Vallis has the lowest albedo, but intermediate thermal inertia. Mineral assemblage assessment from TIR data estimates that this unit has a mineralogical makeup of 30% feldspar, 25% pyroxene [80% high calcium pyroxene (HCP) and 20% low calcium pyroxene (LCP)], 25% high silica materials, and 20% other. The lowest thermal inertia unit corresponds to the layered channel walls and the plains units to the east, consisting of 20% feldspar, 15% pyroxene (80% HCP, 20% LCP), 35% high silica materials, and 30% other. Finally, the highest thermal inertia unit, which is a cliff-forming unit of fractured, in place rock, has a mineralogical composition of 45% pyroxene (>90% HCP), 25% olivine, and 30% other. In particular, this unit was reported to have a feldspar content below the detection limit, which is a striking characteristic as compared to the other units in the area. Rogers et al.  report that the olivine-rich unit is similar in thermophysical character to other olivine-bearing terrains on Mars, except the mineralogy reflects more pyroxene and less plagioclase than the other olivine-bearing terrains. Finally, Rogers et al.  did not include a pigeonite spectrum in their deconvolutions, which may result in an overestimation of the HCP component.
 Mars Express and the Mars Reconnaissance Orbiter (MRO) have been returning complimentary spectral data with OMEGA and CRISM and imagery that we use in the present study to consider these units in more detail, with the intent of further constraining the geological significance of the previously identified units and investigating the composition of Noachian-aged units.
3 Data Sets and Methods
3.1 Global Data Sets
 Global data and higher-resolution imagery were combined using a Geographic Information System (GIS) for the bulk of the analysis. Global data used include the Thermal Emission Imaging System (THEMIS) daytime infrared mosaic and Mars Orbiter Laser Altimeter (MOLA) [Smith et al., 2001] topographic data set, both acquired from the U.S. Geological Survey (USGS), at 256 and 128 pixels per degree, respectively. Gridded elevation data from the global MOLA map were used to create the terrain-corrected topographic profiles shown in Figure 3. Age units used in the GIS, from Tanaka et al. , were acquired from the USGS online database. Select units from the north polar map are shown in Figures 1b and 2.
 High-Resolution Science Experiment (HiRISE), Context Camera (CTX), and High-Resolution Stereo Camera (HRSC) imagery were used to assess different scales of the region's morphology. HRSC has a spatial resolution of about 10 m/pixel [Neukum et al., 2004]; the specific images analyzed herein have a resolution between about 11 and 34 m/pixel. CTX images the surface of Mars at approximately 6 m/pixel [Malin et al., 2007]; HiRISE images the surface at a much higher spatial resolution than CTX, at 25–130 cm/pixel in gray scale and color (channels are in the red, NIR, and blue–green ranges) [McEwen et al., 2007]. Along with the global data sets, CTX and HRSC images along the entirety of Ares Vallis were compiled into the GIS. HRSC, CTX, and HiRISE images used in this study are outlined in Table 1. Figure 4-6
Table 1. IDs for Imagery Used, Listed by Figure Number
 The Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) is a hyperspectral imager that collects data in two modes. The multispectral mapping data consist of 72 wavelengths which are sampled at 200 m/pixel; with full spatial and spectral resolution, CRISM can collect data at 15–19 m/pixel over 544 channels from 362–3920 nm at 6.55 nm/channel sampling [Murchie et al., 2007]. The most current publicly released calibration version of the data, TRR3 [Seelos et al., 2011], is used for images HRL00005B77 and FRT00012CED (Table 2). Atmospheric effects and photometric corrections for these data are described by Seelos et al. . The atmospheric correction was applied by dividing the scaled volcano observation 61C4, using the Pelkey 2-wavelength method for volcano scan scaling (2,011 nm/1,899 nm). The functions for photometric and atmospheric correction are included in the CAT software menu provided by the CRISM team. Spectra were extracted from regions in the CRISM and OMEGA data that were defined on the basis of spectral parameter maps [Pelkey et al., 2007] using either 5 × 5 pixel averages or regions of interest drawn manually to cover the smaller hydrated mineral units or irregular surface exposures. Residual artifacts after calibration and atmospheric removal are suppressed by ratioing the spectra to a spectrally neutral region (e.g., Figures 7 and 8). Spectrally neutral or featureless surfaces are recognized as showing no features in the summary spectral parameters. For CRISM data, ratios are generated using a neutral/featureless spectrum from the same columns because of the nature of 2-D detectors. To ensure consistency in ratios, both S and L data were ratioed prior to being joined. While the ubiquitous red slope in the visible wavelength region can lead to interpretation difficulties, that was not considered important for these data.
Table 2. IDs for Imagery Used, Listed by Figure Numbera
Center Pixel Location
Number of Pixels
aSpectral data used are referenced by their center pixel locations and the number of pixels averaged is provided. Pixel locations on each line refer to numerator and denominator spectra used in the spectral ratios, respectively of the unprojected images.
 In addition, spectral data from OMEGA, on the Mars Express spacecraft, are used in conjunction with CRISM. OMEGA is well suited for surveying the large plains area, since the coverage is more spatially extensive than that for the CRISM targeted observations. OMEGA collected data in 352 spectral channels over three separate detectors and covers the visible and near-infrared wavelength range from 0.35 to 5.09 µm; the spectral resolutions of the VNIR (0.35–1.05 µm), SWIR (0.94–2.70 µm), and LWIR (2.65–5.2 µm) detectors are 7.5, 14, and 21 nm, respectively [Bibring et al., 2004]. Basic reduction of OMEGA data from radiance to reflectance includes applying known instrumental corrections, conversion of DN (digital number) to radiance, division by the solar flux adjusted to the Mars distance to reduce the data to I/F, and lastly dividing by the cosine of the incidence angle relative to the aeroid [Bibring et al., 2005]. Atmospheric removal proceeds as with CRISM by assuming that the surface and atmospheric contributions are multiplicative and that the atmospheric contribution follows a power law variation with altitude [Bibring et al., 1989]. The model of atmospheric transmission was derived from OMEGA observations from the top and bottom of Olympus Mons. This transmission spectrum is then scaled to each OMEGA observation by the strength of the CO2 absorption and ratioed to the data to remove atmospheric contributions. The ratio technique is also used for examining OMEGA data; unlike CRISM, OMEGA is a linear array of spectrometers; thus, the featureless denominator can be from anywhere in the image (e.g., Figure 4). All spectra included in this study are listed by extraction location in Table 2.
 Reflectance spectra of the igneous minerals studied herein (cf. Figure 9), olivine and pyroxene, possess reflectance minima near 1 and 2 µm due to Fe2+ crystal field absorptions [Burns, 1970b, 1993] in the CRISM and OMEGA visible–near-infrared spectral ranges. Three overlapping absorptions near 1 µm due to crystal field transitions of Fe2+ in the distorted M1 and M2 octahedral sites [Burns, 1993, and references therein; Sunshine and Pieters, 1998] define the presence of olivine and shift to longer wavelengths with increasing Fe content [e.g., Burns, 1970a; Adams, 1975; King and Ridley, 1987; Burns, 1993; Sunshine and Pieters, 1998; Isaacson et al., 2011], although the separate absorptions are rarely discernable in orbital spectra of olivines on Mars. Absorptions near 1 and 2 µm define the presence of Fe2+ in the M1 and M2 sites in pyroxene, respectively; band centers are around 0.9 and 1.9 µm for LCP and 1.05 and 2.35 µm for HCP [Cloutis and Gaffey, 1991; Cloutis, 2002]. The band centers for pyroxenes vary systematically [Burns, 1970b, 1993] from shorter to longer wavelengths with increasing calcium and iron content [e.g., Adams, 1974; Cloutis and Gaffey, 1991].
 In Figure 4, the strength of the olivine detection for OMEGA data was measured using a modified version of the OLINDEX2 parameter from Salvatore et al. , which was originally intended to be used on CRISM data. This updated parameter takes spectral slope into account, which more accurately measures the 1 µm band depth. We replaced the wavelengths in the CRISM parameter with an average of three OMEGA wavelengths centered at the OMEGA wavelength closest to the CRISM wavelength used in Salvatore et al. . The equation for calculating OLINDEX2 for OMEGA data, with the proper OMEGA wavelengths in nanometers, is as follows:
where RC is the reflectance value at the expected center of the continuum and R is the reflectance value at the wavelength specified. The weighting of each part of the parameter is the same as from Salvatore et al. .
 Detection of hydrated minerals are made primarily on the basis of fundamental bending and stretching vibrational absorption bands, particularly near 1.4 and 1.9 µm where the 1.4 µm band is due to the structural OH stretching overtone [Clark et al., 1990; Bishop et al., 1994, 2002] and the 1.9 µm band is due to a combination tone of the bending and stretching vibrations of the H2O molecule. The positions of these bands and other diagnostic bands in the 2.1–2.5 µm range shift, depending on the cation present in the structure. For example, absorptions near 2.2 µm are typically due to Al-OH vibrations and 2.28–2.35 µm due to Fe/Mg-OH vibrations [Bishop et al., 2002, 2008].
 All library spectra of igneous minerals and rocks used in this study are shown in Figure 9. The olivine is sample KI3189, a Fo60 sample at <60 µm particle size, taken from the USGS spectral library [Clark et al., 2007]. The pyroxenes, clinopyroxene/diopside C1PP69 ~ 1 and orthopyroxene/enstatite C2PE30 ~ 2, as well as the Hawaiian basalt sample CCBA01 ~ 2 were taken from the CRISM spectral library [Murchie et al., 2007]. The mixed layer clay library spectrum used is from Milliken and Bish .
4.1 Noachian Plains Material
 The Noachian-aged plains to the east of Ares Vallis consist two separate units, as mapped by Tanaka et al. : Nn and Nl (Figures 1b and 2). The older of the two units (early to middle Noachian), Nl, is more heavily cratered and is described as reworked sedimentary and volcanic material with the possibility of containing primordial crust. The younger of the two units, Nn, is middle to late Noachian in age and is described as reworked volcanic and sedimentary material, including various types of resurfacing. This unit fills in and breaches the rims of larger craters and leaves the impression of ghost craters (Figure 2). In addition, this unit retains small craters quite well (Figure 5), indicating an indurated or relatively competent material. Some of the plains, regardless of being assigned to unit Nn or Nl, exhibit day thermal inertia values ~400–625 J m−2 K−1 s−1/2 and night thermal inertia values ~400–815 J m−2 K−1 s−1/2, as mapped using thermal emission spectrometer (TES) data by Putzig and Mellon . The day and night thermal inertia value differences on a per-pixel basis are likely a function of vertical heterogeneity on a decimeter scale [Putzig and Mellon, 2007] but generally indicate surfaces that can be combined with layers of rocky and nonrocky material. Although the modeling by Putzig and Mellon  finds vertical heterogeneity exists almost everywhere on Mars, we cannot determine the number of layers involved, layer thickness, or whether the rockier layer is exposed at the surface or beneath the surface.
 VNIR data from OMEGA reveal that, in both of the Nn and Nl units, olivine dominates the spectral signature (Figures 4 and 8). Taytay crater samples a portion of the crater-filling unit Nn. Taytay crater is an approximately 18 km diameter crater located on the southern rim of a larger, ~40 km diameter unnamed crater (Figures 2 and 8). The unnamed crater has been infilled by the younger of the two Noachian units (Nn). Taytay crater possesses an arcuate ridge on the wall of its northern side which gives rise to olivine-bearing signatures in both CRISM and OMEGA data (Figures 4 and 8) [Wilson and Mustard, 2010]. Because the strongest olivine signatures originate from the ejecta and the ridge on the northern interior part of Taytay, it is clear that Taytay samples the infilling material of the larger crater. A small area of hydrated material, detected in the central peak and near the southern portion of the crater, also shown in Figure 8, indicates that Taytay also excavates complex Noachian basement rocks [Wilson and Mustard, 2010]. Furthermore, we used the scheme of Grieve and Pilkington  to measure the maximum stratigraphic uplift (SU) for terrestrial cases, which has also been applied as a viable way of measuring the depth of material exposed in central peaks in Martian craters [e.g., Michalski and Niles, 2010; Caudill et al., 2011]. From the relationship,
using an average final rim-to-rim crater diameter (D) of 18.2 km, we find that the maximum stratigraphic uplift (SU) in Taytay Crater is approximately 1,700 m; this indicates that the hydrated material is excavated from around this depth, thereby constraining the thickness of the fill unit being sampled in the larger, unnamed crater to be less than ~1,700 m. Since crater degradation serves to decrease the apparent final rim to rim diameter, this estimate should be considered as a lower limit to the upper bound of the fill thickness.
4.2 Ares Vallis Floor Material
 The floor of Ares Vallis, just to the east of Aram Chaos, exhibits quite a different texture than that of the plains units. The floor of the channel in this region is relatively flat, and its top-most surface is dark, appearing to embay preexisting fluvially carved features, as shown in Figure 6. The relatively darker tone of this surface shows similarities to the widespread mafic cap recognized elsewhere on Mars [e.g., Noe Dobrea et al., 2010; Michalski and Noe Dobrea2007; Ehlmann et al., 2009] where the morphology of the mafic cap is a relatively thin obscuring layer that shows weak pyroxene absorptions in CRISM VIS/NIR data. The layer is often so thin that the underlying morphology is readily apparent, but compositional contacts are obscured. The flat floor unit exhibits a polygonal texture that is unlike the rest of the channel floor. The olivine-bearing material is observed on lighter toned material that is, in places, relatively high standing. These may be low mounds formed during the Ares Valles outflow event and later embayed or represent small-scale deformation following emplacement of the floor material. Removal of a thin mafic cap would represent small erosional windows exposing a lighter-toned material (as seen in HiRISE and CTX imagery). These lighter-toned outcrops (Figures 3 and 6) occur in association with small mounds that can be seen in HiRISE anaglyph imagery (see Figure S1 in the supporting information). The mounds are much smoother in a topographic sense than the other fluvial features exposed in the floor of the channel (e.g., compare Figure 6 and Figure S1).
 It has been noted by Rogers et al.  that, in addition to the light-toned regions on the Ares Valles floor, there are cliff-forming units standing above the valley floor that are olivine bearing in THEMIS data [Rogers et al., 2005, Figures 1 and 2]. These outcrops and units unfortunately have not been observed by CRISM and are too small to be resolved by available OMEGA data.
 The exposed polygons range in diameter from <1 m to a few tens of meters and increase in size from the edges of the channel toward its center; they also increase in size as the depth of erosion increases (Figure 6b). Polygon shape and joint types are consistent with those described by Aydin and DeGraff  as a tetragonal pattern with T and curved T joints, respectively, which are features consistent with thermal contraction. Hexagonal patterns are not observed in this unit. The fractures that define the polygons are more pronounced in the light-toned material, although the fractures are apparent in the dark-toned material that superposes the light-toned unit here, giving the appearance of a thin, semitransparent cover that overlies it close to the edge of the present-day erosional scarp. This cover is thinner in some areas than others. Superposed on the smaller polygons are larger, irregular cracks, and where erosional windows do not exist, the surface is riddled with small pits and/or craters (Figure 6a).
 Furthermore, one of the small mounds within the light-toned, polygonally fractured material appears to be cone-like. This feature exhibits a breached crater or pit on the side of its peak measuring ~42 m in diameter and has an average basal diameter of ~220 m; this yields a crater/base ratio of ~0.19 (Figure 6c), which is much smaller than other identified conical features on Mars [e.g., Meresse et al., 2008, Table 1, and references therein].
 Ridge-like features can also be identified in the floor of Ares Vallis (Figure 7c). Although these ridges continuously crosscut the entire channel, the topographic relief (on the order of up to 10 m in the MOLA gridded topography) of the ridges is consistent with the floor material as being extremely competent, enough so to preserve small-scale topography coinciding with running water. Additionally, the thermal inertia values for day (~550–900 J m−2 K−1 s−1/2) and night (~500–1000 J m−2 K−1 s−1/2) [Putzig and Mellon, 2007] indicate that rocky material is dominating the thermal characteristics in the floor of Ares Vallis where the aforementioned observations were made.
 For well-exposed light-toned outcrops on the floor, CRISM spectral data (where available) bear signatures of olivine and pyroxene with absorptions near 1 and 2 µm (Figure 3). The area where the unit is not eroded seems to be mafic in composition, but the spectral signatures are not nearly as strong as those present from the freshly exposed surfaces. This is similar to the character of the mafic cap elsewhere on Mars [Ehlmann et al., 2009; Michalski and Noe Dobrea, 2007; Noe Dobrea et al., 2010]. The band center of the broad 2 µm pyroxene absorption is approximately 2.13 µm. Because we can expect exsolved pyroxenes to act as intimate mixtures of two pyroxenes and because band center does not vary with absolute modal abundances [Sunshine and Pieters, 1993], we report the pyroxene composition relative to the amount of HCP versus LCP. Accordingly, by inspecting the 2 µm band center, we find that this material composing the light-toned units is indicative of a pyroxene that is composed of an approximately 50:50 mixture of HCP and LCP.
5.1 Origin of the Ares Vallis Floor Material
 There are a number of possibilities that could lead to the presence of an olivine-pyroxene basaltic signature on the floor of Ares Vallis including prechannel emplacement or postchannel emplacement of the igneous material. Other ideas, such as formation of outflow channels by lava, have been proposed [e.g., Leverington, 2011]; however, specific examples for Ares Vallis have not been documented extensively enough to support this conclusion over a water-carving scenario, and this hypothesis is not considered here.
5.1.1 Prechannel Formation of Ares Vallis Floor Material?
 We could attribute the flat character of the Ares Vallis floor to infill by fluvially deposited sediment, aeolian material, volcanism, or a combination of these factors. The simplest way to explain the igneous signature on the floor of Ares Vallis is by assuming that the current depth of the channel represents the depth of excavation during channel formation. A detailed study compared Ares Vallis to the Lena River in Siberia and showed that the width to depth ratio and discharge rate of Ares Vallis is consistent with those of rivers in periglacial environments [Costard et al., 2007]. However, this does not address the detail of the channel floor morphology between streamlined islands.
 If the surface of the floor of Ares Vallis is, indeed, a solely erosional surface, a plausible explanation for heterogeneity in this area can be accounted for through studies of Martian impact basins by Schultz and Glicken , Schultz and Schultz , and Schultz et al. . The authors identify many basins on Mars that have been covered, only to be reexpressed on the surface by endogenic processes and features. In particular, Schultz et al.  detail that the Ares Vallis region of Mars has been affected by two basin-forming impacts currently identified as Chryse Basin and Aram Chaos. According to this work, deep-seated faulting as a result of basin formation is reexpressed by chaos and outflow channels, among other features. This can be seen by examining the oddly angled bends that Ares Vallis exhibits (Figure 1a). Schultz et al.  also state that these deep-seated, preexisting listric faults can be conduits for volcanism. The fact that igneous signatures are expressed on the floor of the Ares Vallis between two crosscutting wrinkle ridges (in Figure 1, the area where the Aram and Chryse basin rings overlap), at a proper depth to find the listric faults (~2 km), in relation to the surrounding crust could be consistent with magma that ascended through the faults formed due to the Aram impact, intruded the prechannel crust, and was exposed by channel formation.
Edwards et al.  imply that olivine-rich, high thermal inertia deposits in this area may be linked to those in Valles Marineris at similar elevation by global flooding of lava early in Mars history. This scenario presents some challenges in explaining our observations. Here we have documented that the older plains to the east of Ares Vallis are more olivine-rich than that of the floor of Ares Vallis. The differing mineral assemblage between the plains materials and the floor of Ares Vallis, as well as the age of the channel floor and the fact that the unit appears to fill the channel, may point to a time-related compositional change in igneous character. One might expect a cataract of sorts to be formed if water flowed over either the contact of two different materials with differing competency or the competent material itself (cf. cataracts identified by Warner et al.  in a tributary to Ares Vallis). No such feature has been identified in the main channel in this study.
5.1.2 Postchannel Formation of Ares Vallis Floor Material
 The floor of Ares Vallis may be a volcanic deposit emplaced postchannel formation. The high thermal inertia of the area in which spectral detections are made relative to the remainder of the channel implies that there is a relatively competent/rocky material present. Furthermore, although polygons alone are not definitive evidence of a volcanic deposit, the thermal inertia and spectral signatures are strongly consistent with volcanic materials. The ridges that crosscut Ares Vallis may not be related to the volcanic unit on the floor of the channel; however, they do indicate that in order for such small relief to be preserved, the material must be very competent. Although the light-toned outcrops that show strong olivine-pyroxene signatures occur on low, local topographic rises (Figure S1), we see embayment relationships along the edges of the channel, as well as at the contact with the previously carved fluvial features. The local topographic highs may have had a thin obscuring, mafic cap removed to permit observation of the underlying mafic signatures. This embayment relationship can be clearly seen in Figures 7a and 7b. It is not clear whether all light-toned outcrops (e.g., those shown in Figure 7 as opposed to those shown in Figure 6) occur on low rises, but these can be easily explainable. The low rises have gradual slopes and topography on the order of 5–6 m, as observed in MOLA gridded elevation data; a lava flow draped over previously present topography could form undulating rises as seen here through deflation. It is possible that postemplacement structural deformation or settling of the lava flow could have formed these mounds. The postemplacement deformation hypothesis is not favored because a significant structural orientation trend does not exist for other streamlined features in the channel; in addition, other streamlined features that are much higher in elevation are embayed by the unit that appears to contain the olivine-bearing unit. Therefore, it would appear that the lava fill reached some intermediate height that covers smaller floor features, but did not reach the top of the taller streamlined features. Rogers et al.  noted that cliff-forming units at the base of some streamlined islands showed THEMIS signatures that they are olivine-bearing. Unfortunately, no CRISM or OMEGA data exist at a resolution to confirm that observation.
 Some researchers [e.g., Robinson et al., 1996] have suggested that the mouth of Ares Vallis has been filled with Hesperian volcanic material. It is possible that a volcanic flow originating from the Chryse region could have filled a large portion of Ares Vallis; however, according to current topography, the area where spectral detections are made is coincident with a surface that would have required the lava to flow uphill from the Ares Vallis mouth (cf. Figure 10). Also, VNIR CRISM data from the Acidalia/Chryse region [Salvatore, 2010] show that the compositions of the pyroxenes there are more HCP-rich than those identified on the floor of Ares Vallis. Therefore, this hypothesis is not preferred. Since source vents for Hesperian volcanism are rarely identifiable, it is likely that the source for this material may be obscured.
5.2 Importance of the Noachian-Aged Plains Material
 If it is possible to document Noachian lava flows or igneous outcrops that all have the same general spectral character, we may be able to constrain a more global process for extensive production of olivine-rich igneous materials. Earth's surface rarely exhibits true signatures of melts directly from the mantle due to crustal assimilation and extensive fractional crystallization, but work by McSween et al.  shows that Mars' surface may be dominated by primary magmas and thermal modeling by Baratoux et al.  implies that we can assume that Hesperian-aged volcanism on Mars should be sourced directly from the mantle. Following that logic, if Hesperian-aged terrains experienced volcanism that was sourced directly from the mantle, it is likely that Noachian terrains did also. Because the olivine phase field expands as pressure decreases, we can generally assume that all melts produced by mantle melting are olivine saturated and will crystallize olivine first, regardless of mantle temperature. Dunite channels in peridotite on Earth [e.g., Kelemen et al., 1995] give clear evidence for this process. For a mantle that is hotter in comparison to one that is colder, melts would have to fractionally crystallize more olivine to get to pyroxene and plagioclase saturation; in addition, it is plausible that the lithosphere may be thinner for a hotter mantle as opposed to a cooler mantle. If it can be demonstrated that very olivine-rich volcanic material is increasingly present in Noachian igneous terrains, a scenario where either a hotter mantle and/or a thinner lithosphere in the Noachian as compared to the Hesperian is more likely to be the process that formed the trend in volcanism from more LCP enriched to more HCP enriched, as noted by Mustard et al. . If this is indeed the case, then confirming the lack of plagioclase in the unit studied herein, as noted by [Rogers et al., 2005], would be a compliment to the pyroxene distribution seen through time by Mustard et al . In addition, being able to discern the composition of the olivines we detect can be a third confirmation that the Noachian volcanism sourced from a hotter mantle (i.e., melts from a hotter mantle should be more Fo-rich) than that of Hesperian volcanism.
 In this study, we analyze the morphology, stratigraphy, and VNIR spectral properties of volcanic deposits from the Noachian and Hesperian time periods in the Ares Vallis region. We interpret these data to show that at least some of the most recent volcanic deposits on the floor of Ares Vallis were emplaced in the Hesperian. We interpret the diagnostic VNIR crystal field absorptions, observed in the reflectance spectra and exposed in light-toned mounds that may be erosional windows from beneath a thin cover, as olivine-rich basalts, containing a compliment of pyroxenes with an approximately 50:50 mixture of LCP and HCP. This composition is more typical of other Hesperian basalts on Mars which contain LCP, HCP, and olivine, as opposed to Noachian igneous terrains which contain much less HCP. Alternatively, these light-toned outcrops could be extensions of the olivine-bearing, cliff-forming units described by Rogers et al. , in which case they are units that predate the formation of Ares Valles. We feel the available data favor the former interpretation, but additional high-resolution compositional data would be required to completely resolve.
 The Noachian-aged plains on the plateaus to the east of Ares Vallis have also likely been volcanically resurfaced and exhibit distinct olivine-rich compositions in deposits that fill craters and low-lying areas—a signature which is fundamentally different than that of the floor of Ares Vallis. We interpret the mineralogy to reflect an olivine basalt composition which is not typical of Hesperian lavas in that it shows no definitive spectral evidence for the presence of pyroxene in the VNIR.
 If the units exposed on the floor of Ares Vallis are Hesperian-aged volcanics, similar to the large volcanic units to the west and north of the mouth of Ares Vallis, these analyses capture an evolution in the mineral assemblage from olivine-rich Noachian volcanics to HCP-plagioclase dominated Hesperian volcanics. This difference conforms to the trend observed in VNIR data with LCP as the dominant mafic mineral in older terrains and HCP is a more prominent component in younger terrains. Better constraints on the ages of the units in this region and additional instances of compositional-temporal differences of volcanic terrains across the Noachian-Hesperian boundary will be important to test the hypotheses of a significant change in Martian volcanism at this time. Could the pyroxene composition of the basalt in the floor of Ares Vallis represent a “missing link” to the LCP to HCP transition over time? These two observational conclusions, combined with similar observations across the globe, imply that some igneous compositional change has occurred between the Noachian and Hesperian, but this change is only now being assessed. The change in pyroxene type (LCP to HCP) is corroborated through modeling by Baratoux et al. . More observations and analysis are needed to evaluate these proposed trends and especially to make inferences for mantle properties.
 We are very grateful for the dedicated work of the NASA MRO project team and the excellent job done by the CRISM Science Operations Center (SOC). This work was supported by NASA through NASA MDAP contract NNX07AV41G to JFM and subcontract JHAP852950 to the Johns Hopkins University Applied Physics Laboratory.