Slope-streak formation is one of the most active surface phenomena presently observed in equatorial Mars. Here, we use for the first time color HiRISE and hyperspectral CRISM images to characterize the spectral properties of individual slope streaks formed during the past decade in the Olympus Mons Aureole, constrain their composition, and discriminate between ‘spectrally permissible’ formation hypotheses. The slope streaks investigated form a spectrally distinct class of active surface features with a compositional uniqueness that is inconsistent with previously suggested slope-streak formation mechanisms, such as exposure of pre-existing substrate, persistent soil moisture, or textural effects alone. Instead, a dry transparent surface coating and/or effective enrichment in low-albedo ferric oxides emerge as the most likely surface coloring mechanisms. This raises the intriguing possibility that the spectrally unique class of slope streaks identified here may represent low-volume seeps of brines that evaporate shortly after their emergence, leaving behind dry and diagenetically ‘stained’ surfaces.
 Formation of low-albedo slope streaks within seasonal time scales as short as several months represents one of the most active surface phenomena presently observed on Mars [Sullivan et al., 2001; Malin and Edgett, 2001; Schorghofer et al., 2007]. Slope streak occurrence is exclusive to equatorial regions with low thermal inertia (TI) values [Sullivan et al., 2001; Schorghofer et al., 2002] commonly attributed to a fine-grained surface layer of unconsolidated material (i.e., dust) [Putzig et al., 2005]. New slope streaks typically appear as <10% darkening [Sullivan et al., 2001] of otherwise intact surrounding slopes and characteristically exhibit point-source elongated flow-like morphologies rarely exceeding 3 km in length and 0.3 km in width (Figure 1). The origin of Martian slope streaks remains controversial. ‘Dry’ ‘dust avalanching’ that reveals a pre-existing low-albedo substrate [Sullivan et al., 2001; Baratoux et al., 2006; Chuang et al., 2007; Phillips et al., 2007; Chuang et al., 2010] is the more ‘conventional’ mechanism suggested, while ‘wet’ surface modification from liquid seeps [Ferguson and Lucchitta, 1984; Ferris et al., 2002; Miyamoto et al., 2004; Kreslavsky and Head, 2009] is regarded as the ‘less conventional’ suite of processes suggested. Previous efforts to discriminate between these mechanisms through photo-interpretation of high-resolution images [Sullivan et al., 2001; Malin and Edgett, 2001; Chuang et al., 2007; Phillips et al., 2007; Chuang et al., 2010], characterization of their occurrence environments [Schorghofer et al., 2002], theoretical modeling [Miyamoto et al., 2004; Kreslavsky and Head, 2009] and comparison to Earth analogs [Head et al., 2007] resulted in physical constraints, but did not provide direct evidence for unambiguously resolving the ongoing debate. For example, of the thousands of slope streaks imaged so far with resolution as high as 0.3 m/pixel by HiRISE (High Resolution Imaging Science Experiment [McEwen et al., 2007]), only several cases of low-albedo slope streaks display evidence [Chuang et al., 2007; Phillips et al., 2007; Chuang et al., 2010] for the required topographic relief in the ‘dust avalanche’ model between undisturbed surroundings and streak interior, where dust removal and substrate exposure are invoked. Thus it appears that in the vast majority of the cases such relief, if exists at all, is at sub-pixel cm-mm scales unresolved by HiRISE. In this study we present a new approach for discriminating between slope-streak formation mechanisms by using for the first time spectral orbital images to constrain slope-streak compositions and possible sub-pixel effects. We focus on two locations in the Olympus Mons Aureole (OMA) where overlapping color HiRISE and hyperspectral CRISM (Compact Reconnaissance Imaging Spectrometer for Mars [Murchie et al., 2007]) images provide unprecedented high-resolution spectral data to study fresh slope streaks formed during the last decade, and which are thus assumed to have experienced minimal post-formation modifications.
 In both of the OMA locations selected for this study (Figure 1) false-color HiRISE images acquired with three spectral channels at 0.54, 0.69 and 0.87 μm provide unambiguous evidence against substrate exposure and the ‘dust-avalanche’ mechanism. At the first location, within a ∼30-km-diameter crater in Lycus Sulci (Figure 1a), meter-sized depressions in the surface dust layer that formed as bounce marks from a rolling boulder appear differently colored adjacent to and within a slope steak formed between 6/2008–8/2009. Invoking substrate exposure for this slope streak would require exposure of that dark-toned substrate by the bounce marks outside the streak as well. At the second location, along the southern basal scarp of Olympus Mons (Figure 1b), slope steaks formed between 12/2005–4/2007 display an invariable grey color as they traverse distinct blue-appearing continuous outcrops. Invoking substrate exposure in this case would require the streak color to vary accordingly at the point where the streaks cross the blue-appearing layers. With substrate exposure ruled out for the OMA slope streaks in Figure 1, we examine surface-structure effects, such as sub-pixel unresolved roughening of the dust, which has been previously proposed as a possible surface darkening mechanism for low-albedo slope streaks [e.g. Kreslavsky and Head, 2009]. In fact, sub-pixel roughness measurements derived from HRSC stereo images [Mushkin and Gillespie, 2006] indicate several % of increased unresolved shadowing resulting from increased ∼mm-cm scale roughness within slope streaks relative to their background surfaces (Figure S2 of the auxiliary material). However, HiRISE I/F reflectance values in the Lycus Sulci study site (Figure 1a) demonstrate that a fully shadowed undisturbed dusty slope remains ∼25% lighter than a fully shadowed streak surface. This implies that compositional darkening of the surface, in addition to possible shadowing, is also required for this streak. Further testing of such sub-pixel effects is also facilitated by the separable color characteristics of slope streaks in HiRISE color images (Figures 1a and 1b) and application of spectral mixture analysis (SMA) [Adams et al., 1986], a common method used to study sub-pixel compositional heterogeneity on the Martian surface [e.g., McCord et al., 2007]. In SMA the measured spectrum from each pixel is described as a linear mixture of spectrally distinct scene elements (spectral endmembers) that comprise it. Thus, slope streaks resulting from substrate exposure or sub-pixel roughening of the dust layer should fall within the spectral mixing volume defined by exposed substrate, dust and shadow. However, the slope streaks at the OMA study sites plot unambiguously outside these mixtures (Figures 1c and 1d and Figures S1a and S1b) as unique spectral endmembers that cannot be attributed solely to mixtures of the exposed surrounding scene elements. The spectral ‘darkening-by-roughness’ trend in Figure 1d also indicates that the several % of increased unresolved shadowing found within slope streaks (Figure S2) has only a minor secondary role in streak coloring, which is therefore requires 1st order compositional effects.
 CRISM 18 m/pixel images, in which individual slope streaks can be resolved, and the >300 usable CRISM spectral channels between 0.4–2.5 μm facilitate unprecedented characterization of slope-streak composition relative to surrounding slopes. In such CRISM images the OMA slope streaks in Figure 1 and their background dusty surfaces appear to be dominated by a strong absorption edge between 0.40–0.75 μm, a wide trough between 0.85–1.1 μm (interrupted by CRISM's detector change at ∼1.0 μm) and a reflectance maximum between ∼1.3–1.4 μm (Figure 2), all indicating spectral dominance of ferric oxides (FeOx) as previously determined for Martian dust [Bibring et al., 2007]. Spectral ratios commonly used to reduce residual atmospheric and calibration artifacts and to enhance subtle in-scene spectral and compositional variations reveal that, relative to their undisturbed dusty environment, these slope-streak spectra: 1) decrease steadily from ∼0.4–0.7 μm, 2) remain fairly ‘flat’ from ∼0.7 μm onwards with a possible decrease towards longer wavelengths, and 3) lack detectable spectral absorption features. In contrast, CRISM streak/background spectral ratios for other similar-scale active low-albedo surface features on Mars, such as Gusev crater wind streaks (Figure 2c) reveal olivine + pyroxene absorption features at ∼1.0 and 1.95 μm consistent with formation of the Gusev wind streaks by eolian exposure of a darker-toned basaltic-composition substrate as verified by the rover ‘Spirit’ [Lichtenberg et al., 2007]. Thus, the spectrally featureless darkening of the OMA slope streaks revealed by CRISM distinguishes them as a unique class of active low-albedo surface features. The absence of diagnostic water or ice absorption bands at 1.4–1.5 and 1.9–2.0 μm in the CRISM slope streak/background OMA spectral ratios also indicates no detectable H2O enrichment at the time of measurement and unambiguously rules out darkening by persistent soil moisture, as would be expected under current surface conditions on Mars.
 In both OMA locations described above, the leading hypotheses for slope-streak formation, substrate exposure, textural modifications following wet seeps and/or soil moisture, are all inconsistent with the available spectral observations. Thus, within the realm of previously established and documented physical processes, we propose two spectrally feasible darkening mechanisms for the OMA slope streaks: H1) A transparent coating, such as silica, with a refractive index intermediate between that of the dust grains and the atmosphere, which could result in spectrally featureless darkening, or H2) An increase in the abundance of a low-albedo material with the same spectral absorption features as the dusty background. In H2, low-albedo FeOx, which dominate the spectral signature of Martian dust [Bibring et al., 2007] and are a group of ubiquitous surface minerals on present-day Mars, appears as a spectrally viable option. Thermal infrared multispectral THEMIS images of resolved slope streaks do not reveal the diagnostic 9.1 μm emissivity silica band, and yet we cannot rule out this option due to the fine particulate nature of Martian surface dust. Nonetheless, three additional independent observations favor H2: 1) Surface enrichment in cm-scale FeOx (mainly hematite) has been previously implicated in similar spectrally featureless darkening of the Opportunity landing site area in Meridiani Planum, relative to nearby bright plains [Arvidson et al., 2006] (Figure S3a); 2) Enrichment in FeOx is consistent with the decrease in the streak/background ratio spectrum <0.7 μm, because darkening by low-albedo FeOx will be weaker closer to the Fe-O charge-transfer band centered at the UV wavelengths, which is already near saturation in Martian dust; and 3) Similarities between CRISM streak/background spectral ratios and laboratory coarse/fine hematite spectral ratios (Figure S3b) suggest that apparent FeOx enrichment within slope streaks could result from increasing the effective grain size of pre-existing FeOx minerals. For example amalgamation of nanophase FeOx minerals in the dust layer into larger FeOx grain aggregates could effectively increase FeOx volume scattering as well as the mm-cm scale roughness within slope-streak surfaces (Figure S2).
 The occurrence of transparent grain coatings, FeOx precipitation and/or FeOx amalgamation during time periods as short as several months and within point-source flow-like morphologies, appears to be strongly suggestive of a diagenetic process involving a liquid phase. Diagenetic precipitation of FeOx from low-temperature, near-surface brines has been previously suggested for early Mars [Squyres et al., 2004] and in the geologic record on Earth [Chan et al., 2004], where it is also observed as a presently active process in wet acidic environments [Benison et al., 2007]. However, the lack of detectable H2O absorption bands in the CRISM slope streak/background spectral ratios (Figure 2) implies that such liquids, if they occur on Mars, are transient and evaporate shortly after surfacing. We therefore propose a formation model for the OMA slope streaks in which the low-TI surface dust layer provides a thin surficial micro-environment where peak summer daytime temperatures are sufficiently warm to allow transient down-slope seepage of volatile liquids having salinity-depressed freezing temperatures [Chevrier and Atheide, 2008; Kreslavsky and Head, 2009; Fairen et al., 2009], atop a less permeable interface (Figure 3). Propagation of the brines may occur within the pore spaces of the dust layer stopping when the warm condition cease and the brines evaporate or freeze and sublime shortly thereafter. The dry surfaces left behind are stained by a silica coating, amalgamated clusters of nanophase FeOx, and/or newly precipitated low-albedo FeOx. The lack of relief between the OMA slope streaks and their surroundings, even at 0.3 m/pixel HiRISE resolution (Figure 1), indicates insignificant erosion during seepage and thus a low flux of brines. At this stage, the source and nature of the implied brines remains as an intriguing open aspect of the proposed model, although the characteristic point-source morphology of these slope streaks suggests localized tapping of a confined source.
 Previous studies have suggested that a gamut of formation and post-formation processes may characterize the ∼800,000 [Aharonson et al., 2003] slope streaks estimated to occur on Mars [Ferris et al., 2002; Phillips et al., 2007]. Accordingly, we propose that the orbital spectral data used here are sufficient to distinguish a unique class of recently formed slope streaks identified at two locations in OMA. This class of low-albedo slope streaks: 1) cannot be explained by exposure of pre-existing substrate, 2) are spectrally and compositionally distinct from their surroundings, and 3) display spectrally featureless darkening relative to their surrounding dust-mantled slopes. From a spectral perspective, a thin dry transparent coating and/or effective enrichment in FeOx emerges as the most likely surface-darkening mechanisms for this class of slope streaks. The short time-scales and flow-like morphology associated with this class of OMA slope streaks appear to support a diagenetic origin, thus raising the provocative possibility that low-volume seeps of short-term transient liquids may have recently occurred in the OMA locations discussed above and conceivably across other equatorial regions of Mars, where slope streaks are presently observed forming. We suggest the spectral approach described in this paper as a viable strategy to further study and subcategorize the large and apparently diverse population of slope streaks on Mars, which includes relief-displaying slope streaks and light-toned slope streaks, in order to gain a better understanding of the possible origin and environmental significance of these active surface features.