ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) thermal infrared imagery over a 389 km × 387 km area in northern Nevada (38.5°–42°N, 114°–118.5°W) was analyzed to evaluate its capability for accurate and cost-effective identification and mapping of quartz and carbonate minerals at regional to local scales. The geology of this area has been mapped at a wide range of scales and includes a diversity of rock types and unconsolidated surficial materials, many of which are composed primarily of quartz and carbonate minerals. This area is also endowed with a wide variety of economically and scientifically important ore deposit types that contain an array of commodities (Au, Ag, Pb, Zn, Cu, Mo, W, Sn, Be, F, Mn, Fe, Sb, Hg, and barite). The hydrothermal systems that generated these deposits frequently deposited large amounts of quartz where fluids cool, and generally smaller amounts of calcite or dolomite by other mechanisms.

To identify and map quartz and carbonate minerals, band ratioing techniques were developed based on the shapes of laboratory reference spectra and applied to ASTER Level 2 surface emissivity products of 108 overlapping scenes. These mineral maps were mosaicked into a single coverage that was overlain with published, vector-format geologic maps of various scales to determine which geologic terranes, formations, and geomorphic features correspond to identified quartz or carbonate. Where quartz or carbonate minerals were mapped in rocks composed primarily of other minerals, they were inferred to be hydrothermal in origin and compared to known occurrences of hydrothermal alteration and mineralization.

The ASTER-based quartz mapping identified thick sequences of quartzite, bedded radiolarian chert, quartz sandstone, conglomerates with clasts of quartzite and chert, silicic and/or altered rhyolites, and silicic welded tuffs. Alluvial fan surfaces, sand dunes, and beach deposits composed of quartz and/or carbonate are prominent, well-mapped features. Quartz was also identified in smaller bodies of jasperoid, quartz-alunite, and quartz-sericite-pyrite alteration, hot spring silica sinter terraces, and several diatomite and perlite mines and prospects. The ASTER-based carbonate mapping identified thick sequences of dolomite, limestone, and marble, as well as small hot spring travertine deposits. Eolian carbonate was identified in several playas. Dolomite exhibited a stronger carbonate response than calcite, as predicted based on their thermal spectral characteristics. Quartz was detected at lower concentrations than carbonates because of the greater strength of the quartz reststrahlen features in the thermal infrared compared to the bending-related spectral features of carbonates. The 90 m ground pixel size of the ASTER thermal imagery prevents the identification of small-scale features. Despite this limitation, numerous bodies of hydrothermal quartz were detected in or near known Carlin-type gold deposits, distal disseminated Au-Ag deposits, high- and low-sulfidation epithermal Au-Ag deposits, and geothermal areas. Detection of hydrothermal carbonate was rare and mainly in geothermal areas.

The ASTER-based thermal quartz and carbonate mapping demonstrated here can be used in well-studied or frontier areas to verify the accuracy of existing geologic maps, guide future detailed stratigraphic and structural mapping in lithologically complex terranes and allochthons, and identify hydrothermal features for exploration and resource assessment purposes.


The Great Basin physiographic province is the world's second leading producer of gold and is also host to a wide variety of ore deposits that contain large resources of silver, base metals, and other important metallic and industrial minerals (Hofstra and Wallace, 2006). Because hydrothermal silicification accompanies mineralization in many metal deposits, the identification and mapping of quartz in rocks composed mainly of other minerals is of great value for exploration and assessments of resource potential. Identification of quartz and carbonate in rocks and unconsolidated surficial materials across large areas holds promise for assessing the potential for industrial rock-mineral resources including aggregate, sand, gravel, caliche, and calcrete. The regional mapping of these minerals also facilitates evaluation of existing geologic maps and recognition of locales in need of more refined mapping or topical investigation.

Data from spaceborne remote sensing systems have been applied to prospecting in terranes with requisite geologic and tectonic frameworks for decades, but the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor aboard the Earth Observing System (EOS) Terra satellite launched in 2000 provides revolutionary new capabilities for cost-effective mineral and landcover mapping over large areas. The ASTER sensor acquires multi-spectral data in 14 bands, including 5 bands in the mid-infrared, or thermal (8–14 µm), region of the electromagnetic spectrum, in addition to another band with backward-looking geometry in the near-infrared (NIR, 0.76–0.86 µm) to provide capability for stereoscopic observation and digital elevation model generation. The multi-spectral thermal infrared (TIR) data of ASTER are vital for detecting nonhydrous varieties of quartz that are not spectrally identifiable in the visible, near-infrared, and shortwave-infrared (SWIR, 1.4–2.5 µm) spectral regions because of a lack of diagnostic absorption features. The ASTER TIR data also provide capability for the remote identification and mapping of areally extensive occurrences of other minerals, including carbonates, although the ASTER SWIR data are more sensitive at detecting hydrous quartz (e.g., opal and chalcedony) and differentiating carbonate mineral species. Despite its potential utility, the mineral mapping capabilities of ASTER TIR data are often underutilized in scientific studies.

This study evaluates the utility of band ratio analysis techniques applied to ASTER thermal infrared emissivity data for the identification and mapping of quartz and carbonate minerals across a 3.5° × 4.5° (389 km × 387 km) area of the central Great Basin in northern Nevada. The quartz and carbonate mineral maps were overlain with geologic maps of a wide range of scales (1:500,000–1:6000) to evaluate their reliability and to identify previously unknown occurrences of quartz and carbonate to guide future investigations. Of particular relevance to metallogeny was evidence of hydrothermal quartz, including jasperoids or other quartz associated with argillic and phyllic alteration, in areas of known mineralization. Also evaluated was the utility of the mineral maps for identifying quartz and/or carbonate anomalies within complex, deformed terranes for which only coarse-scale geologic mapping is available. Such anomalies can enhance understanding of these terranes by delineating stratigraphic marker horizons and structural features in need of more detailed geologic mapping.


Quartz and carbonate minerals are spectrally characterized by strong vibrational absorption features within the 8–14 µm atmospheric window (Salisbury and D'Aria, 1992a; Hook et al., 1999) that is measured by the five thermal infrared bands of the ASTER sensor. 01Table 1 shows the wavelength centers and band widths of the five ASTER thermal bands, which have a ground instantaneous field of view (GIFOV), or ground spatial resolution, of 90 m. Figures 1 and 2 exhibit the spectral features of quartz, calcite, and dolomite in this spectral region at original laboratory resolution and convolved to ASTER sampling and bandpass 01(Table 1). The emissivity spectra shown in these figures were created by applying Kirchoff's Law, E = 1 – R, where E and R are emissivity and reflectance, to reflectance spectra from a mid-infrared laboratory spectral library (Salisbury et al., 1991). Kirchoff's Law inverts reflectance peaks into emissivity troughs commonly referred to as emissivity “absorption” features. As the laboratory spectra were measured in biconical, rather than hemispherical, reflectance, the generated emissivity spectra can be used to predict only the spectral shapes, and not absolute emissivity values, of the mineral spectra. In both Figures 1 and 2, the emissivity spectra convolved to ASTER sampling and bandpass are shown in red. In Figure 1, the emissivity absorption features of quartz at ASTER bands 10 and 12 are related to fundamental asymmetric Si-O stretching vibrations (reststrahlen bands). The reststrahlen bands of quartz are the strongest of any silicate mineral (Salisbury and D'Aria, 1992b). In Figure 2, the emissivity absorption features of calcite and dolomite at ASTER band 14 are related to out-of-plane bending modes of the CO3 ion (Clark, 1999). Note that dolomite exhibits a greater decrease in emissivity between bands 13 and 14 than calcite. This characteristic is caused by the greater width and shorter wavelength position of the bending feature of dolomite at 11.15 µm relative to the bending feature of calcite at 11.27 µm.

In the 8–11 µm spectral region measured by ASTER, some fabricated construction materials such as paving concrete and asphalt (macadam) have absorption features that are similar to that of quartz at ASTER sampling and bandpass (ASTER Spectral Library, 2002), possibly due to quartz sand in the aggregate used to make these materials. Aggregates are often either quartz (or granite, which is dominated spectrally by quartz) or limestone, or a mixture of the two. As concrete and macadam weather, the cementing matrix becomes progressively less exposed, and the aggregate more so. As a result, the TIR emissivity signatures of concrete and macadam will become more similar with age (J. Salisbury, 2007, personal commun.). Quartz signatures (and those of limestone) have been identified elsewhere in asphalt road aggregate using the Spatially Enhanced Broadband Array Spectrograph System (SEBASS), an airborne hyper-spectral thermal imaging system (Kirkland et al., 2002). Many soils contain abundant quartz (e.g., ultisols), and thus will also have TIR emissivity signatures similar to that of quartz (Salisbury and D'Aria, 1992b). Therefore, surface features other than quartz-bearing rocks may be identified as quartz using ASTER thermal data.

Salisbury and D'Aria (1992b) described how grain size, soil moisture, and mineral mixtures can have significant effects on the shape of the TIR emissivity spectra of rocks and soils. In quartz-bearing soils, coarse grain sizes show much larger increases in emissivity between ASTER bands 12 and 13, and between ASTER bands 10 and 11, than do finer grain sizes. Dry quartz-bearing soils show larger increases in emissivity between ASTER bands 12 and 13 than do moist soils. Mixtures of quartz with other minerals can also affect spectral shape. Figure 3 shows emissivity spectra of several minerals common in rocks and soils that, if present in sufficient amounts, can obscure the presence of quartz or carbonate. The presence of these minerals in sufficient concentrations will decrease the prominence of the diagnostic quartz emissivity peak at band 11 or create an emissivity absorption feature at those wavelengths. The presence of montmorillonite, kaolinite, or muscovite, among other minerals, will result in a substantial decrease in emissivity between bands 10 and 12. Minerals such as phyllosilicates (clays and micas), carbonate, and sulfates (including gypsum) can be accurately identified using ASTER data of the SWIR spectral region (1.4–2.5 µm; Rowan et al., 2005; Rockwell et al., 2006). Therefore, ASTER SWIR data (30 m GIFOV) can be used to identify many rock-forming minerals that will affect the TIR spectral response.

ASTER Level 2 surface emissivity data products (AST_05) from the EROS Data Center are produced from Level 1B data using MOD-TRAN-based atmospheric correction and a temperature-emissivity separation (TES) algorithm developed by Gillespie et al. (1998, and references therein). The Level 1B data were processed from the original Level 1A format by the Earth Remote Sensing Data Analysis Center (ERSDAC) of Japan. A total of 108 ostensibly cloud-free scenes of ASTER emissivity data acquired between 1 May and 30 September 2000–2006 were obtained from EROS and resampled to the geographic projection using a nearest-neighbor kernal based on satellite ephemeris data contained within the data files. The May–September dates were chosen to avoid snow cover and assure highest possible solar elevation angles, thus maximizing solar illumination.

Based on the TIR emissivity signatures of quartz and carbonate minerals shown in Figures 1 and 2, indices using compound band ratios were developed to identify these mineral groups.

Equation 1 is the index formula for quartz, and equation 2 is the index formula for carbonate.

Equation 1 (quartz index):  
Equation 2 (carbonate index):  

This quartz index exploits the emissivity absorption features at ASTER bands 10 and 12 relative to bands 11 and 13. Expression (a) of the index will yield bright values for areas that are high in band 11 relative to the sum of bands 10 and 12, and expression (b) will yield bright values for areas that are high in band 13 relative to band 12. The quartz index is the product of these two expressions, both of which yield high values for quartz-bearing surfaces. The utility of the type of compound band ratio used in expression (a), in inverted form, for isolating specific absorption features in imaging spectrometer data was described by Crowley et al. (1989). A different quartz index that exploits the same spectral features as expression (a), and the carbonate index of equation 2, was described by Ninomiya et al. (2005) for application to ASTER TIR radiance-at-sensor data. High digital number (DN) values for these indices indicate spectral signatures similar to those of the particular mineral group they were designed to map, and should indicate the presence of quartz or carbonate if the ASTER spectral signatures are unambiguous with those of other surface materials. Given the greater decrease of emissivity between bands 13 and 14 for dolomite relative to calcite, dolomitic rocks should exhibit higher carbonate index values than limestone and other calcite-bearing surfaces. The indices were applied to the 108 scenes of ASTER emissivity data on a scene-by-scene basis using automated batch processing.

The resultant mineral index maps were then histogram matched and mosaicked into single coverages, which were then georeferenced to the Universal Transverse Mercator (UTM) projection using a rubber-sheeting transformation and control points from an orthocorrected coverage of Landsat Thematic Mapper (TM) imagery acquired from the U.S. Geological Survey Seamless Data Distribution system. Control points from vector geology coverages (Ludington et al., 2005; Crafford, 2007) were used in some areas where more detailed control was required. ASTER scenes acquired with significantly off-nadir viewing geometries (>∼6°) have higher positional errors in the ephemeris data than do scenes acquired with near-nadir geometries, requiring the use of the rubber-sheeting, rather than polynomial, transformation. Rubber sheeting is based on triangle-based finite element analysis (Watson, 1992). Because of the variable positional accuracies of the satellite ephemeris data, the 90 m GIFOV of the ASTER thermal data that made control point selection difficult, and the nature of the rubber-sheeting transformation, geometric distortions may be locally present in the mineral index maps. Several small and isolated cumulus clouds were masked from the carbonate mineral index map.

To determine the minimum DN levels corresponding to accurate detections of quartz or carbonate by the mineral indices, the georeferenced indices were overlain with 1:24,000 scale vector-based geologic map data of the Alligator Ridge and Bald Mountain gold districts in White Pine County (Nutt, 2000; Nutt and Hart, 2004). These districts are known to contain large jasperoids of multiple ages, hydrothermal alteration, and a wide range of lithologies including abundant carbonate rocks (Nutt and Hofstra, 2003, 2007). Minimum DN value thresholds were chosen for each index at the level at which pixels began to occur outside of geologic units known to contain the indexed mineral. To verify the accuracy of the thresholds, ASTER emissivity spectra of surfaces with high index DN values (Fig. 4) were extracted from the original data to determine their similarity with the reference laboratory spectra (Figs. 1 and 2). Figure 4 shows that the mid-infrared spectral features of quartz and carbonate are present in the selected spectra, and that there is little consistent variation in spectral shape between quartz-bearing units of sedimentary and hydrothermal origin (e.g., jasperoids). The populations of index pixels selected as accurate mineral detections showed strong spatial contiguity with each other. Spatial contiguity thus served as a secondary, qualitative means of choosing index detection thresholds. ASTER emissivity spectra of sand dunes and hot springs that were identified as either quartz or carbonate bearing elsewhere in the study area were also extracted to determine if the thresholds were accurate across the entire study area (Fig. 5). Such spectra also corresponded well with reference laboratory spectra. Index pixels with DN values below the detection thresholds were set to zero, and color tables were applied to index pixels above the thresholds. Pixels with the highest index values for quartz and carbonate were set to brightest red and green, respectively. Successively darker shades of these colors were assigned to pixels with decreasing DN values. The gradational color tables enable the interpretation of the index level for each pixel selected as a mineral detection. The index values of a pixel are affected by abundance of that particular mineral or mineral group, degree of exposure relative to soil and vegetation cover, carbonate mineral type (e.g., calcite or dolomite), grain size distribution, moisture content, and/or areal or intimate mixtures with other minerals or materials.


General Characteristics of Mineral Index Maps

Figure 6 shows an overview of the quartz and carbonate mineral index maps generated from the ASTER thermal infrared data. The index maps are provided here in GeoTIFF raster format (quartz map, carbonate map).1 Although the results show general consistency across the study area, noise and inter-scene variations in index levels are locally present. The carbonate index map is subject to both random speckle noise and striping in the across-track (west-northwest–east-southeast) direction, as expected using a density-sliced ratio of two spectral bands. The speckle effects are caused by low signal-to-noise levels of the data enhanced by the ratioing process, and the striping is caused by interdetector performance variation within the 10 detector arrays for each TIR band in the whiskbroom scanning system. The index maps contain some sharp breaks in index values across boundaries of scenes acquired on different dates, especially in the northwest quadrant of the area near long 117°W. These breaks occur most often along scene boundaries oriented in the along-track (north-northeast–south-southwest) direction. The breaks are most likely caused by subtle, uncorrected atmospheric and/or soil moisture variation between adjacent scenes, possibly exacerbated by potential errors in the automated histogram matching process used to create the mosaic of mineral index maps. Inter-scene variability related to the ASTER instrument, solar irradiance, and/or atmospheric and surface scattering effects was described by Hewson et al. (2005) as related to difficulties in generating seamless geological maps from data acquired on different overpass dates.

Plate 1 shows the mineral index maps in vector format overlain on a background of Landsat TM data. The index maps were simplified prior to vectorization using a 3 × 3 majority filter. The locations of important ore deposits are indicated by deposit type and commodity. Hot springs are also shown. This map enables interpretation of surface characteristics in areas identified as either quartz or carbonate. Many large areas disturbed by active mining have high quartz index values (Fig. 6), especially along the Carlin Trend (CN), the Independence Trend (IN), the Twin Creeks mine at the northeast end of the Getchell Trend (GT), the Lone Tree (LT) and Marigold (MG) mines along the Battle Mountain–Eureka Trend (BM/EK), the Florida Canyon (FC) gold mine in Humboldt County, and the Robinson copper district (RO) near Ruth and Ely. Tailings deposits and heap leach piles at many mines show high quartz index values. Generally, quartz is very abundant in these areas, given the quartz-rich nature of the ore-bearing rocks. In contrast, along most of the Battle Mountain–Eureka Trend southeast of the Marigold mine, and at Round Mountain (RM, Fig. 6), the principal mines are characterized by only localized areas of quartz and carbonate. Vegetation cover (shown in shades of blue) within areas identified as quartz or carbonate ranges from none to moderate density. Abundant vegetation and/or poor exposure commonly preclude remote mineral detection. For example, patchy occurrences of quartz were identified along the western flanks of the Diamond Mountains (DM, Fig. 6) in eastern Eureka County within the Upper Mississippian Diamond Peak Formation, a conglomerate with clasts of chert and quartzite that consistently shows strong quartz index values across the eastern half of the study area. Quartz was identified principally on south-facing slopes of east-striking drainage divides in this area. South-facing slopes typically have significantly less vegetation cover than north-facing slopes in the Northern Hemisphere due to increased annual solar illumination.

Based on typical compositions of the lithologies and features identified with the ASTER TIR data, we estimate that the minimum detection limits of mineral abundance for the ratio-based methodology presented here is ∼70% of bulk composition for quartz and 85%–90% for carbonate.

Map of Mineral Indices Overlain on 1:500,000 Scale Geology

Plate 2 shows the mineral index maps in the context of the statewide geology of Ludington et al. (2005). Color coding of the geologic units is based on the primary lithology attribute of the vector geology coverage, and secondary lithologies from the coverage are listed in the explanation in parentheses. Quartz identified using the ASTER data corresponds well with units described as primarily quartzite (colored gold) and chert (colored black), and in alluvium and/or colluvium derived from them. Identified carbonate corresponds well with both limestone (colored blue) and dolomite (colored cyan) units, with the highest index values most often corresponding with dolomitic units, as expected given the greater decrease in emissivity between ASTER bands 13 and 14 for dolomite (Fig. 2). Many of the mapped quartz and carbonate anomalies are also located within units described as primarily shale or felsic volcanic rocks. Most, but not all, of these anomalous occurrences of quartz and carbonate reflect the coarse scale and generalized descriptions of the primary lithologies of the geologic units in the statewide geology coverage. Units described as primarily chert, especially within the Roberts Mountain allochthon, contain units of quartzite, chert, and shale, and the mapped quartz most often corresponds to the quartzite units within these complex terranes. The quartz-rich Diamond Peak Formation has been grouped with associated shale units in the statewide geology coverage (e.g., along the west flank of the Diamond Mountains mentioned above). Several occurrences of identified carbonate within units described as felsic volcanic rocks also reflect the broad nature of the described primary lithologies. For example, within the Sand Springs Range in Churchill and Mineral Counties east and north of the Rawhide Ag district (RH, Fig. 6), mapped carbonate corresponds to shelf limestones intercalated with volcaniclastic rocks. The mineral maps demonstrate both the merits and problems inherent in such small-scale lithologic maps with heterogeneous units. Hence, lithostratigraphic (rather than chronostratigraphic) geologic maps of the largest possible scale should be used when interpreting mineralogic information derived from remote sensing data.

Map of Mineral Indices Overlain on 1:250,000 Scale Geology

Plate 3 shows the mineral index maps overlain with 1:250,000 scale geology from Crafford (2007) derived from county geologic maps of the same scale. Polygon outlines of geologic units have been color coded to reflect lithologies relevant for interpreting the mineral index maps. This color coding was based upon descriptions of the regional geologic units developed by Crafford (2007), and other information from the county maps. The color coding should only be used as an approximate guide to lithology, particularly for the coding related to units containing quartzites and/or conglomerates with quartzite clasts (colored magenta). Although units composed primarily of chert have been color coded in purple, minor chert units may also be contained within the quartzite and/or conglomerate grouping.

The more detailed scale of the Crafford (2007) geology coverage permits finer interpretation of the remote sensing results. Excellent correlation of the mineral index maps with units known to contain quartz or carbonate is shown in 201Tables 2202 and 033. For example, Figure 7 exhibits the accurate mapping of quartz within units of quartzite (metamorphosed Ordovician Eureka Quartzite, Ocqm) in the Wood Hills of Elko County, and foliated metaquartzite (in part Precambrian to Cambrian Prospect Mountain Quartzite, €Zqm) in the nearby East Humboldt Range. Figure 7 also shows how carbonate was detected within lower Paleozoic calcite marbles (O€cm), dolomite and graphitic marbles (DOcm), the Ely Springs Dolomite and Hanson Creek Formation (SOc), and limestones of the Devils Gate and Guilmette Formations (Dc). Note how the highest carbonate index values and most complete coverage of mapped carbonate pixels spatially correspond with dolomitic rocks. This effect is partly due to better exposure of the dolomitic units caused by lower densities of vegetation cover than present on calcite-bearing rocks (Plate 1), and its spectral response relative to calcite described above.

The utility of the ASTER-derived mineral index maps for guiding future detailed geologic mapping in complex and/or undifferentiated terranes is illustrated in Figures 8 and 9. Figure 8 shows how the detail of geologic mapping in Ruby Mountains (Coats, 1987) is significantly greater than in the East Humboldt Range to the northeast. The Ruby Mountains are underlain by late Precambrian to early Paleozoic metasedimentary rocks, mainly calcite marbles (O€cm) and metaquartzites (€Zqm). In contrast, the East Humboldt Range is underlain by an undifferentiated metamorphic unit (TAgn) consisting of gneiss, schist, and migmatite of a wide range of ages from Archean to Oligocene. Quartzite and quartzitic schists within this unit are delineated by the quartz index map, as are the metaquartzites in the Ruby Mountains. Carbonate minerals were detected in only a small portion of the calcite marble in the Ruby Mountains. These areas, some of which had high carbonate index values in discrete patches, could correspond with either very thick bedded, well exposed, or more dolomitic sections of the marble unit.

The geology of the Independence Mountains (F, Fig. 6) is shown in Figures 9 and 10. The Ordovician McAfee Quartzite and Cambrian Prospect Mountain Quartzite are highlighted by the quartz mineral index map (Fig. 9). Within the Independence Mountains, the McAfee Quartzite is part of a complex, undifferentiated lower Paleozoic unit (D€s) consisting of shale, chert, quartzite, greenstone, and limestone that composes the Roberts Mountain allochthon of the Mississippian Antler orogeny in this area. The D€s unit is part of the Basin assemblage of Crafford (2007). Similarly, within the undifferentiated, allochthonous Golconda terrane (GC) of Crafford (2007) located in the northwestern Independence Mountains, the quartz mineral index map differentiated units of chert and/or argillite within the Devonian–Permian Schoonover Sequence. Occurrences of quartz in this area parallel the southwest-northeast strike of numerous thrust faults formed during the Permian–Triassic Sonoma orogeny (Miller et al., 1984).

An anomalous occurrence of quartz was identified in the vicinity of Coffin Mountain in the southern Piñon Range in Elko County (40°16′02″N, 116°00′19″W; L, Fig. 6; 201Table 2202). The anomaly consists of two lobes that together are >6 km long from north to south, and as much as 3 km wide. On the 1:250,000 scale map of Crafford (2007), the area is underlain by unit Dcd, consisting of dolomite, sandstone, and limestone, and the county-level data for the area lists Sevy, Simonson, and Nevada Formations, all Devonian units of primarily dolomitic composition. Inspection of a 1:62,500 scale map of the area (Smith and Ketner, 1978) revealed that the mapped quartz corresponds to the Oxyoke Canyon Sandstone Member of the Nevada Formation, the only sandstone in the autochthonous Devonian Carbonate Shelf sequence of Crafford (2007). Although the bulk of the mountain is underlain by the Beacon Peak Dolomite Member (Dnb) of the Nevada Formation, the Oxyoke sandstone forms the mountain summit and also occurs in a crude ring 1–2 km away from the summit to the north, east, and south. Quartz-bearing colluvium and/or float derived from the Oxyoke sandstone at the summit of the mountain conceal exposures of Beacon Peak Dolomite on sections of the western and eastern mountain flanks. Significant carbonate was detected within the Upper dolomite member (Dnu) of the Nevada Formation ∼1.6 km southwest of the summit of Coffin Mountain, and within the Beacon Peak Dolomite, which is well exposed 1.5 km south of the summit. High quartz index values correspond to thick, ridge-forming exposures of the Oxyoke sandstone, and lesser index values correspond to colluvium and float derived from them. This example demonstrates how large-scale lithostratigraphic maps are required to accurately interpret the results of the ASTER thermal infrared mineral mapping, and, conversely, how the mineral index maps can serve as a cost-effective means of both guiding and verifying geologic field mapping.

Hydrothermal Quartz and Carbonate

Plate 3 shows that there is a high correlation between identified quartz and jasperoids of uncertain age and undefined relation to ore deposits (outlined in orange). Such correlation is especially high in White Pine County, where jasperoids were well mapped and differentiated by Hose et al. (1976). Below, mineral indices are compared to geologic maps of various scales (1:250,000–1:6000) from mining districts and geothermal areas that are known to contain large exposures of hydrothermal quartz. In a few areas, detected quartz prompted literature investigations of known mineralization. Sizable occurrences of quartz formed by hydrothermal processes were identified that are associated with Carlin-type Au (Hofstra and Cline, 2000; Cline et al., 2005), distal-disseminated Au-Ag (Cox, 1992; Theodore, 1998), and epithermal high- and low-sulfidation Au-Ag (John, 2001) deposits. Siliceous sinters associated with modern hot springs (Shevenell and Garside, 2005) were also mapped (Fig. 11).

Other than several occurrences of travertine associated with hot springs (Fig. 11), carbonate of hydrothermal origin is difficult to recognize on the carbonate index map because there is either an insufficient percent and volume of carbonate present to be detected, or it occurs within carbonate sedimentary rocks such that there is no spectral contrast. In volcanic rocks, calcite-bearing propylitic alteration has not yet been conclusively identified using ASTER TIR data, as the calcite is typically present in only low abundance and mixed with other minerals such as chlorite and epidote. In the TIR, chlorite and epidote are characterized by a decrease in emissivity between ASTER bands 11 and 13, and a slight increase in emissivity between bands 13 and 14 (Salisbury et al., 1991). These spectral properties are similar to those of mafic and/or ultramafic minerals and suggest that the compound ratio (band 11 + band 12 + band 14)/(band 13 × 3), or previously developed indices for mafic and/or ultramafic minerals including (band 12 + band 14)/(band 13 × 2) (Rowan et al., 2005) and (band 12 × band 14)/(band 13) (Ninomiya et al., 2005), can be used to isolate chlorite and epidote from carbonate minerals in areas with very sparse to nonexistent vegetation cover and limited weathering, and where the minerals are not intimately mixed. Given the spectral resolution of ASTER TIR data, it is unlikely that they can be used to accurately differentiate chlorite and epidote from mafic and/or ultramafic minerals such as hornblende, actinolite, tremolite, biotite, olivine, augite, and diopside. It is also possible that ASTER visible, NIR, and SWIR data can be used to differentiate chlorite and epidote (as a group) from carbonates by exploiting the deep absorption caused by ferric/ferrous iron centered around 1.0 µm that is present in chlorite and epidote (and many mafic minerals) but not in pure carbonates (Hewson et al., 2005). In other areas, propylitic calcite and epidote have been identified using ASTER SWIR data (Rockwell et al., 2006), and are readily discernable with spectroscopic data such as AVIRIS (Airborne Visible/Infrared Imaging Spectrometer) (Rockwell et al., 2000, 2005; Rowan and Mars, 2003; Cunningham et al., 2005), HyMap, or HyperSpecTIR, although intimate mixing of these minerals can cause difficulties (Dalton et al., 2004). In Paleozoic carbonate rocks, the distribution of hydrothermal dolomite occurrences can be used to map flow paths of hot basinal brines that generate SEDEX (sedimentary exhalative) and MVT (Mississippi Valley type) deposits (e.g., Diehl et al., 2005). Dolomite is nearly impossible to differentiate from its host using ASTER TIR data alone. Since dolomite usually exhibits higher carbonate index values than calcite, it is possible that dolomitization within a limestone unit could be recognized as discrete areas of high carbonate index values, although no examples have yet been found. ASTER SWIR data are better suited the problem and have been used to distinguish calcite from dolomite (Rowan and Mars, 2003).

Carlin-Type Au Deposits

Many Carlin-type gold deposits are associated with jasperoids formed by pervasive silicification of carbonates, shales, or conglomerates (Hofstra and Cline, 2000). Some of the largest jasperoids are associated with gold deposits in the Alligator Ridge trend (Nutt and Hofstra, 2003) and neighboring deposits in east-central Nevada (Maher, 1997). In these deposits, the largest jasperoids are stratabound and are frequently present in the Chainman Shale, Joana Limestone, Pilot Shale, and Devonian Guilmette Formation (Dg) limestone, and especially at contacts between these formations. The Jerritt Canyon district, in the Independence Mountains, also has some large jasperoids, especially in the lower part of the Silurian and Devonian Roberts Mountains Formation and upper part of the Ordovician and Silurian Hanson Creek Formation (Hofstra et al., 1999). In the largest gold trends (Carlin, CN, Fig. 6; Battle Mountain–Eureka, BM/EK; Getchell, GT), jasperoids are generally smaller, but the largest jasperoids present also are typically developed in certain limestone units and along shale-limestone contacts. Consequently, quartz identified using ASTER TIR data in the above-mentioned units is likely to be jasperoid. Several examples from these trends and districts are described below.

In the adjacent Bald Mountain and Alligator Ridge districts (Fig. 12), jasperoids of four different ages have been recognized: (1) small stratabound jasperoids hosted in Cambrian–Ordovician Formations that predate Jurassic intrusions, (2) small discordant jasperoids associated with Jurassic reduced intrusion-related gold deposits hosted in Cambrian–Mississippian Formations, (3) large stratabound jasperoids associated with Eocene Carlin-type gold deposits hosted in Devonian–Mississippian Formations, and (4) large jasperoids hosted in a chaotic breccia unit of Eocene–Miocene age (Nutt and Hofstra, 2003, 2007; Nutt and Hart, 2004). Of these, only the latter two are large enough to discern between areas disturbed by mining using the ASTER TIR data. Those associated with Carlin-type deposits, such as Winrock (W, Fig. 12), Galaxy (G), Horseshoe (H), Saga (S), and others [JB(1–3) and J1, Fig. 12], are localized by north-striking faults along Mooney Basin and intersecting northwest-striking faults of the Bida trend (BT, Fig. 12). The Bida trend also served to localize the Jurassic Bald Mountain stock and related deposits. Jasperoids exposed in and adjacent to the Carlin-type deposits are hosted in Chainman Shale (Mch of Nutt and Hart, 2004), Joana Limestone (Mj and jasperoid Mjj), and Pilot Shale (MDp). In the Bida trend, between the Eocene Horseshoe deposit and the Jurassic Top deposit (T, Fig. 12), quartz was identified on a low ridge in an area underlain by Pilot Shale and Chainman Shale (J2, Fig. 12). ASTER SWIR data indicate that these rocks have been highly argillized to kaolinite. The large size and typical host of the J2 quartz anomaly suggest that it may be a remote manifestation of the Eocene Carlin-type hydrothermal system that extends into the older Bald Mountain district. Farther south, between the Vantage and Yankee deposits, a large jasperoid (J3, Fig. 12) hosted in the Guilmette Formation (Dg, limestone and dolomite), Pilot Shale, Joana Limestone, and Eocene conglomerate (not shown) was identified. ASTER SWIR data show that some of the jasperoids in the Pilot Shale are accompanied by significant argillization (mainly kaolinite), while in adjacent areas sericite (most likely muscovite) was identified. A large quartz anomaly was mapped on the west side of the Bald Mountain district in the chaotic breccia unit [JB(4), Fig. 12]. Although some of this quartz could be due to brecciated Diamond Peak Formation and silicified Chainman Shale, jasperoid in the hanging wall of the north-northwest–striking Ruby listric normal fault is likely Miocene in age (C. Nutt, 2007, personal commun.). An ASTER emissivity spectrum for an exposure of the jasperoid in Water Canyon shows a strong quartz signature (Fig. 4).

In the area surrounding the Illipah Au deposit at the north end of the White Pine Range (J, Fig. 6), the large jasperoid hosted in Eocene conglomerate and underlying Joana Limestone, described and photographed by Nutt and Hofstra (2003), was mapped using the ASTER TIR data (Fig. 13). This jasperoid forms the cap rock of the flat-topped ridge 2 km southeast of the Illipah mine. High quartz indices correspond well with the mapped jasperoid in that area from Crafford (2007). South of the jasperoid, the ridge is underlain by unaltered Joana Limestone (Mj within the composite MDcl unit of Crafford, 2007) that was identified as carbonate. In this area (Fig. 13), the Diamond Peak Formation (Mdp) consistently shows moderate to high quartz index values.

In the Easy Junior mining district (E, Fig. 6), the same composite Devonian to Mississippian MDcl unit hosts a series of jasperoids mapped by Crafford (2007, br, outlined in orange in Plate 3) and Carden (1991) that were all mapped using the ASTER TIR data (Fig. 14). Likewise, the Devonian Devils Gate Limestone (Dd) within the Dc unit of Crafford (2007) was identified as carbonate 03(Table 3).

In the Independence Mountains (F, Fig. 6), a massive jasperoid hosted in interbedded chert and carbonate rocks of the Ordovician–Silurian Hanson Creek Formation (unit 1, Hofstra et al., 1999) was mapped (Fig. 9). While this ridge-forming jasperoid (Fig. 10) would be considered to be large by most economic geologists, it is relatively small in terms of the ASTER TIR data (16 90 m pixels identified as quartz, including surrounding float). Most of the other large jasperoids associated with Carlin-type Au deposits in the Jerritt Canyon district were detected in areas disturbed by mining, as discussed later.

In the Carlin Trend, quartz was mapped in the Silurian–Lower Devonian Roberts Mountain Formation 3.3 km north of the main open pit of the Gold Quarry mine along the northeast side of Maggie Creek Canyon (GQ, Fig. 6; Plates 1–3). This northwest-southeast–trending zone of quartz appears to have high albedo suggestive of bleaching, and corresponds in part to hydrothermal alteration mapped along a thrust fault of similar strike (Fig. I-10 in Harlan et al., 2002). The composition of the Roberts Mountain Formation ranges from a silty limestone to a calcareous siltstone, but no carbonate was detected on this outcrop using the ASTER data. Where decalcified, the rocks consist mainly of fine-grained, platy siltstone with high quartz content. Therefore, these outcrops with low to moderate quartz index values may be decalcified and/or silicified. Smaller quartz-bearing areas with low index values also were detected in an area of mapped alteration 1 km to the east, in the vicinity of the Rainbow deposit (Harlan et al., 2002).

Pluton-Related Distal-Disseminated Deposits

Silicification was generally not detected along pluton contacts (Plate 2) or near proximal pluton-related ore deposits (Plate 1). However, small jasperoids are frequently present in or near open pit mines that produce gold from distal-disseminated deposits hosted in sedimentary rocks (e.g., Robinson district, Maher, 1996; Lone Tree and Marigold mines in the Battle Mountain district, Theodore, 2000; Bald Mountain district, Nutt and Hofstra, 2007). Quartz was detected in each of the above mining areas, but it was difficult to find exposures of jasperoid in adjacent areas that were undisturbed. For example, Maher (1996) mapped hydrothermal silicification within the marble halo at the Robinson porphyry Cu-Au deposit near Ely in White Pine County, but we were unable to spatially discriminate quartz in outcrop from quartz in the pits and tailings. At Lone Tree (LT, Fig. 6) and Marigold (MG), the deposits are hosted in Havallah Sequence, Edna Mountain Formation, Battle Formation, and Valmy Formation that contain quartz sandstones, siltstones, and chert, making it nearly impossible to distinguish hydrothermal quartz. At Bald Mountain (Fig. 12), most of the discordant jasperoids associated with the Jurassic distal disseminated deposit are small, difficult to resolve, and only a few occur outside areas disturbed by mining. Quartz with moderate index values (SS1, Fig. 12) was identified on the summit ridge of Big Bald Mountain several kilometers north of the Top Pit area within the Ordovician Pogonip Group. At this locality, the Pogonip is composed primarily of silty limestone with thin beds and lenses of resistant quartz sandstone (C. Nutt, 2007, personal commun.). Hence, the identified quartz likely corresponds to the sandstone, although silicification and/or decalcification may also be present.

The small Kinsley mining district in eastern Elko County (C, Fig. 6) has small exposures of jasperoid that occur both within and outside the pits. Comparison with the geologic map from Lapointe et al. (1991) shows that both mined and undisturbed jasperoids were successfully identified using the ASTER data (Fig. 15). This example also shows that tiny jasperoids, such as those located 2–3.5 km north-northeast of the main mining area (“Main Zone”), are too small to be resolved with the ASTER TIR data. Quartz was also identified in a heap leach pile located 2 km east of the main mining area. The Eocene pluton inferred to be responsible for the mineralization is located 3.2 km south of the main mining area along the southern edge of the maps shown in Figure 15.

High-Sulfidation Au-Ag Deposits

In the study area, the only high-sulfidation ore deposits of economic significance associated with magmatic-hydrothermal acid-sulfate alteration in volcanic rocks are the Paradise Peak Au-Ag-Hg deposit (K, Fig. 6) in Nye and Mineral Counties and the Santa Fe Au-Ag-Cu-Pb-Sb-W district in Mineral County. Deposits of this type are restricted to the western andesite assemblage of the Great Basin (John, 2001). Mineralization in the Paradise Peak area is most likely the result of two hydrothermal systems (John et al., 1989; Sillitoe and Lorson, 1994). The earlier of these systems created porphyry-type quartz stockworks enveloped by quartz-sericite-pyrite alteration with Au ca. 22 Ma in the County Line and East Zone deposits in the western section of the area. A later system formed intense silicic (quartz-pyrite ± opal), quartz-alunite, argillic (quartz + clays), and propylitic (chlorite + clays) alteration with Au and Hg ca. 19–18 Ma in the Paradise Peak, Ketchup Flat, Ketchup Knob, and Ketchup Hills deposits in the eastern section of the area. Figure 16 shows that quartz formed by both systems was identified using the ASTER TIR data. The highest quartz index values correspond to tailings deposits just south of the Paradise Peak, Ketchup Flat, and Ketchup Hill deposits to the east. Moderate quartz index values are associated with the quartz-alunite (QA, Fig. 16) alteration within the Middle Tuffs (Tmt, 23 Ma) on and around Newman Ridge in the central part of the district. The quartz-sericite-pyrite alteration (QS, Fig. 16) is expressed by low quartz index values and emissivity features suggestive of a mixture of quartz and illite or muscovite (i.e., reduced prominence of emissivity peak at ASTER band 11 and greater decrease in emissivity between bands 10 and 12 relative to quartz, Fig. 3). Quartz-bearing alluvium and/or tailings derived from the County Line deposit (CL, Fig. 16) have been shed to the northwest into Gabbs Valley along Fingerock Wash (Plates 1–3). Silicification was also identified in the Santa Fe district located 20 km southwest of the Paradise Peak area in the Gabbs Valley Range (Fig. 6; Plates 1–3). Silicification identified with the ASTER TIR data that is directly associated with advanced argillic or phyllic hydrothermal alteration is commonly characterized by low to moderate quartz index values, whereas quartzites, sinters, and jasperoids with higher concentrations of quartz commonly show moderate to high index values.

Low-Sulfidation Au-Ag and Hg Deposits

Outside the study area, a large body of relict siliceous sinter associated with the Hasbrouck Peak low-sulfidation Au-Ag deposit in Esmeralda County is readily identifiable with ASTER TIR data. It is therefore possible that areas with high quartz index values in the vicinity of the McDermitt hot spring–type, low-sulfidation Hg deposit (MD, Fig. 6) in Humboldt County are sinters, although the surface disturbance caused by mining makes discrimination of such features difficult.

Silicified rocks associated with low-sulfidation deposits in the study area are for the most part too small to be directly identified using the ASTER TIR data, although several pixels containing quartz may be found that are expressions of more extensive subsurface silicification and/or silicified rocks that have been exposed by mining (e.g., Rawhide Au-Ag district, Mineral County; RH, Fig. 6). An exception is in the well-exposed high Jarbidge Mountains of Elko County (JB, Fig. 6), where abundant quartz was identified above treeline within the Miocene Jarbidge Rhyolite (Plates 1–3; Tr3 of Crafford, 2007). These quartz occurrences are some of the largest associated with volcanic terranes in the study area, and are located several kilometers southeast of the Jarbidge low-sulfidation epithermal Au-Ag district within the bimodal volcanic assemblage. The Jarbidge deposit (ca. 14 Ma) is hosted within rhyolite lava flows and domes, and consists of laminated quartz-adularia-calcite veins, stockworks, and breccias within phyllically altered wallrock (Coats, 1964; Lapointe et al., 1991; Bernt, 1998). The occurrences of quartz within the Jarbidge Rhyolite are highly anomalous, ridge forming, and locally associated with argillic and/or phyllic alteration detected with the ASTER SWIR data. The north-northwest–south-southeast trend of the main quartz occurrence in the Jarbidge Mountains parallels local faults (Plate 2), and corresponds to the “east vein system” of Au- and Ag-bearing quartz-adularia veins described by Schrader (1923) and Coats et al. (1977). Numerous prospect pits and adits are located high on the ridge along the system, principally along shear zones with quartz veins and limonitic staining that can extend for as much as 900 m. The size of the quartz anomaly is increased by colluvium, landslide deposits, and glacial moraines derived from the altered rocks forming the ridge. While it is possible that the identified quartz is related to highly silicic and perhaps glassy and/or perlitic flows or domes of the Jarbidge Rhyolite, the relationships described above suggest it is likely due to silicification of rhyolite along shear zones, at least in part.

Surficial Geologic Features Containing Quartz and Carbonate

Quartz was identified within playa lake deposits and in sand dunes. While most of the quartz occurrences in and adjacent to playa lakes in the western half of the study area are likely to be thick deposits of beach sand, some may represent sinter deposits or eolian dusts derived from them. For example, excellent correlation exists between the outline of a large playa in Edwards Creek Valley in Churchill County (playa center at 39°38′27″N, 117°39′12″W; Ludington et al., 2005; Crafford, 2007) and quartz mapped using the ASTER data (Plates 1–3). The identification of these sand deposits could be useful for future geomorphology studies and surficial and/or Quaternary geologic mapping. Figure 17 demonstrates how quartz identified in Silver State Valley in Humboldt County corresponds to the Winnemucca Dunes. These dunes were identified as quartz based mainly on the basis of the steep increase in emissivity between ASTER bands 12 and 13 (Fig. 5), as the spectral shape between bands 10 and 12 does not correspond well to that of quartz (Fig. 1). The significant decrease in emissivity between bands 10 and 12 suggests that other minerals, possibly clay minerals and/or gypsum (Fig. 3), are present in the dunes. Abundant loess with clays ± micas derived from deflation of pluvial lakes is present in soils and within the matrix of colluvial deposits in the Osgood Mountains (Getchell Trend, GT, Fig. 6) and exposed in the drainages of Kelly and Evans Creeks to the east (Rockwell, 1991). Sand dunes containing quartz and gypsum are common in areas with abundant pluvial lakes (Chen et al., 1995; Wilkins, 2000). Preliminary analysis of corresponding ASTER SWIR data of the Winnemucca Dunes suggests that montmorillonite and gypsum are present in variable concentrations, with montmorillonite being generally more abundant. The use of only expression (a) of equation 1 as a quartz index would likely eliminate the dunes from the quartz index map, as the TIR spectra of the dunes in bands 10–12 differ substantially from that of quartz, as mentioned above. However, as the emissivity increase between bands 12 and 13 exploited by expression (b) of equation 1 is so strong for most quartz exposures, eliminating expression (b) from the quartz index would likely also result in an index map degraded in terms of overall accuracy and detail.

Carbonate was identified in playas within Buena Vista Valley in Pershing County (BV, Fig. 6) and in the northeastern and eastern parts of Carson Sink in Churchill County (CS, Fig. 6). Fine eolian deposits of silicates (clays and illite/mica), calcite, and soluble salts are present on the Carson Sink surface, and the carbonate and soluble salts increase in abundance downwind, toward the northeast (Chadwick and Davis, 1990). The calcite is most likely an eolian deposit derived in part from Tertiary limestones and/or Quaternary tufa from the upwind areas of the Carson Sink to the southwest, including the Hot Springs Mountains region (Kratt, 2005).

Quartz and Carbonate Related to Anthropogenic Features

In addition to many areas disturbed by mining such as pits, tailings deposits, and heap leach piles, quartz was identified along Interstate 80 (Figs. 7 and 8; Plates 1–3), within most urban areas, and on one of the two intersecting runways of the airport near the town of Battle Mountain (Plate 1). Quartz was not identified along smaller roads because they were unresolvable given the 90 m GIFOV of the ASTER TIR data. It is presumed that the spectral similarity of these features with the reference spectrum of quartz (Fig. 1) is caused by the presence of quartz in aggregate used to make concrete and asphalt (macadam), as described above in Methods. However, it is unknown if uncorrected temperature effects from the TES calibration process (Gillespie et al., 1998) play a role in the detection of quartz in some or all of these features.

Several large occurrences of apparently quartz-bearing soils were identified 40 km north of the town of Battle Mountain and west of Six-mile Hill in western Elko and eastern Humboldt Counties, including the Evans Creek drainage (QS, Fig. 6). The edges of these occurrences are sharp and most are oriented north-south and east-west, indicating that they represent areas that have been cleared of rangeland vegetation, possibly for grazing or agricultural use. The emissivity spectra of these areas are similar to those of the Winnemucca Dunes shown in Figure 5, but most show a slight emissivity peak at ASTER band 11 indicative of quartz. It is possible that clearing has exposed sandy soils containing quartz and loess.

Quartz was mapped at the Colado/Perlite diatomite mine and associated ore storage areas along Interstate 80 4 km south of Woolsey in Pershing County (CP, Fig. 6). Other diatomite mines such as the Trinity and Brady's mines in Humboldt and Pershing Counties (outside of study area) also showed strong quartz index values. Most diatomite deposits occur within the sedimentary unit Ts3 on the statewide geology map (Ludington et al., 2005), although most undisturbed diatomite deposits from the Mineral Resource Data System (MRDS) (2007) showed no appreciable quartz index response. Diatom phytoplankton commonly contain frustules of opaline silica, and thus could be identified in greater detail using SWIR remote sensing data, including those of ASTER.

Quartz was also mapped within rhyolitic units (Tr3 and Tt2, 201Table 2202) locally associated with perlite deposits. In addition to the possibly perlitic zones of the Jarbidge Rhyolite, quartz was mapped within and near the Palisade Canyon perlite prospect (PC, Fig. 6) in Eureka County, and within the Monotony Tuff and Lava, northern Nye County, near the Pamela Placer perlite prospect (PP, Fig. 6; MRDS, 2007).


This research demonstrates several characteristics of ASTER TIR data, the processing and analysis methods applied to them, and the utility of the data for mapping quartz and carbonate minerals. The generally high correlation between identified quartz and carbonate and corresponding lithologies, alteration types, unconsolidated surficial material, and anthropogenically disturbed areas containing these minerals is testament to the radiometric stability and resolution of the ASTER data in this wavelength region. The accuracy of the automated TES algorithm (Gillespie et al., 1998), the strong and unambiguous nature of the mid-infrared spectral features of quartz and carbonate at ASTER sampling and bandpass, and the utility of the ratio-based analysis methodologies presented here to exploit these spectral signatures also contributed to the success of this effort.

These analysis methods are ideal for cost-effective mineral mapping using ASTER data. For example, they can be used as a tool for guiding field mapping and sampling in frontier areas, and as baselines for more targeted airborne imaging spectrometer surveys. For maximum effectiveness, however, spectral analysis of ASTER visible and NIR (for mapping of iron minerals, vegetation, and volatiles including water) and SWIR data (for phyllosilicates, sulfates, carbonates, amphiboles, some sorosilicates, and hydrous silica) should accompany TIR analysis, especially when characterizing hydrothermal alteration. The semiautomated application of these methods to many scenes acquired on different dates, though largely successful, resulted in some sharp inter-scene variations in mineral index values. However, when applied to individual ASTER scenes, the mineral detection thresholds for density slicing the index maps can be further optimized for the atmospheric, soil moisture, noise, viewing geometry, and solar illumination characteristics of that scene.

Because of the low spectral and spatial resolutions of the ASTER TIR data and the nature of mineral spectral features in the mid-infrared, minimum mineral detection limits are high, and thus only the largest and most thickly bedded exposures of quartz and carbonate minerals could be identified. Despite the limitations of resolution, the results show that several different types of hydrothermal systems deposited bodies of hydrothermal quartz of sufficient size and mineral abundance to be detected. Notable among these are Carlin-type gold deposits, distal disseminated deposits, high- and low-sulfidation epithermal deposits, and sinter in geothermal areas. TIR data of increased spectral and spatial resolution would provide results of much greater detail and accuracy, and allow for the identification of a greater variety of pure and mixed minerals and rock types at lower abundances. Airborne thermal sensors such as the MODIS (Moderate Resolution Imaging Spectroradiometer)/ASTER (MASTER) simulator (25 TIR spectral bands, 5–50 m GIFOV; Hook et al., 2001) and SEBASS (256 SWIR and TIR spectral bands, 2+ m GIFOV; Kirkland et al., 2002) currently offer better mineral mapping capability at local scales that ASTER (e.g., Vaughan et al., 2005). SWIR remote sensing data, including those of ASTER, can be used to obtain maps of much greater spatial and mineralogical detail regarding pure and mixed assemblages of carbonate minerals, and hydrous quartz minerals including opal and chalcedony.

This paper also demonstrates that TIR spectral features of quartz formed by hydrothermal and sedimentary and/or metamorphic processes are quite similar at ASTER spectral resolution (Figs. 4 and 5). Variations in emissivity levels and rest-strahlen band depths do not appear to be systematic enough to permit the reliable discrimination of quartz by genetic process type. Recognition of hydrothermal quartz or carbonate minerals is mainly a problem of geologic context in that their identification is facilitated in rocks composed mainly of other minerals. As such, comparisons of mineral index maps with large-scale lithostratigraphic maps can reduce the uncertainty of interpretations. Future experimentation is warranted regarding the use of known exposures of hydrothermal quartz as training sites for supervised classification techniques, including Spectral Angle Mapper (Kruse et al., 1993; Rowan et al., 2005). This analysis approach has the potential for discriminating hydrothermal and sedimentary and/or metamorphic quartz based on statistical measures of overall spectral shape on a scene-by-scene basis.

The ASTER-derived quartz and carbonate mineral maps provide essential, reconnaissance-level information relevant to future assessments of mineral resource potential and detailed geologic mapping of rocks, surficial deposits, and anthropogenic disturbances related to mining and remediation activities. The results from this well-studied area of Nevada demonstrate some of the strengths and weaknesses of existing geologic maps compiled at various scales. The results suggest that much remains to be learned about the distribution and compositions of unconsolidated surficial material in the study area and their relations to exhumation, erosion, and fluvial and lacustrine processes. The ability to distinguish individual quartz- and carbonate-bearing units and features, especially within complex and/or undifferentiated terranes, is of obvious utility for poorly studied or frontier areas where it can both guide field mapping programs and aid mineral resource investigations.

1 If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00126.S4 and http://dx.doi.org/10.1130/GES00126.S5 or the full-text article on www.gsajournals.org to access the quartz (00126_sf01.tif) and carbonate (00126_sf02.tif) index maps in Geo-TIFF file format (for use in GIS systems).

We thank the U.S. Geological Survey (USGS) Land Processes Distributed Active Archive Center (LP DAAC), the National Aeronautics and Space Administration Earth Observing System Data and Information System (EOSDIS), and Japan's Earth Remote Sensing Data Analysis Center (ERSDAC) for providing the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data that were the foundation of this research. We thank Paul Denning of the USGS for providing the Landsat data from the USGS Seamless Data Distribution System, the shaded-relief digital terrain model, and a portion of Elizabeth Crafford's geology coverage of Nevada that was subsequently modified for use in this paper. We also thank Jack Salisbury for information regarding the thermal spectral properties of road surfaces, David John for information on the Paradise Peak deposit and Santa Fe district, and Alan Wallace for data on diatomite deposits. Special thanks to Elizabeth Crafford, Bernard Hubbard, and Alan Wallace for their thorough and insightful reviews of this manuscript.