The U.S. Geological Survey tested the utility of imaging spectroscopy (also referred to as hyperspectral remote sensing) as an aid to regional mineral exploration efforts in remote parts of Alaska. Airborne imaging spectrometer data were collected in 2014 over unmined porphyry Cu deposits in the eastern Alaska Range using the HyMap™ sensor. Maps of the distributions of predominant minerals, made by matching reflectance signatures in the remotely sensed data to reference spectra in the shortwave infrared region, do not uniquely discriminate individual rock units. However, they do highlight hydrothermal alteration associated with porphyry deposits and prospects hosted mostly within the Nabesna pluton. In and around porphyry Cu deposits at Orange Hill and Bond Creek, unique spectral signatures are related to variations in chlorite and white mica abundance and their chemical composition. This is best revealed in the longer-wavelength 2,200-nm Al-OH absorption feature positions in pixels spectrally dominated by white mica proximal to porphyry deposits. Similar spectral signatures of chlorite and white mica wavelength positions were also recognized away from the porphyry deposits; follow-up sampling identified these satellite areas to also contain Cu-Mo-Au mineralized rock. Our study confirms that airborne imaging spectroscopy has application for regional mineral exploration in exposed mountainous terrain in Alaska.


Imaging spectroscopy (also referred to as hyperspectral remote sensing) is a technology that has been utilized with success for mapping vegetation and mineral abundances over many areas of the Earth’s surface (e.g., Clark et al., 2003; Kokaly et al., 2009, 2013). Spectroscopy can be used to identify certain minerals based on their electronic and vibrational absorptions in the reflected solar range (400–2,500 nm). While not effective in identifying many rock-forming minerals such as quartz, feldspars, and pyroxenes, the shortwave infrared (SWIR) region (2,000–2,500 nm) is key to identifying carbonates and hydrous minerals (e.g., micas and clays; Clark, 1999; Thompson et al., 1999; Swayze et al., 2014) that are often products of hydrothermal alteration. Aircraft- or spacecraft-borne platforms are particularly effective for recognizing the spatial distributions of these minerals associated with important mineral deposit types (Crosta et al., 1998; van der Meer, 2006; Hubbard et al., 2007; Bedini, 2012; van Ruitenbeek et al., 2012; Mars, 2013). Most published studies have focused on areas at midlatitudes, with the technology infrequently applied at higher latitudes because data collection and analysis are more difficult due to short collection seasons and low sun angle resulting in poor solar illumination.

Alaska is well known for its precious and base metal mines and mineral deposits. Major mines include the Pogo, Fort Knox, and Kensington Au deposits, the Greens Creek Ag-Au-Pb-Zn volcanogenic massive sulfide deposit, and the Red Dog Zn-Pb-Ag deposit. World-class deposits, including the Donlin Creek Au and the giant Pebble porphyry Cu-Au-Mo, remain undeveloped. It is likely that additional economic mineral resources will be discovered, and this drives continued exploration activities across the state (e.g., Freeman et al., 2015; Athey et al., 2016). However, the vast size, remoteness, and rugged terrain, coupled with a relatively short summer field season, hamper these efforts. New methods like remote imaging spectroscopy (hyperspectral surveying) could help assess large tracts of land and provide focused target areas for ground-based mineral exploration.

The U.S. Geological Survey (USGS) conducted an airborne imaging spectroscopy study using the HyMap sensor to test this technology for mineral exploration at high latitudes in Alaska. The selected study area extends from the Nabesna Glacier to the Canadian border in Wrangell-St. Elias National Park, eastern Alaska Range (Fig. 1). This area was selected because (1) the region has reconnaissance-level geologic control based on historical USGS mapping (e.g., Richter, 1973; Richter et al., 1975a), (2) much of the uplands contain bedrock exposure for effective spectral mapping, and (3) the area includes several undisturbed porphyry Cu deposits and prospects. Porphyry deposits are typically characterized by broad alteration zones with distinctive mineralogy (e.g., Lowell and Guilbert, 1970; Seedorff et al., 2005; Sillitoe, 2010), including mineral(s) that can be remotely sensed using spectroscopy (e.g., Berger et al., 2003; John et al., 2010; Mars, 2013). These factors, along with reconnaissance ground sampling completed as part of this project, form a framework against which our imaging spectroscopy survey results can be compared and validated.

In this paper, we present mapping results, including spectrally predominant mineral and mineral composition variability maps from the western part of the survey area that contains the largest exposed porphyry Cu-Au-Mo deposits—Orange Hill and Bond Creek. We examine whether geologic units can be distinguished based on mineral predominance mapping and whether the porphyry deposits can be identified. We establish that different chlorite and/or white mica spectral signatures are directly related to mineral chemistry and that these differences are useful for identifying known as well as previously unknown Cu-Au-Mo–mineralized areas. Finally, we discuss utilization of the technology in other portions of the state of Alaska and in other northern latitude areas.

Regional Geology

The geology of the southern margin of Alaska records continental growth. The Wrangellia superterrane (composed of the Peninsular, Alexander, and Wrangellia island arc terranes; Fig. 1B) was accreted to the paleocontinental margin in the Jurassic to Cretaceous (Plafker and Berg, 1994). Sediments deposited into the intervening closing ocean basin are now preserved in a discontinuously exposed belt of flysch basins, including the Nutzotin basin in the general study area (Fig. 1A, B; Berg et al., 1972; Plafker and Berg, 1994; Manuszak et al., 2007; Goldfarb et al., 2013). The Nutzotin rocks onlap the Wrangellia terrane and extend northward to the Denali fault, but their exposures pinch out to the west (Fig. 1B). The Denali fault displays net dextral displacement of ~350 to 400 km presumed to be mostly of Cenozoic age and separates the Mesozoic rocks from Paleozoic and older Yukon Tanana terrane rocks of ancestral North America (Plafker and Berg, 1994). A number of Early Cretaceous plutons intruded deformed Nutzotin flysch and the adjacent leading edge of the exposed Wrangellia terrane presumably after the main contractional event, as evidenced by their lack of deformation. Recent U-Pb age data indicate emplacement between ~126.4 ± 1 and 113 ± 0.5 Ma (Graham et al., 2016), similar to and slightly older than previous K-Ar and Ar-Ar cooling ages (~120–105 Ma [K-Ar, Ar-Ar]; Richter et al., 1975a; Snyder and Hart, 2007). Younger Cenozoic plutonic and volcanic rocks are locally abundant (Richter, 1973). The study area lies just southwest of the Totschunda fault (a splay off the Denali fault) whose Pleistocene movement has disrupted the original configuration (Fig. 1B, C).

Orange Hill-Bond Creek Study Area

The Orange Hill-Bond Creek study area, which encompasses about 300 km2 (~115 mi2) of the original survey area, is located approximately 450 km east-northeast of Anchorage and 100 km south-southwest of Northway, Alaska (Fig. 1A). The rugged terrane varies from approximately 900- to more than 2,750-m elevation above sea level. The western margin of the study area is the Nabesna Glacier valley floor. Orange Hill is a conspicuous iron-stained knob that rises several hundred feet above the eastern side of the valley. It is separated from the higher ridge to the east by California Gulch (Figs. 1C, 2A). To the east and north of the higher ridge are the west, middle, and east forks of Bond Creek, which flow north and west. Vegetation, including trees, scrub brush, and tundra, is present in the glacial valley and at Orange Hill, giving way to sparsely vegetated tundra at elevations above approximately 1,500 m. Outcrop, rubble crop, and talus predominate at higher elevations where lichen cover is minimal.

The geology of the study area features Early Cretaceous rocks of the Nabesna pluton to the north that intrude Wrangellia terrane to the south (Fig. 1C). A number of geologic investigations were completed in this area (e.g., Mendenhall and Schrader, 1903; Moffit and Knopf, 1910; Pilgrim, 1930; Moffit and Wayland, 1943; Gillespie, 1970; Linn, 1973; Richter, 1973). The most comprehensive mapping was completed by Richter (1973), from which the following geologic descriptions of the plutonic and arc rocks are derived. The Nabesna pluton consists predominantly of biotite-hornblende granodiorite with lesser quartz monzonite, biotite quartz diorite, and hornblende diorite. There are sparse exposures of medium- to coarse-grained trondhjemite and light-gray quartz-plagioclase porphyry. To the south of the roughly E striking contact, the Wrangellia terrane rocks include volcanic and volcaniclastic rocks, limestone, siltstone, and the Nikolai Greenstone. The Pennsylvanian and/or Permian volcanic rocks include fragmental volcanic rocks, tuffs, and andesite flows intruded by silicic quartz-eye effusive or shallow intrusive bodies. Overlying the volcanic rocks are Permian limestone and Permian to Triassic calcareous and carbonaceous siltstone and argillite; the carbonaceous argillite has a basal silicic tuff unit and has locally been intruded by dikes, sills, and irregular bodies of Triassic gabbro (as much as 50% of the section). The Nikolai Greenstone comprises amygdaloidal basalt flows with thin interleaved volcaniclastic beds. Amygdules are filled with quartz, calcite, chlorite, epidote, pumpellyite, zeolite minerals, and locally Cu.

Several Cu prospects and deposits have been identified, mostly within the southwestern portion of the Nabesna pluton (Fig. 1C; Mendenhall and Schrader, 1903; Moffit and Knopf, 1910; Pilgrim, 1930; Van Alstine and Black, 1946; Linn, 1973; Richter et al., 1975b, Hudson, 2003). The best described are the Early Cretaceous Orange Hill and Bond Creek porphyry Cu-Mo(Au) deposits. The deposits and prospects are characterized by distinctive limonite staining from surficial weathering of pyritic altered rock (Fig. 2A). Reported alteration zonation at Orange Hill includes a 400- × 2,000-m zone of potassically altered rocks that contain biotite, quartz veinlets, minor K-feldspar, chlorite, and sericite and an outer 1,000- × 3,000-m zone of chlorite and minor sericite (Richter et al., 1975b). Much of this altered rock is not exposed, and zonation cannot be observed at the surface. Stockwork quartz chalcopyrite-pyrite ± molybdenite veins are exposed up to 1,400 m to the northeast (Fig. 2B, C). Other alteration minerals include calcite and kaolinite (Van Alstine and Black, 1946).

Alteration associated with the Bond Creek deposit is evident in both the Nabesna pluton and adjacent Wrangellia terrane volcanic arc rocks (Fig. 2D). Within the deposit, chalcopyrite ± molybdenite-bearing quartz-rich and quartz-poor veins are present (Fig. 2E). A central zone with abundant chlorite, minor biotite and K-feldspar, and spotty sericite is enveloped in a 2,000- × 3,000-m zone with minor chlorite, epidote, and anhydrite (Richter et al., 1975b). Both the Orange Hill and Bond Creek deposits are cut by late anhydrite veins (now gypsum).

Historical non-NI 43-101 compliant estimates at Orange Hill vary significantly from 86 Mt at ~0.3% Cu and 0.015% Mo (Linn, 1973) to 320 Mt of rock 0.35% and 0.02% Mo (Richter et al., 1975b). Richter et al. (1975b) estimated Bond Creek to contain 500 Mt of rock with average grade of 0.3% Cu and 0.02% Mo at Bond Creek. Both deposits contain locally elevated Au concentrations. Other prospects in the area include porphyry prospects (e.g., Nike, Neil), skarn (Copper King), and polymetallic vein occurrences (Fig. 1C); a younger, ca. 22 Ma unnamed Mo occurrence lies northeast of Neil (Fig. 1C; Silberman et al., 1977). Limited descriptions of these systems are available (Richter et al., 1975b; Hudson, 2003). Follow-up investigations based on our spectral results led to recognition of additional mineralized areas hosted in the Nabesna pluton, including at the headwaters of the east fork of Bond Creek (Fig. 2F). Here, pyrite ± chalcopyrite and molybdenite in veins and fracture coatings in glacial debris were observed, sometimes with K-feldspar alteration halos (Fig. 2G, H); these areas are discussed in the text.


Our study integrated airborne-, field-, and laboratory-based spectroscopic data with field- and laboratory-based geologic studies. Airborne imaging spectrometer data in the Nabesna field area were collected over two days (July 14 and 21, 2014) and coincided with the first of three approximately one-week-long field work sessions, completed during the 2014 to 2016 summer field seasons. Field spectroscopy was geared toward collecting calibration site data for atmospheric correction of airborne survey data and verifying and validating airborne results. Geologic samples were collected at and around mineral occurrences and used for laboratory-based imaging spectroscopy and geochemical characterization to establish the distribution of mineralized rock and improve sample density within historical sediment geochemistry datasets. Chemistry from targeted sediment and soil sampling in particular provides important information for interpreting possible relationships between imaging spectrometer data and mineralized areas (both previously known and newly recognized) discussed in this paper. The following sections summarize salient aspects of methods. More detailed methods descriptions and complete spectral and geochemical results can be found in the Digital Appendix and in USGS data releases (Kokaly et al., 2017a, 2018; Graham et al., 2017; Hoefen et al., 2018).

Spectral data collection and processing

Approximately 1,000 line km of imagery were collected using a HyMap sensor (Cocks et al., 1998) mounted on a modified Piper Navajo aircraft. The aircraft was flown at an altitude of approximately 5,050 m, resulting in average ground spatial resolution of approximately 6 m. HyMap measured reflected sunlight in 126 narrow channels that cover the wavelengths of 455 to 2,483 nm. The full set of airborne data, extending beyond the focus area of this study, is available in Kokaly et al. (2017a).

Ground-based spectral collections were completed using an Analytical Spectral Devices FieldSpec® 4 standard resolution (ASD FS4) field spectrometer. The ASD FS4 measures 2,151 channels that span the 350- to 2,500-nm reflected solar range using three detectors. Reflectance data were collected from four calibration sites (with coverage of 86–210 HyMap pixels) in broad alluvial-fluvial gravel bars that were minimally vegetated and mostly lichen free; these calibration sites were used to calibrate flight lines (following procedures in Clark et al., 2002; Kokaly et al., 2013; and Kokaly and Skidmore, 2015). Although the rock types in these areas were mixed and varied at the fine-spatial scale, at the HyMap 6-m pixel scale the calibration areas were spectrally homogeneous.

Corescan’s Hyperspectral Core Imager Mark III™ imaging spectrometer (www.corescan.com.au/services/the-corescan-system; hereafter referred to as Corescan imaging spectrometer) was used to scan approximately 30 hand specimens from the Orange Hill and Bond Creek porphyry Cu deposits during the fall of 2015 in order to map the spectrally predominant minerals and wavelength positions for white mica. Spatial resolution was approximately 0.5 mm. The Corescan imaging spectrometer measures 514 channels that span the 450- to 2,500-nm wavelength range.

The wavelength and bandpass characteristics of each spectrometer used in this study were evaluated using a set of reference materials in order to check and cross calibrate data from different instruments. HyMap and ASD evaluations agreed with manufacturer-reported values for channel wavelength position and bandpass. Based on our analysis of measurements of the reference materials, the reported channel positions of the Corescan imaging spectrometer were adjusted, notably by shifting them ~1.4 nm to longer wavelength for channels in the 2,200-nm region (for additional details on the wavelength evaluation and adjustment see Hoefen et al., 2018).

Reflectance conversion and predominant mineral classification: The HyMap data were converted from radiance to reflectance using a multistep calibration process adapted from the procedures in Kokaly et al. (2013). Reflectance images from HyMap and Corescan were processed using the Material Identification and Characterization Algorithm (MICA), a module of the USGS PRISM (Processing Routines in IDL for Spectroscopic Measurements) software (Kokaly, 2011). PRISM is a freely distributed software (https://speclab.cr.usgs.gov/spectral-lib.html#software) programmed in Interactive Data Language (IDL; Harris Geospatial Solutions, Broomfield, Colorado). The MICA analysis identifies the spectrally dominant mineral(s) in each pixel of imaging spectrometer data by comparing continuum-removed spectral features in its reflectance spectrum to continuum-removed absorption features in reference spectra of minerals, vegetation, water, and other materials. Continuum removal is a technique to isolate an absorption feature from background spectral variations (Clark and Roush, 1984). The reference spectra used in the MICA command file are available to the public in the USGS spectral library (Kokaly et al., 2017b). The MICA command file is provided in the Digital Appendix. Output mineral classes from the MICA analysis were combined into the grouped classes depicted in mineral predominance maps (see Table A1). Representative spectra from the ground-calibrated HyMap reflectance data for the major classes discussed in this study are included in Figure A1 of the Digital Appendix.

The selection of reference spectra in the MICA command file was guided by previously published applications of hyperspectral data for mineral mapping (notably, Clark et al., 2003) and criteria relevant to this study, including the following: (1) likelihood a mineral would be present in great enough abundance to be detected at the pixel size of the imagery; (2) reliability in identifying the spectral features of a mineral using HyMap, given its wavelength range and spectral resolution; and (3) importance of a mineral to lithology and known mineral resources for the areas covered by the 2014 HyMap survey. In analyzing the large-area hyperspectral coverage of Afghanistan, Kokaly et al. (2013) established a set of mineral spectra relevant to the first two criteria. To that MICA command file, we added more reference spectra of chlorites (clinochlore and thuringite), clays (nontronite and hectorite), and topaz to address the third criterion. Finally, additional linear spectral mixtures of various combinations of white mica, clays, chlorites, and carbonates were computed and added to the MICA command file (see Kokaly et al., 2018, for details on the adaptation of the Afghanistan MICA command file).

Mineral composition mapping: In addition to mineral mapping, subtle changes in SWIR absorption feature positions can provide evidence of compositional variations within mineral species (Post and Noble, 1993; Swayze, 1997). Examples of previous studies include spectral mapping of white mica Al composition and/or chlorite of varying compositions to characterize metamorphic history (e.g., Duke, 1994) and geology and zoning about mineral deposits (e.g., Herrmann et al., 2001; van Ruitenbeek et al., 2012; Harraden et al., 2013; Laakso et al., 2015). In our study, we identified pixels with that had highest MICA fit to muscovite or illite in our HyMap and Corescan imaging spectrometer data and calculated the 2,200-nm Al-OH absorption feature for each of these pixels. For our computations, we fit a parabola to three channels within the white mica 2,200-nm absorption feature (the channel with a minimum in continuum-removed reflectance and one channel on either side). The wavelength value of the axis of symmetry from the fitted quadratic function was used to model the central wavelength position of the feature. The images of continuous values of wavelength position were converted to classification images with classes in 1-nm increments. Representative HyMap spectra for various white mica wavelength classes are shown in Figure A1 of the Digital Appendix. Four billets from three rock samples imaged with the Corescan imaging spectrometer were scanned to determine the wavelength positions of white mica across the billets.

These data were used to select locations for electron probe microanalysis (EPMA) of white mica chemistry in the respective thin sections, as described below.

Sediment/soil/rock chemistry and mineralogy

Sediment data from the Alaska Geochemical Database (Granitto et al., 2011) were used to establish the regional geochemical framework. We collected an additional 96 rock and 66 soil and sediment samples for geochemical analysis. The fine-sediment fraction of soils and sediments (–80 mesh) and the rock samples were pulverized to –200 mesh (<0.074 mm) and analyzed at SGS Laboratories for concentrations of 55 major, rare earth, and trace elements by inductively coupled plasma-atomic emission spectrometry-mass spectrometry (ICP-AES-MS) after sodium peroxide fusion. Data were considered acceptable if recovery for all 55 elements was ±15% at five times the lower reporting limit and the calculated relative standard deviation (RSD) of duplicate sample analysis was no greater than 15%. Gold was determined by fire assay. Complete results can be found in Graham et al. (2017).

Select soil samples and four rock samples were also analyzed by X-ray diffraction (XRD) to validate the presence of spectrally identifiable minerals identified in mineral predominance mapping from MICA analysis. The XRD scans were collected on a PANalytical X’Pert Pro MPD diffractometer with Bragg Bertano optics using Cu radiation after following sample preparation methods of Moore and Reynolds (1997) (App.; Table A2).

White mica chemistry

Electron probe microanalysis (EPMA) was completed on white mica (n = ~159 spots) and lesser chlorite and kaolinite from a total of 14 areas on four thin sections from three hand specimens from the Orange Hill deposit area. White mica within these billets spanned the approximate range of white mica 2,200-nm wavelength positions observed in our airborne and Corescan data. The JEOL JXA-8900 Superprobe at the USGS Central Mineral and Environmental Resources Science Center, Denver, Colorado, USA, is outfitted with five wavelength dispersive spectrometers. The microanalyzer was operated at 15 kV and 20 nA, with a beam diameter of less than 1 µm. Calibration was checked using well-characterized silicate and oxide standards. The full complement of EPMA results is provided in Graham et al. (2017).


Airborne spectroscopy

Mineral predominance mapping: Approximately 51% of the Orange Hill and Bond Creek study area produced interpretable mineral-related spectral signatures (Fig. 3). Vegetated ground, shown in tan shades, accounts for an additional 21% of the area. Vegetation occurs mostly below 1,500 m and covers much of the Orange Hill deposit. Glacial ice and snow and wet ground are mapped at higher elevations to the north and east in the glacier field (22% of classified pixels). A small percentage of the area is unclassified (5%, shown as black pixels), mainly due to steep terrane or clouds/cloud shadows that result in poor illumination or ice and meltwater from glaciers that cause interference in spectral signatures.

The mineral predominance map indicates complex lateral variations in spectral signatures in the study area (Fig. 3). In general, the Nabesna pluton east of the middle fork of Bond Creek and north of the east fork of Bond Creek is dominated by white mica signatures (shown in orange). In the porphyry deposit and prospect area to the south of the east fork of Bond Creek, the spectral patterns are more variable. Exposures at Orange Hill map as white mica, with clinochlore + white mica more predominant in the southwest portion of the hill and abundant montmorillonite (smectite) and gypsum to the northwest (Figs. 3, 4). Localized kaolinite-bearing white mica signatures characterize the upper reaches of California Gulch. Zones of white mica, clinochlore + white mica, and montmorillonite (smectite) signatures are also indicated in the airborne data over variably iron-stained igneous rocks that crop out in E-W–trending drainages to the northeast. The mostly plutonic rocks between Orange Hill and the west fork of Bond Creek have abundant carbonate, chlorite/epidote, and kaolinite signatures. The altered zone at Bond Creek, as mapped by Richter (1973) (loc. 2, Fig. 3; central area, Fig. 5), is dominated by white mica signatures in both plutonic (labeled Kg) and volcanic rocks (labeled PlPv) to the south. Outside of this zone, more abundant clinochlore, chlorite/epidote, and carbonate classes are present relative to white mica. Jarosite + white mica and montmorillonite (smectite) map on the northeastern flank of the alteration zone. The XRD analyses of soils from the Orange Hill and Bond Creek areas yield variable proportions of these same minerals (see Table A2; Graham et al., 2017).

Mapping of the Wrangellia terrane delineates areas of white mica, chlorite + white mica, clays, and zones of chlorite/epidote generally similar to the porphyry deposit/prospect area (Fig. 3). The MICA analysis indicates the Nikolai Greenstone (Trn) signatures are dominated by chlorite/epidote and carbonate (albeit not a pattern restricted to that rock unit). Several small areas had spectra that best match serpentine group reference standards. XRD analyses of four rock samples collected to investigate this spectral signature establish that the rocks actually contain amphibole and chlorite (pale green; Fig. 3). The initial classification in HyMap data was the result of serpentine and amphibole both having spectral features near 2,320 and 2,390 nm (Kokaly et al., 2017b), but only serpentines are included as reference spectra in the MICA analysis.

For other units, thin limestone-bearing units within the volcanic arc rocks (unit Pl) map predominantly as calcite. In addition, some pixels match the reference spectrum of calcite + dolomite (or a mixture of calcite and epidote or chlorite). Systematic variations in predominant mineral signatures suggest exposure of different laterally continuous lithologic units in the volcanic units (PIPV) along Nikonda Creek with clay, white mica, chlorite + white mica, and local pyrophyllite signatures. The XRD results verify pyrophyllite in soils from areas mapped with pyrophyllite signatures (Table A2; Graham et al., 2017). Systematic spectral trends are not obvious to the east.

Chlorite classes: The predominant mineral classes containing chlorite, epidote, and clinochlore occur in distinct regions of the study area with little overlap (Fig. 6). Most clinochlore pixels are intermixed with abundant chlorite/epidote pixels (bright green and dark green, respectively) in the Nikolai Greenstone southeast of Orange Hill, and pixels with chlorite + white mica (purple) signals coincide with volcanic arc rocks. Clinochlore + white mica signatures dominate in the vicinity of the porphyry cluster (Fig. 6). Outside of the main cluster, moderate abundances of the clinochlore + white mica pixels are mapped in several areas, including on the north side of and at the headwaters of the east fork of Bond Creek (locs. 7, 8; Fig. 6) and in a zone mostly along the edges and terminus of a glacier that cut volcanic rocks at the southern end of our survey area (loc. 9). The latter zone is at the toe of a glacier and could be reflecting materials transported from the south, where an undifferentiated mafic intrusion interpreted as a probable marginal phase of the Nabesna pluton is exposed (Richter, 1973).

White mica wavelength position mapping:Figure 7 shows the variation in wavelength position for the 2,200-nm Al-OH absorption feature measured in white mica-dominated pixels (see Fig. 3). Calculated absorption feature positions are grouped into 12 classes ranging from <2,196 to >2,207 nm (grading from purple to red, respectively). The majority of pixels contain shorter-wavelength positions less than 2,202 nm (blue and purple), irrespective of the geologic unit analyzed. However, white micas at longer wavelengths (green, yellow, and red) are highly concentrated over a broad area in the Nabesna pluton, extending from the Orange Hill area, through Bond Creek to east of the middle fork of Bond Creek and north to the eastern fork of Bond Creek (area 2; Fig. 7). This signature incorporates the Nike and Neil porphyry prospects (locs. 3, 4) and the vicinity of the unnamed porphyry Mo prospect (loc. 5; Fig. 7). At the Bond Creek deposit, the feature extends into the Wrangellia terrane volcanic rocks. Rock exposures at Orange Hill and the northeastern drainages (loc.1, 1a) yield mixed short to long Al-OH absorption feature positions. Some intermediate- to long-wavelength signatures (classes depicted with green, yellow, and red) map in the vicinity of the Copper King skarn (loc. 6).

Two notable satellite areas with longer-wavelength white mica occur north of the east fork of Bond Creek and at the headwaters of the east fork (locs. 7, 8; Fig. 7), the same areas dominated by clinochlore + white mica pixels as described above. A narrow but prominent zone of longer-wavelength Al-OH absorption features (loc. 10) is present in a small drainage southwest of location 7. In the volcanic rocks south of the contact with the Nabesna pluton, localized intermediate to long wavelengths are observed, but they primarily occur in small isolated areas. The most extensive of these areas is on the southern edge of our survey on the west fork of Bond Creek (loc. 9) and also coincides with clinochlore + white mica signatures (Fig. 6). These spectrally distinct areas in Wrangellia terrane rocks (loc. 9 and isolated areas) are not definitively associated with mineralized rock.

Corescan images and mineral chemistry

The white mica wavelength positions calculated from Corescan imaging spectrometer data collected on hand specimens and billets of altered igneous rocks from Orange Hill span a range similar to those of the HyMap results. They also demonstrate spectral complexity at the hand specimen scale. Even within macroscopically uniform rock samples containing primarily white mica, spectral variations can be abrupt (e.g., rock sample 14HYPORH013A in Fig. 8). In other cases, they can be gradational at the millimeter scale (e.g., the 15HYPORH002 billet in Fig. 8). Such variation, likely related to overprinting alteration/mineralization events and weathering, provides the opportunity to examine white mica wavelength position in relation to chemical changes.

We categorized the chemistry of white micas in the four sections into three discrete categories: (1) longest-wavelength positions (≥2,207 nm), (2) shortest-wavelength signatures (≤2,202 nm), and (3) two categories of mixed short- to long-wavelength signatures (2,203–2,204 nm). Calculated unit formulae for white mica with short- and long-wavelength signatures have >6 silicon atoms and insufficient K or Al in their unit cell to be strictly considered muscovite (Deer et al., 1992). The octahedral site for both end members exceeds 4 atoms per unit cell, possibly but not necessarily due to some component of mixed-layer clays (Gaudette et al., 1964), as suggested by R1 ordered illite/smectite clays in many of the XRD analyses (see Table A2).

Chemical compositional differences of white micas with short- vs. long-wavelength positions in the thin sections confirm that the wavelength position is at least in part related to mineral chemistry (representative analyses shown in Table 1). White micas from regions of billets displaying the longest-wavelength Al-OH absorption features have higher median Si, Mg, and Mn and lower median Al (± Na) than those with the shortest-wavelength positions (Table 1; Fig. 9A, B). When abundance (as atoms per formula unit [apfu]) of total Fe + Mn + Mg + Si is compared to total Al, there is a systematic inverse relationship attributed to Tschermak substitution (AlVI + AlIV↔[Fe, Mg, Mn]VI + SiIV) with a greater phengitic component (and lower Al) in micas with longer-wavelength Al-OH features (Fig. 9C).

Sediment and soil chemistry: A large number of soil and stream sediment samples collected proximal to the Orange Hill and Bond Creek porphyry systems, including historical samples and those collected as part of the current study, contain anomalous concentrations of Cu, Mo, and/or Au (15.5× median value of soils in the conterminous U.S.; Smith et al., 2013). Nearly all samples (n = 25) at Orange Hill and the two drainages to the northeast contain anomalous Cu and Mo; of these, eight samples contain ≥60 ppb Au (areas 1, 1a, Fig. 10A, B). Most samples from both the west and middle forks of Bond Creek immediately upstream and downstream of the Bond Creek deposit (loc. 2, Fig. 10A) also contain more than 15 times the median value for Cu, and they contain concomitant highly anomalous Mo and/or Au concentrations (Fig. 10B).

North and east of the area of known mineral occurrences/deposits, samples are also anomalous in Cu, Mo, ± Au. At location 7 (Fig. 10A, B), soil samples from alluvial fans and one stream sediment sample contain anomalous Cu (3 samples at 269–300 ppm), two contain 15 to 26 ppm Mo, and one contains 30 ppb Au. Three samples collected at the upper reaches of the east fork of Bond contain highly anomalous Cu (235–434 ppm), Au (80–510 ppb), and Mo (8–10 ppm) (loc. 8, Fig. 10A, B). Three samples from drainages northeast of the headwall to east Bond Creek that flow to the northeast contain anomalous Au (50–260 ppb) but are not anomalous in Cu or Mo. Finally, sediment from a narrow drainage on the south side of the middle fork of Bond Creek (loc. 10) contains highly anomalous Mo (18 ppm) and Au (150 ppb) and elevated Cu (220 ppm).

Multielement anomalies are largely absent in the Wrangellia terrane rocks and Nabesna pluton outside of the areas discussed above. Although sporadically distributed samples with highly anomalous concentrations of Cu (>7.5× background) are common, anomalous Mo and/or Au concentrations are only found immediately adjacent to the Bond Creek deposit (loc. 2), downstream of the Copper King skarn (loc. 6), and locations 11 and 12 (Fig. 10B).

Altered and mineralized rock samples confirm local bedrock sources of anomalous metals. Rocks from within and in the vicinity of the Orange Hill and Bond Creek deposits contain high Cu, Mo, and Au concentrations (median concentrations of 500 ppm, 18 ppm, and 13 ppb, respectively), consistent with chalcopyrite and molybdenite observed in hand specimens and with previous studies (Van Alstine and Black, 1946; Linn, 1973). Mineralized samples were also collected from the drainages northeast of Orange Hill and the headwaters of the east Bond Creek drainage. A mineralized grab sample from an alluvial fan below location 7 contained 8,400 ppm Cu, 3,780 ppm Mo, and 143 ppb Au. The three highest-grade rock samples collected from glacial debris at the headwaters of east Bond Creek (loc. 8, Fig. 10) contain 800 to 3,900 ppm Cu, <2 to 51 ppm Mo, and 36 to 200 ppb Au. A grab sample of quartz-sulfide–veined volcanics from the southern end of our survey yielded a polymetallic signature including 740 ppb Au, 1,120 ppm Cu, 11 ppm Mo, 9,440 ppm Pb, and 21,000 ppm Zn (loc. 9).


Regional mineral predominance maps as a supplement to geologic mapping

Mineral predominance mapping as a proxy for geologic mapping in the Wrangellia terrane has yielded mixed results. On the northeast side of Nikonda Creek (Fig. 3), distinct SE-trending linear strips of different mineral spectral signatures (from carbonate to white mica to chlorite + white mica) delineate laterally continuous stratigraphic units with mineralogical variation. However, elsewhere in the mapped area, mineral predominance does not show systematic variation. The lack of patterns is likely a consequence of substantial surface material movement across mapped boundaries and in part due to shared spectrally identifiable minerals in multiple geologic units. For example, whereas calcite is a major component of the Pl unit, which includes limestone, calcareous siltstone, and limy shale, it is also an important alteration mineral (along with chlorite and epidote) in the Nikolai Greenstone. Consequently, parts of both units map as calcite (Fig. 3). The widespread andesitic and dacitic volcanic and volcaniclastic units of the Tetelna volcanics are distinct in their lack of carbonate and abundant white mica and locally clay signatures. But in general, there is no consistent pattern to discern discrete subunits, and patches of chlorite/epidote that are present in the volcanics are indistinguishable from the Nikolai Greenstone. Geologic mapping is not detailed enough to explain the non-systematic spectral mapping, but the patterns could reflect lateral facies changes, structural complexity, mass wasting, and overprinting by hydrothermal activity.

In the Nabesna pluton, the widespread spectral predominance of white mica with lesser calcite and clay throughout the northern and eastern parts of the pluton suggests a pervasive background signature from deuteric alteration or weathering (Fig. 3). The abundant chlorite + white mica, chlorite/epidote, and clay minerals (sometimes reflecting late dikes) in the vicinity of the porphyry cluster appear similar to those of the Wrangellia terrane rocks, and without independent geologic control, those rocks could be misinterpreted to be part of Wrangellia. Overall, the spectral results do not directly correspond to systematic differences within individual geologic units at the regional scale, precluding geologic mapping based only on mineral predominance. However, the variations between spectral data and geologic mapping provide important indications of secondary (potentially hydrothermal) influences. Utilization of other analytical methods, for example, supervised classification using end-member spectra for different units, has proven useful for lithologic mapping elsewhere at northern latitudes (e.g., Harris et al., 2005; Feng et al., 2018).

Deposit-scale mapping: Bond Creek and Orange Hill

At the deposit scale, distributions of spectrally dominant mineral(s) reflect both primary geology and magmatic-hydrothermal overprinting. In the well-exposed Bond Creek deposit area, a >1.5- X 2-km zone dominated by white mica and lesser chlorite + white mica signatures closely corresponds to the altered zone mapped by Richter (1973) (dashed zone; Fig. 5). The zone grades outward to the north and south to clinochlore + white mica and then clinochlore + white mica with irregular chlorite/epidote and carbonate signatures, consisting of a possible phyllic core flanked by chlorite + white mica to propylitic zones. This pattern is consistent with a fairly deeply eroded idealized porphyry system (e.g., Lowell and Guilbert, 1970; Sillitoe, 2010). Similar mineral assemblage zonation has been reported in multispectral studies (e.g., John et al., 2010). The predominance of jarosite along with montmorillonite (smectite) on the steep eastern side of the Bond Creek ridge is conspicuous, with jarosite likely an important indicator of weathering of sulfide-bearing rock.

Large-scale zoning patterns are not obvious within the Orange Hill deposit owing to extensive vegetative cover (Fig. 4). However, Linn (1973) identified at least three phases of the Nabesna pluton at Orange Hill: a diorite body in the southwestern part of the hill in contact with the locally predominant tonalite on the northern and eastern parts and quartz-feldspar porphyry exposed in the headwaters of California Gulch (Fig. 4). Spectrally, clinochlore + white mica signatures are widespread in the diorite (and some adjacent tonalite), montmorillonite (smectite) and gypsum occur in the more decomposed tonalite on the northwest side of the ridge, and kaolinite and white mica signatures predominate at and around quartz-feldspar porphyry in the headwaters of California Gulch. Kaolinite + white mica signatures to the east by the Copper King skarn coincide with additional quartz-feldspar porphyry (Fig. 4). Therefore, as a first order, the spectral signatures at Orange Hill empirically reflect alteration products associated with different rock units.

The different spectral signatures in tonalite and diorite appear to relate to intensity of alteration as well as initial composition. Both smectite and chlorite, common alteration minerals in porphyry Cu deposits, can form by alteration of ferromagnesian minerals, calcic feldspars, and volcanic glasses (Ross and Hendricks, 1945). Montmorillonite (smectite) signatures coincide with broader aprons of friable tonalite on the northwest side of Orange Hill, suggestive of more intense alteration than the clinochlore- and white mica-bearing diorite to the southwest. Whereas mostly near vertical gypsum veins occur in both tonalite and diorite, the more friable character of the tonalite results in dispersal of gypsum across the ground surface. The quartz-feldspar porphyry in the head of California Gulch is compositionally different and lacks mafic minerals. Extensive quartz veining in and around the porphyry indicates widespread hydrothermal fluid circulation. Alteration and weathering of feldspar and mica within this unit and the rocks it intruded produced secondary minerals with kaolinite rather than smectite.

Both clinochlore + white mica and kaolinite + white mica zones are also mapped in the mineralized drainages to the northeast of Orange Hill (Fig. 4) and similarly reflect variably altered and mineralized dioritic to granodioritic rocks. Importantly, the abrupt change in spectral signatures in the drainages aligns with and thus locally defines the Bryner fault (Fig. 4). The predominance of chlorite/epidote and carbonate signatures in plutonic rocks farther uphill to the east suggests distal propylitic alteration associated with the Orange Hill porphyry system.

Chlorite and white mica signatures and their relationship to mineral occurrences

Coherent patterns of the three main chlorite-bearing mineral predominance classes (clinochlore, clinochlore + white mica, and chlorite + white mica) strongly support fundamental differences in chlorite composition or relative proportions of chlorite and white mica related to lithology or magmatic-hydrothermal processes. The distribution of clinochlore pixels in Nikolai Greenstone and widespread chlorite + white mica pixels in the volcanic and volcaniclastic rocks of the Wrangelia terrane suggest lithologic control (Fig. 6). In contrast, the distribution of the clinochlore + white mica class is inferred to be related to magmatic-hydrothermal processes based on (1) the close spatial association to the porphyry cluster in the Nabesna pluton, (2) the fact that it extends across lithologic boundaries at the Bond Creek deposit, and (3) the near absence of this (or any chlorite-dominant) class elsewhere in the Nabesna pluton. The similar but more broadly distributed intermediate- to long-wavelength white mica (compare Figs. 6, 7) and the linkage of long-wavelength signatures to phengitic chemistry in our microprobe analyses (Table 1) strongly suggest the phengitic white mica is also of magmatic-hydrothermal origin.

The spatial association of the clinochlore + white mica and longer-wavelength white mica classes to multielement Cu-Mo-Au anomalies further supports a causative relationship with hydrothermal activity (Fig. 11). At the Bond Creek deposit, anomalous concentrations of Cu, Mo, and Au occur in the upstream portion of the west and middle forks of Bond Creek in the immediate vicinity of longer-wavelength white mica. Where this zone extends to the east across the middle fork of Bond Creek, metal concentrations remain high (e.g., loc. 4b, Fig. 11). The mixed spectral signatures at Orange Hill and the gullies to the northeast have highly anomalous metal concentrations.

Anomalous Cu, Mo, and/or Au coincide with clinochlore + white mica and long-wavelength absorption feature signatures in several zones outside of the main porphyry Cu cluster (locs. 7–10; Fig. 11). Elevated metal concentrations in some rock samples from these locations with up to 8,000 ppm Cu, >50 ppm Mo, and >500 ppb Au indicate additional Cu-Mo-Au mineralized areas that were previously unknown or unreported.

The absence of significant clinochlore + white mica and long-wavelength white mica signatures in most of the Wrangellia terrane rocks coincides with a general lack of anomalous multielement Cu-Mo ± Au concentrations. Only sporadic Cu anomalies are present. Most of the occurrences described in the Wrangellia terrane are poorly documented and described as small polymetallic vein occurrences (e.g., locs. 11, 12; Fig. 11) and occur in areas without rock exposure (e.g., loc. 12). The abundance of mafic rock (inherently higher Cu concentrations) and weathering of the Nikolai Greenstone (Richter, 1973) could account for sporadic high-Cu concentrations in stream sediment samples.

Absorption feature shifts in porphyry deposits

Although airborne imaging spectroscopy data have been used for porphyry Cu exploration, there are few detailed interpretations of these data in the literature (e.g., Cudahy et al., 2001; Berger et al., 2003, Coulter et al., 2007); however, there are several studies at the hand-specimen scale (e.g., Alva-Jimenez, 2011; Cohen, 2011; Cohen et al., 2011). Halley et al. (2015) described a model of spectral variations related to changes in white mica chemistry associated with changing temperature and pH across an idealized porphyry deposit. In the core of the deposit, where low-pH conditions result from sulfur dissociation from rising, cooling, oxidized, sulfate-rich fluids, formation of white micas with near-end-member muscovite composition are favored. Below dissociation depths and distal from the main magmatic fluid flux, where fluids are near neutral pH, phengitic compositions (increased Fe, Mg, and Si and lower Al) are favored. Spectroscopically, the phengitic compositions have longer-wavelength 2,200-nm Al-OH absorption feature positions.

Unlike the above idealized simple system model, many porphyry deposits form by multiple overprinting episodes of alteration (e.g., Seedorff et al., 2005; Sillitoe, 2010). Spectral investigations on drill core from the Cretaceous Pebble porphyry Cu-Au-Mo deposit in southwestern Alaska established that alteration zones within the deposit can be identified by classifying the wavelength position of the 2,200-nm feature of the contained micas (Harraden et al., 2013). In agreement with the model by Halley et al. (2015), short Al-OH absorption features of white mica (2,190–2,201 nm) and pyrophyllite (2,160–2,170 nm) predominate in the high-grade Cu-Au zone associated with low-pH advanced argillic alteration in the eastern part of the deposit. However, longer-wavelength illite-rich zones (with chemistry and spectral signatures similar to our phengitic white micas [≥2,205–2,210 nm; Table 1]) that locally overprint potassic alteration (Harraden et al., 2013, fig. 7A) also contain significant metals. Overprinting illite has also been recognized at some other porphyry deposits (e.g., Bingham, Utah; Parry et al., 2002; Red Chris Cu-Au porphyry deposit, B.C., Canada; Norris, 2012). Although wavelength positions are not reported, the illite (± kaolinite) zones can be quite extensive (up to 600–800 m thick and lateral extents of at least 500 × 500 m at Red Chris; Norris, 2012). In each case, the presence of illite (and the longer-wavelength Al-OH absorption feature position that was identified at Pebble) is interpreted to record incursion of lower-temperature, near-neutral pH waters either during intermineral (e.g., Pebble; Gregory et al., 2013) or waning stages of magmatic-hydrothermal activity (Red Chris; Norris, 2012).

The widespread longer-wavelength absorption feature positions at the Orange Hill and Bond Creek deposits are consistent with either a deep-level or distal fringe exposure of the deposit (below main mineralized shell) in the Halley et al. (2015) model. Alternatively, by analogy with reported studies at Pebble, Bingham, and Red Chris, and based on the mica chemistry and spectral signatures at Bond Creek and Orange Hill, relatively low temperature fluids may have overprinted earlier alteration minerals. The additional clinochlore + white mica signature and high Cu, Mo, and/or Au in sediment samples draining areas with long-wavelength features support this latter interpretation. Regardless, the phengitic compositions of the micas associated with porphyry formation are on the whole sufficiently compositionally different to produce the longer-wavelength 2,200-nm absorption features that are distinct from those found in plutonic and volcanic arc rocks not affected by the magmatic-hydrothermal fluids. As has been observed in this and other studies, white mica wavelength position maps from imaging spectrometers can be used to better understand hydrothermal systems and potentially serve as a vector to mineralization in some systems (Bierwirth et al., 2002; Laukamp et al, 2011; Yang et al., 2011; van Ruitenbeek et al., 2012; Swayze et al., 2014; Guo et al., 2017).

Exploration implications

Satellite remote sensing techniques have been used for several decades to map alteration mineral patterns and identify exploration targets, commonly at 15- to 90-m pixel size (Goetz and Rowan, 1981; van der Meer et al., 2012). High spectral resolution imaging spectroscopy largely carried out on airborne platforms has been used commercially for approximately 20 years (Cocks et al., 1998; Coulter et al., 2007). The power of these high-resolution imaging spectrometers lies in the dense spectral sampling, which permits the identification of subtle changes in spectral signatures related to mineral chemistry, as observed in this study.

There have been limited published airborne imaging spectroscopy studies in the Canadian arctic (e.g., Rogge et al., 2014; Laakso et al., 2015, 2016), and to our knowledge, there are no published accounts of the application to geologic or mineral deposit studies in Alaska. Difficulties that have likely prevented such studies in the past include the requirement of clear weather during collection and the need for exposed bedrock. Furthermore, whereas lithologic determinations can be made on rocks with significant lichen cover (~80% by surface area; Rogge et al., 2014), mineral compositions can be masked (e.g., Laakso et al., 2015). However, our study demonstrates that there are areas in Alaska where this technology could provide direct information for mineral exploration. Despite rugged, poorly illuminated terrane, the survey data were sufficient to establish distributions of spectrally dominant minerals and to identify subtle changes in mineral chemistry that characterize known porphyry deposits. These signatures can be broad in areas with extensive exposure, as seen in Bond Creek, but can also be recognized in more poorly exposed areas (e.g., Orange Hill). Importantly, focused geochemical sampling to investigate areas with similar chlorite and white mica spectral signatures resulted in the identification of two previously undocumented Cu (Au-Mo) occurrences north of the main porphyry deposit cluster (locs. 7, 8; Fig. 11).

Numerous investigations of other deposit types such as volcanogenic massive sulfide (VMS; Jones et al., 2005; van Ruitenbeek et al., 2012; Laakso et al., 2015), orogenic Au (Huntington et al., 2006; Arne et al., 2016), epithermal (Bierwirth et al., 2002; Bedini et al., 2009), and iron oxide Cu-Au deposits (Laukamp et al., 2011) from around the world indicate large alteration halos associated with mineral chemistry variations. These chemistry variations can be identified by imaging spectroscopy in drill core, outcrop, or airborne studies. The geologic and tectonic settings of mountainous regions in Alaska with significant exposure are permissive of (and in some cases known to host) a number of these other deposit types (e.g., VMS and orogenic- and intrusion-related Au systems). Design of appropriate resolution surveys in these regions could provide invaluable targeting information in these remote areas. Follow-up sampling and investigations can then be focused into specific areas, greatly reducing subsequent exploration costs.

Our study focused on analysis of spectral features in areas of exposed rock and sediment with clear mineral absorption features. Because significant portions of the Orange Hill deposit are covered by vegetation, follow-on studies should examine vegetation and soils within mineralized areas in comparison to nonmineralized areas. Variations in soil nutrient and cation content have the potential to lead to variations in vegetation cover, species composition, and biochemical concentrations (pigments, nitrogen, lignin, cellulose), resulting in contrasting spectral signatures that could be used to more completely define boundaries of mineralized zones. Because much of Alaska is vegetated, identification of geobotanical signatures over mineralized rock would be important to utilizing remotely sensed imaging spectroscopy over more areas. Although it may be difficult to establish a robust signature, current remote sensing technology offers spectral, spatial, and temporal sampling advantages over past applications of remote sensing to this problem. Advantages of current remote sensing technologies include airborne imaging and field-deployable imaging spectrometers and multispectral sensors on multiple satellite platforms that offer high spatial resolution (for example WorldView3 with 30-cm pixels), long-term time series (30 yr of data from Landsat series satellites), and the possibility for repeat data collection within the short Alaska growing season by many satellite sensors (e.g., Landsat 8, Sentinel-2). Planned satellite missions such as HyspIRI (Lee et al., 2015) and EnMAP (Guanter et al., 2016) have the potential to provide imaging spectrometer data at 30-m pixel size.


This study demonstrates that remotely sensed spectral data can be used to identify porphyry Cu-related alteration and mineralized rock in a remote part of Alaska. Zonation of mapped minerals and mineral groups such as white mica, clinochlore + white mica, and chlorite/epidote are consistent with porphyry models and other studies of porphyry deposits. Several key conclusions can be drawn from this work:

  1. The distributions of white mica, chlorite/epidote, carbonates, and clays utilizing solely the spectrally dominant mineral map derived from imaging spectrometer data only crudely correspond to mapped rock types. However, when combined with reconnaissance mapping, spectral data can help identify areas of potential (magmatic-) hydrothermal alteration.

  2. Two spectrally distinct chlorite-white mica classes (clinochlore + white mica vs. chlorite + white mica) predominate across the study area. The spatial association of the clinochlore + white mica class with porphyry mineralization in rocks from both terranes supports a relationship with hydrothermal processes.

  3. Longer-wavelength white mica 2,200-nm absorption feature positions coincide with the clinochlore + white mica assemblage in the Nabesna pluton and altered volcanic rocks at Bond Creek. The longer-wavelength signatures (≥ 2,206 nm) are a consequence of distinctive chemistry (lower Al and Na and higher Si and Mg) compared to the chemistry of white mica with a shorter-wavelength position (≤2,202 nm).

  4. Focused sampling of stream sediments guided by the map of longer-wavelength white mica yielded anomalous concentrations of Cu, Au, and Mo in areas of previously unrecognized mineralized rock.

  5. An evolving understanding of mineralogical and spectral variations within and among deposit types is important for interpreting airborne imaging spectrometer data; our study indicates that imaging spectroscopy (hyperspectral imaging) could prove useful in identifying a number of different deposit types in appropriate parts (e.g., exposed mountainous regions) of Alaska.


This project was funded through the U.S. Geological Survey (USGS) Mineral Resources Program. We want to acknowledge the leadership roles played by USGS colleagues Richard Goldfarb (retired) and Trude King in the proposal stage of the project. William Benzel, Heather Lowers, and David Adams (USGS) provided XRD and EPMA analyses of samples and helpful discussion of the results. We would like to thank the U.S. National Park Service for granting us access for helicopter-supported fieldwork in Wrangell-St. Elias National Park. We are grateful to Drs. Anupma Prakash and Marcel Buchhorn for providing logistical assistance throughout the deployment of the airborne hyperspectral sensor during the rainiest summer on record in Fairbanks, Alaska. We also thank the Alaska Science Center, and Andy Allard, Mike McKinnon, and Fenumiai Ilalio in particular, for their logistical support. The manuscript was greatly improved based on reviews by John “Lyle” Mars (USGS), Steve Ludington (USGS emeritus), and Benoit Rivard (University of Alberta, Canada). Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Garth Graham is a research economic geologist with the U.S. Geological Survey. He received his Ph.D. degree in economic geology from the Colorado School of Mines, Golden, Colorado, in 2011. Most of his research has been focused around base and precious metals deposits in Alaska. Research interests include exploration geochemistry, ore genesis studies, and imaging spectrometry in mineral exploration.

This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Supplementary data