See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/354558666

Remote Sensing in Mineral Exploration

Article in SEG Discovery · July 2004

DOI: 10.5382/SEGnews.2004-58.fea

CITATIONS READS
7 1,858

1 author:

Richard Bedell
Independent Researcher
21 PUBLICATIONS 298 CITATIONS

SEE PROFILE

All content following this page was uploaded by Richard Bedell on 28 June 2022.

The user has requested enhancement of the downloaded file.

SEG
NEWSLETTER
SOCIETY OF ECONOMIC GEOLOGISTS
JULY 2004 NUMBER 58

Remote Sensing in
Mineral Exploration
Richard Bedell (SEG 1994 F), AuEx, Inc., Suite 17, 940 Matley Lane, Reno, NV, 89512†

ABSTRACT
The proliferation of remote sensing platforms has resulted in unprecedented opportunities for ore deposit vectoring.
Importantly, remote sensing technology is now beyond the vague identification of alteration, and can accurately map specific
minerals and directly contribute to the understanding of ore systems. Remote sensing is making discoveries of new alteration
zones within classic and previously well mapped ore systems, as well as outlining their geometry and mineralogy. Confining
this review to the geologically important reflected-light remote sensing systems, there are four main categories of sensors read-
ily available to economic geologists, including the following: (1) submeter resolution panchromatic satellites that offer little
spectral information but provide base maps; (2) multispectral Landsat satellites that can map iron and clay alteration; (3) the
new ASTER satellite that can map important alteration groups and some specific minerals; and (4) hyperspectral airborne
scanners that can provide maps of specific mineral species important to detailed alteration mapping.
At the core of comprehending this plethora of technology is the difference between spectral and spatial resolution. This
review will provide an understanding of the more fundamental aspects of remote sensing systems that will help field geolo-
gists to interact better with and leverage this rapidly evolving technology.

INTRODUCTION Recent articles in the SEG Newsletter Analogue camera systems employ
(Grunsky, 2003) and the publication lenses that focus reflected light onto a
The ability to map different minerals
Economic Geology (Berger et al., 2003; focal plane containing film (Fig. 1). The
or mineral groups relies on the ability
Rowan et al., 2003) have provided some spatial resolution of this system is largely
to image a spectrum with diagnostic
insights into the availability of data and determined by the grain of the film,
absorption features. Old and new detec-
some of the results that can be obtained. which can be as small as 40,000 lines
tors have a range of abilities to map
However, to understand what remote per millimeter for quality aerial photog-
these diagnostic spectra at varying dis-
sensing technology can and cannot do raphy film. By comparison, digital cam-
tance from the surface using low-flying
in an exploration context it is important era systems use an array of charged cou-
aircraft to satellites. Thus, there is a cost
first to examine the implications of spa- pled devices (CCD) instead of grains in
vs. spatial vs. spectral resolution con-
tial and spectral resolution. the film. The size of the CCD on the focal
text. A fourth axis is the signal-to-noise
plane is the limiting factor in spatial res-
ratio, and this may be more important
olution for digital systems, and currently
to mineral discovery than other consid- SPATIAL RESOLUTION reaches tens of CCD per millimeter. For a
erations. This article reviews various
When an image is enlarged on the com- given CCD array on the focal plane, the
opportunities for using remote sensing
puter screen, it eventually breaks into lens system is optimized for that resolu-
data in mineral exploration and deci-
small squares called pixels (picture ele- tion. Thus, it is not possible to add a 2×
phers some of the terminology associ-
ments). These pixels have a specific magnification lens to double the spatial
ated with this field. Because of space
value related to the amount of reflected resolution, although the platform can be
limitations, the article is confined to
energy at a given wavelength. As a first flown at a lower altitude (if it is not a
applications using reflected light and
consideration, spatial resolution can be satellite). Despite this resolution limita-
does not discuss RADAR, thermal emis-
equated to the size of the pixel, so it is tion, the advantages of digital data are
sion, or other remote sensing systems.
important to consider the properties of numerous.
a remote sensing system that determine One major advantage
†E-mail: [email protected] pixel size. of the digital method is to page 11 . . .
JULY 2004 • No 58 SEG NEWSLETTER 11

... from 1 Remote Sensing in Mineral Exploration (Continued)

center. Modern systems, but the utility of digital data

airborne remote for image acquisition, image stability,
sensing systems processing, and delivery usually make
have GPS-con- digital remote sensing the preferred
trolled pixel loca- method. In addition, the availability of
tions on each scan submeter pixel satellite data is a cost-
line so the final effective alternative to mobilizing air-
product can be craft for aerial photo acquisition.
reconstructed
automatically with
high precision. SPECTRAL RESOLUTION
An important In multispectral data, each pixel has a
consideration of value for the different wavelengths
higher spatial res- detected which is proportional to the
olution is the scale amount of energy reflected by that
difference due to respective portion of the surface. The
FIGURE 1. Reflected light remote sensing systems focus reflected
light through a lens onto a focal plane. For digital systems, the light
elevation changes. more bands of data, the more spectral
is focused onto a small array of diodes that become individual pix- Control points are information that is available. As remote
els. For an analogue film system, the light is focused onto grains in needed from a ter- sensing sensors get more bands with
the film emulsion. Spatial resolution is compared by the size of the rain model to narrower bandwidths, they become
film grains versus how small the diodes can be made. minimize distor- more like a geochemical detector. Of
tion. This is course, the methods are different, but a
known as ortho- sampling analogy can be made. In geo-
the geometric stability of the image. correction and increases costs consider- chemistry, features that are a minor
The Instantaneous Field of View (IFOV) ably. For a discussion of spatial accu- constituent of the sample can be
is the image viewed at any instant in racy in a mineral exploration context, detected; similarly, minerals that have
time. For a photograph (Fig. 2a), the see Bedell et al. (1997). unique spectra can be identified in
lens focuses the image onto a relatively Other advantages of digital over remotely sensed images if there is a suf-
large and densely information-packed analogue photography include the ease ficient signal-to-noise ratio. It is there-
layer of film. In a digital system, the of data collection and processing. Film fore routinely possible with many remote
CCD are comparatively expensive so a can always be scanned to become digi- sensing systems to detect mineralogical
pushbroom or a line scan device is used tal, but the processing is a costly addi- features that are smaller than one
to focus the reflected energy onto a rela- tional step. For multispectral data, digi- pixel, although at present it is very diffi-
tively small array of diodes (Fig. 2b). tal information is essential in order to cult to say how much of that material is
The larger IFOV of film results in a process the spectra for specific mineral present. Mixing models are available
radial distortion from the lens com- species or assemblages. that can begin to determine relative
pared to the small IFOV of the digital In summary, analogue film has and proportions, but this is currently at a
system that collects only a few pixels of will continue to have for the near relatively crude level.
data at a time using the undistorted lens future, far higher resolution than digital However, new methods to page 12 . . .

(a) (b)

FIGURE 2. The Instantaneous Field of View (IFOV) of a digital camera system is smaller because diodes are on the focal plane rather than
the film. However, film has grains in the emulsion that are much finer and less expensive to produce. (a) With analogue film, the lens takes
the entire image during one instant in time. Geometric problems may result from the radial lens distortion, but variations in platform move-
ment are minimized. (b) For digital systems, data are focused onto a relatively small array of diodes and the image is built up. Radial dis-
tortion due to the lens is essentially eliminated, but perturbations in platform movement may create random distortions.
12 SEG NEWSLETTER No 58 • JULY 2004

... from 11 Remote Sensing in Mineral Exploration (Continued)

will undoubtedly follow from lab and

field spectroscopy that can be more
quantitative because of the control on
the sample being analyzed.
The discovery of significant miner-
alogical features smaller than one pixel
is exciting and is a powerful demonstra-
tion of the evolving technology.
However, the signal-to-noise ratio is
extremely important and geologists
should consider that it may be more
valuable to have larger pixels with bet-
ter signal-to-noise than it is to have
smaller pixels with ambiguous spectral
signatures. For instance, greater signal
strength might distinguish a high-tem-
perature from a low-temperature
argillic assemblage. This is akin to
increasing the sensitivity of a geochemi-
cal analytical technique whereby a FIGURE 3. The electromagnetic spectrum (EMS) is shown in wavelength (mm) on the
diagnostic absorption feature might X-axis from visible through the shortwave infrared (SWIR) and relative reflectance on
become apparent. Thus, smaller pixels the Y-axis. The naked eye sees from about 0.4 to 0.7 mm. Typical reflectance spectra
may not always be the best solution for for vegetation and dry soil are shown. Also shown are the bars representing the differ-
ore vectoring if they decrease the signal- ent sensors, vertically offset for comparison and implying no variation in the amount of
to-noise ratio. energy detected. Notice the broad spectrum covered by panchromatic film. Similarly,
single panchromatic digital bands are broad within the visible to near IR for increased
In summary, there is an important
signal-to-noise.
trade-off between spectral and spatial
resolution. Decreasing the spectral
bandwidth results in a concomitant 1987, chapter 2). The near infra red unfortunate term because there is no
decrease in signal strength, and there- (NIR) is invisible to the human eye and real definition as to when data go from
fore it may be more important to take a ranges from about 0.7 to about 1.5 µm, multispectral to hyperspectral. In gen-
larger sample (i.e., use larger pixels). whereas the short-wave infra red (SWIR) eral, a single pixel spectrum from mul-
The disadvantage of this approach is ranges from about 1.5 to 2.5 µm and is tispectral systems is like a bar graph,
that larger pixels will have more min- rich in spectral features for clays, car- whereas hyperspectral data are dense
eral mixtures. bonates, sulfates, and other minerals enough to outline continuous curves.
Each remote sensing system has a important to ore vectoring. The distinc- Presumably, technology will continue to
variety of pixel sizes, bands, and signal- tion between the NIR and SWIR is not evolve resulting in higher signal-to-
to-noise characteristics. Of critical precise and is quoted as between 1.0 noise ratios and a concomitant decrease
importance is the location of the bands and 1.5 µm. The term SWIR evolved in spectral bandwidth, leading to more
and their bandwidth. A plot of the elec- from remote sensing and is not recog- constrained definitions for multi- vs.
tromagnetic spectrum (EMS) from the nized in physics or other disciplines. hyperspectral data.
visible range (starting near 0.4 µm) Landsat Thematic Mapper (TM)
through short-wave infrared (SWIR; up bands are depicted in Figure 3 with just
to 2.5 µm) light is useful to examine dif- six bands in the visible to SWIR spec-
ferent detectors and what alteration fea- trum, numbered 1 to 5 and 7. Band 6 is SPECTRAL FEATURES
tures might be determined (Fig. 3). At the thermal band at 8 to 12 µm. An
Not all minerals have unique spectra
longer wavelengths, thermal imaging interesting anecdote: the Jet Propulsion
within the range of wavelengths
employs bands from 8 to 12 µm, but Laboratory (JPL) petitioned Congress at
detected by satellites, airborne scanners,
because of atmospheric attenuation the last minute to provide funds for an
or field and lab spectrometers. However,
there is little to detect at intermediate additional 7th band in the SWIR range
most alteration minerals have unique
wavlengths between 2.5 and 8 µm. In as an aid to finding mineral deposits.
spectral signatures in this range.
addition, not all of the spectrum This is why the band numbering
Important considerations for the detec-
between 0.4 and 2.5 µm is available, sequence is out of order on TM.
tion of these alteration minerals by
owing to attenuation by water and ASTER satellite bands are shown on
remote sensing include the following:
other atmospheric gases. The part of the the bottom line of Figure 3. ASTER has
spectrum available to the human visual five bands in the SWIR compared to 1. Are the spectral resolutions high
system (HVS) is limited to a range of Landsat TM’s single band 7. Many air- enough to define uniquely a given
about 0.4 to 0.7 µm; the HVS can borne scanners have hundreds of spec- mineral or group of minerals?
arguably detect hundreds, if not mil- tral bands in this region, and such scan- 2. Are the abundances of the minerals
lions, of color changes, although not ners are sometimes referred to as and the signal-to-noise of the sensor
equally throughout the EMS (Drury, “hyperspectral.” However, this is an good enough to detect them?
JULY 2004 • No 58 SEG NEWSLETTER 13

As an example, the mineral alunite is after its launch in 1982. It has been to factor analysis used in multi-variate
an important mineral for ore vectoring. used to make alteration maps and geochemical analysis. This approach
Figure 4 shows a laboratory spectrum large-area synoptic images for regional has multiple advantages in that it can
for alunite compared with the satellite geology and target selection. Its prede- avoid well known false anomalies such
based ASTER spectra and bars represent- cessor, the Earth Resource Technology as vegetation while preserving the origi-
ing multispectral TM bands. At this Satellite (ERTS; later renamed Landsat nal signal-to-noise ratio.
scale, most “hyperspectral” data would MSS for multispectral scanner) was For an excellent review of remote
look similar to the laboratory spectrum. launched in 1972; it had no SWIR band, sensing image processing techniques in
The Landsat TM satellite can only was coarser in resolution (80 m pixels), mineral exploration, see Sabine (1999).
distinguish if there is absorption within and had relatively little to offer except
the 2.0 to 2.5 µm part of the EMS (i.e., iron oxide detection and a synoptic TM – Pasco Canyon
in the SWIR). However, all clays and view. Landsat 1 through 4 were of simi- Pasco Canyon is a recently discovered
most carbonates and sulfates have this lar configuration, but Landsat 5 gold prospect east of Round Mountain,
signature. In comparison, the ASTER Thematic Mapper (TM), launched in Nevada. Figure 5 shows the TM signa-
satellite has five bands in the SWIR so it 1982, had 30-m pixels plus the SWIR ture generated by a form of FPCS. This
can begin to distinguish different min- band 7. Landsat 7, launched in 1999, anomaly successfully maps the hypo-
erals. In most cases the exact mineral or added a 15 m panchromatic band. gene alteration but not the regionally
mineral assemblage is not known with- With improved image processing meth- pervasive supergene alteration. The
out field spectra or direct outcrop obser- ods, TM became a standard in the min- area mapped is a 400 × 600-m boiling
vation, but combining ASTER data with eral exploration industry. Given the few vein swarm (bladed quartz after calcite)
field spectroscopy is nevertheless a pow- spectral bands available on Landsat and breccia. What is remarkable is that
erful mapping tool. For an excellent TM, the main alteration detected is sim- this is in one of the most heavily
review of practical ore vectoring with ply clay and iron oxide. Images were prospected regions of Nevada, and lies
field SWIR spectroscopy and the identifi- originally generated by simple band within some detailed geologic map
cation of key mineral assemblages, see ratios, but included many false anoma- sheets, but has never been mapped,
Thompson et al. (1999). lies such as vegetation. staked, or drilled. This prospect is on the
A significant breakthrough in pro- edge of the pediment with a magnetic
cessing was achieved by Alvaro Crosta low, and will be first drilled in 2004.
EXAMPLES OF PLATFORMS as part of his PhD thesis (Crosta and After it was decided where to look,
AND APPLICATIONS Moore, 1989), and was independently regionally, this discovery was made
tested with great success by Loughlin purely by TM image analysis, and
Landsat TM (1990, 1991). The technique uses a form demonstrates how data available since
Landsat TM has been a standard of Feature-oriented Principal Component 1982 can still result in
to page 14 . . .
workhorse in our industry since shortly Selection (FPCS), which is comparable discoveries today.

FIGURE 4. A laboratory spectrum for alunite (continuous line) is

similar to the detail of a hyperspectral imaging system at this scale.
The dashed line is the same spectrum subsampled to the resolu- FIGURE 5. Landsat TM hypogene anomaly at the Pasco Canyon
tion of the ASTER satellite. The gray horizontal bars represent the gold project. The gray levels of different pixels outline a boiling
data that would be obtained from Landsat TM. The TM data could zone (bladed quartz after calcite) in a brecciated, low-sulfidation
only indicate the presence of clay or sulfate, whereas the ASTER system anomalous in gold and toxic element geochemistry. The
spectrum could reveal the presence of an advanced argillic alter- background map is from the Pine Ranch 7.5’ topographic quad-
ation assemblage. rangle in northern Nye County, Nevada. The grid is 500 m.
14 SEG NEWSLETTER No 58 • JULY 2004

... from 13 Remote Sensing in Mineral Exploration (Continued)

There are many archives of TM data, additional bands allow for distinction Figure 6 is an image the author pro-
and most of the world is now available between groups of clays, sulfates, car- cessed in early 2001 when ASTER data
in cloud-free images. Costs vary from bonates, and other minerals. Rarely can were first released. This image shows
free to $425 per scene (Table 1). ASTER be specific as to the exact min- stockpiles around a gypsum plant near
eral species, but it can make some Empire, Nevada. The match between
ASTER – Gypsum important distinctions between argillic the spectral library (subsampled from
ASTER is a joint US-Japanese satellite and advanced argillic assemblages, as the USGS lab spectra to the spectral res-
optimized for geologic remote sensing, well as propylitic and phyllic mineral- olution of ASTER) is excellent and
and data have been available to the ogy. ASTER has two bands in the visible highly indicative of gypsum. This was
public since 2001. ASTER has five spec- range, and one near-IR band with pre- an exciting moment because it was the
tral bands in the SWIR (compared with cise positioning useful for distinguish- first time the author had been able to
only one for TM; Fig. 3) and these ing jarosite from other iron oxides. map a distinct mineral species from
space. It is important to note that the
image also mapped two additional
TABLE 1. A Comparison of Remote Sensing Platforms known gypsum prospects and was very
well rectified to 1:24,000 topographic
Satellite maps after a linear shift of about 200
Spectrum Bands Pixel (m) Cost ($US) Website m. ASTER comes in a stable UTM map
QuickBird
projection, but often needs a linear shift
4.45–9.0 1 0.5–1 22.50 sq/km www.digitalglobe.com of about 200 m to rectify it accurately.
4.5–9.0 4 2–5 25.00 sq/km The gypsum example is unusual in
that the spectra are well classified with
IKONOS
the few bands available with ASTER.
0.45–9.0 1 1 18.00 sq/km www.intecamericas.
0.45–0.85 4 4 15.00 sq/km com/satprices.com Usually, ASTER must be combined with
field spectroscopy to determine groups
SPOT 5 of minerals or assemblages in order
0.48–0.71 1 2.5 60 × 60 km 6,700.00 www.spotimage.fr effectively to map alteration.
0.5–0.89 3 10 2,700.00
1.58–1.75 1 20
A recent paper by Rowan et al.
(2002) presents spectra from the ASTER
TM satellite over the Cuprite mining district
0.45–2.35 7 30 180 × 180 km 425.00 edcimswww.cr.usgs.gov/ in Nevada, including muscovite, calcite,
10.45–12.5 1 120 pub/imswelcome
alunite, opal, and other minerals. It is
ETM important to realize that the spectra
0.45–2.35 6 30 180 × 180 km 600.00 landsat7.usgs.gov can only infer these minerals, and field
0.52–0.9 1 15 spectrometry or other data are required
10.4–12.5 1 60 to verify this as fact. ASTER is a very
ASTER
powerful exploration tool but it cannot
0.6–1.2 3 15 60 × 60 km 55.00 asterweb.jpl.nasa.gov be treated in isolation without field-
1.5 – 2.5 6 30 work. To make the most of the data it
8–12 5 90 takes iteration between the field and
data processing. Furthermore, the data
Hyperspectral Satellite
Hyperion cannot be used directly without check-
0.4–2.5 220 30 7.5 × 100 km 250.00 eo1.usgs.gov/instru/ ing the correction to reflectance.
hyperion.asp Atmospheric effects, spectrometer drift,
and other artifacts can render the data
Hyperspectral Airborne
AVIRIS
useless.
0.38–2.5 224 20–5 10-km wide 500.00/ aviris.jpl.nasa.gov/html/ Figure 7 shows a spectral plot from
flight line aviris.quicklooks.html an ASTER pixel north of Reno, Nevada,
where the “apparent correction to
SpecTIR
reflectance” product ordered from the
4.5–2.45 227 0.5–5 13.00 sq/km www.spectir.com
plus mob ASTER web site was compared to data
calibrated and corrected by Empirical
HyMap Line Calibration. This method of correc-
0.45–2.5 126 www.intspec.com tion to reflectance employs a grid of
field spectroscopy data over light and
Notes: the high spatial resolution satellites have one wide panchromatic band for good signal-to- dark targets on the ground. A simple
noise definition, and some other narrower bands in the visible and near IR for color; the amount of line-fitting approach suppresses various
reflected energy available decreases at longer wavelengths, and a different detector with larger
artifacts and corrects the data to
pixels is often used for SWIR and Thermal IR; prices are given with radiometric corrections, but
without rectification; cost is for US data and may be priced differently for other areas; all prices reflectance. Correction to reflectance is
are for archive data except for the high-resolution panchromatic satellites that do not hold essential before the ASTER satellite
archives; requests for data acquisition not in an archive are more expensive spectra can be directly compared to a
JULY 2004 • No 58 SEG NEWSLETTER 15

how, or even if, this crucial step was

accomplished before mobilizing field
crews.
AVIRIS – Goldfield
The Airborne Visible/Infrared Imaging
Spectrometer (AVIRIS) was the first
hyperspectral scanner available for
public use. Its 224 bands provide near-
continuous spectra with a resolution
high enough to see most major absorp-
tion features. AVIRIS was developed by
the Jet Propulsion Laboratory and has
an archive of data flown over small
parts of the United States. Data from
other countries is scarce primarily
because the sensor is mounted on a U2
spy plane.
Figure 8 shows an AVIRIS image over
the classic high-sulfidation gold system
of Goldfield, Nevada. Goldfield has
FIGURE 6. Gypsum identified by ASTER after correction to reflectance. (a) Comparison of undergone almost continuous explo-
the spectra obtained by ASTER and a USGS laboratory library spectrum of gypsum sub-
ration for a century, and yet remote
sampled to the spectral resolution of ASTER. (b) ASTER image of the Empire gypsum mine
sensing is still making discoveries. To
area, Nevada. The mottled dark texture is cloud shadow. The box outlines the area in
image 6c. (c) Closeup of the ASTER 15 m visible band pixels on the gypsum piles. (d) use AVIRIS as a gold vectoring tool, the
Classified image using an algorithm to locate gypsum; two gypsum piles are accurately tailings were analyzed to determine the
classified in this image. dominant alteration mineral associated
with the ore, and revealed a dominant
alunite spectrum. This image was then
spectral library. In Figure 7 it can be reflecting in the ASTER data where they classified by using a laboratory alunite
seen that the original data are com- should be absorbing and vice versa. spectrum as a target and finding similar
pletely wrong when compared to the There are several ways to do these cor- spectra throughout the image (Fig. 8a).
corrected data and the independently rections, and the geologist obtaining Field spectrometer inspection showed
tested data using field spectra. Data are imagery should ask the question as to that the alunite-classified image was
very accurate.
Figure 8b is a pyrophyllite image,
Spectral Profile FIGURE 7. Correction to and in late 2003 this was field-checked,
reflectance: this plot shows
confirming new unmapped pyrophyllite
wavelength in microns on
the x-axis and relative
zones corresponding to this image.
reflectance on the y-axis. Years of recent field mapping and thou-
0.40 The thin line represents an sands of field spectrometer measure-
integrated grid of soil sam- ments have been made in this mature
ples over a known ASTER exploration area (Bob Bennett, personal
pixel. This pixel was sampled commun.), and yet remote sensing can
independently of the pixels still discover important new features.
used to drive the empirical Pyrophyllite represents a relatively
line calibration as a test. The high temperature acidic hydrothermal
0.35 thick solid line represents
event, and field relationships at many
the ASTER spectrum of the
gold projects throughout the world
Value

same pixel. Note the excel-

lent fit, considering that the demonstrate this event to be early and
angular changes in the line not related to gold mineralization.
are the only places where Instead, many gold events within a
ASTER has a spectral band. multiphase alteration system are
0.30
The dashed line shows the related to retrograde collapse.
SWIR bands from the same Nevertheless, in the field, the naked eye
pixel from the “apparent will often chase visible alteration that
corrected-to-reflectance” may not be associated with the gold
image downloaded from
event. For these reasons, when com-
the ASTER website. Note
that the corrected data are
bined with accurate fieldwork, remote
0.25 sensing and field spectroscopy are
very different from the
“apparent” correction. extremely important contributors in
mapping, vectoring, and
to page 16 . . .
discovery. The recent
1.0 1.5 2.0
article in Economic
16 SEG NEWSLETTER No 58 • JULY 2004

... from 15 Remote Sensing in Mineral Exploration (Continued)

(a) (b) 1982), which makes it possible to get

data over recent mining districts prior to
disturbance, and to select cloud-free
images at the right time of year.
For district-scale work, ASTER is
rapidly becoming the standard explo-
ration and mapping tool. ASTER has
three 15-m visible-to-NIR bands, and
the five 30-m SWIR bands that can be
fused to make coherent images at
1:25,000 scale. The five SWIR bands can
make excellent alteration zoning maps
for mineral assemblages. ASTER also
has five 90 m thermal bands that can
map silica, but even with these very
large pixels the signal-to-noise ratio is
poor and results are variable. ASTER
may be all that is required for prospect
FIGURE 8. AVIRIS airborne scanner of the high-sulfidation system at Goldfield, Nevada.
scale work provided it is accompanied
The image is about 6 km across. The brightness of the pixels demarcates (a) alunite and
(b) pyrophyllite. (a) Large, alunite-bearing tailings piles (T) are clearly identified in the NW
by a field spectroscopy survey. However,
corner of the image, but much of the remaining alunite outlines an arcuate structure the ASTER archive is young and many
believed to be related to the main mineralizing event. (b) In contrast, pyrophyllite occu- areas do not yet have acceptable
pies a more central position, and is believed to represent an earlier acidic alteration images in the archive that are cloud
event. free and taken at the right time of year.
Given the time and cost of hyper-
spectral surveys, these are best applied
Geology by Berger et al. (2003) provides cheaply, in conjunction with field spec- where commitments to the property
a detailed case study using AVIRIS over troscopy. As a broad framework, a and district are strong. A hyperspectral
a porphyry copper system in Arizona, hyperspectral survey plus data process- survey can reveal zoning patterns and
and demonstrates how powerful these ing will cost tens of thousands of US dol- distinct mineral assemblages that may
data can be when combined with field lars, whereas ASTER data and process- take years to find (or may never be
mapping and geophysics. ing will cost thousands, and TM data found) with other methods.
AVIRIS is only one example of a and processing will cost about US$1,000 Hyperspectral surveys are typically
hyperspectral scanner commercially or less. All these methods must include flown for 1 to 5 m pixels, and can be
available for flying mineral exploration fieldwork, because discovery cannot be effectively integrated with a drill pro-
targets. AVIRIS is a JPL instrument and made without physical field analysis gram and ground geophysical surveys.
can be hired, but other purely commer- and assays. If remote sensing data are This may require large mobilization
cial systems may be better suited and too expensive and take too long to costs, but inter-company and -agency
cheaper. In addition, any flights done obtain, there may not be time to do all cooperation can share these costs and
with AVIRIS go into the public domain the necessary fieldwork where land greatly minimize the overall survey
within a few months, and because holding costs and seasonal access are expenditure.
AVIRIS is built into a U2 spy plane, limitations.
access to international skies is a prob- Landsat TM can be very effective for
lem. However, an archive is available regional work. With advanced process-
for exploration in the United States. ing techniques to distinguish hypogene SUMMARY
Hyperspectral systems available to geol- alteration, TM has broad bands and Digital data have many advantages
ogists are included in the summary good signal-to-noise ratio, so it can pro- over analogue data, but the spatial res-
Table 1. vide excellent regional images. TM’s 30- olution of analogue film is still far
m pixels and large aerial coverage of beyond current digital technology.
180 × 180-km make for a compelling Nevertheless, the advantages of digital
and cost-effective regional database. technology are multiple and override
PLATFORMS AND PRACTICAL Similarly, when enlarged and pro- the pure resolution benefit of analogue
CONSIDERATIONS cessed, TM images make coherent film. Film can still be scanned and recti-
Although hyperspectral scanners pro- 1:50,000-scale maps. The new ETM fied, but the rectification problems and
vide the greatest amount of informa- satellite data (TM7 launched in 1999) cost are considerable relative to most
tion, they may not be the best choice in have an additional 15-m panchromatic digital systems.
a practical framework. Mobilization, band so that the 15-m texture can be Geologists should consider a sacrifice
data acquisition, and processing costs fused into the other 30 m spectral bands in spatial resolution for better signal-to-
can be considerable. Consider the qual- for effective 1:25,000 scale images. An noise ratio when using spectral signa-
ity of data that can be obtained from important consideration is that TM has tures to map lithology or alteration. For
the ASTER satellite, quickly and a large historic archive (dating from example, alteration and lithologic map-
JULY 2004 • No 58 SEG NEWSLETTER 17

ping in well-vegetated areas is entirely this summary as well as Eric Grunsky, Grunsky, E., 2003, Two papers on the use of
possible if there is some soil exposed Alvaro Crosta, and Jeremy Richards for hyperspectral airborne-satellite imagery in
and the sensor provides superior signal- useful editing and comments in the later mineral exploration: Economic Geology—
to-noise. With digital topographic maps stages. I would also like to thank my an invited commentary on journal papers:
SEG Newsletter no. 55, p. 24–25.
and high-resolution panchromatic colleagues at the Arthur Brandt Labora-
Loughlin, W.P., 1990, Geological exploration
satellites increasingly available, fusing tory of Exploration Geophysics and the in the western United States by use of air-
multi-hyperspectral anomalies into many colleagues I have interacted with borne scanner imagery, in Legg, C., ed.,
these base maps is a very effective map- over the years that make mineral explo- Remote sensing: An operational technology
ping and vectoring tool. ration an exciting profession. for the mining and petroleum industries:
This short review does not allow the London, Institute of Mining and
space or color to emphasize the out- Metallurgy, October 1990, Proceedings, p.
standing products that can be gener- 223–241.
ated by the current array of remote REFERENCES ——1991, Principal components analysis for
sensing systems for exploration and alteration mapping; Photogrammatic
Bedell, R.L., Lei, Y., and Heston, G.S., 1997,
Engineering Remote Sensing, v. 57, p.
mapping. More examples will be avail- Spatial data accuracy and its importance
1163–1170
able in the literature and there are two to effective mineral exploration: Prospec-
Rowan, L.C., Hook, S.J., Abrams, M.J., and
remote sensing sessions in field map- tors and Developers Association of Canada,
Mars, J.C., 2003, Mapping hydrothermally
ping and exploration for remote sens- Proceedings of Exploration 97: Fourth
altered rocks at Cuprite, Nevada, using the
ing at the cosponsored Geological Decennial International Conference on
Advanced Spaceborne Thermal Emission
Society of Nevada (GSN) meeting in Mineral Exploration, Toronto, Ontario,
and Reflected Radiometer (ASTER), a new
Canada, September 14–18, 1997, p. 89–96.
May 2005, “Windows to the World.” A satellite-imaging system: Economic
Berger, B.R., King, T.V.V., Morath, L.C. and
special SEG publication is planned to Geology, v. 98, p. 1019–1027.
Phillips, J.D., 2003, Utility of high-altitude
provide detailed data and analysis of Sabine, C., 1999, Remote sensing strategies
infrared spectral data in mineral explo-
this rapidly evolving technology. for mineral exploration: Remote sensing for
ration: Application to northern Patagonia
the earth sciences, in Rencz, A.N., ed.,
In summary, remote sensing has Mountains, Arizona: Economic Geology,
Manual of remote sensing [3rd edition]:
matured in data quality and processing v. 98, p. 1003–1018.
John Wiley, Inc., v. 3, p. 375–447.
methods to offer many direct contribu- Crosta, A.P. and Moore, J.M., 1989,
Thompson, A.J.B., Hauff, P.L., and Robitaille
tions to the field of economic geology. It Enhancement of Landsat thematic mapper
A.J., 1999, Alteration mapping in explo-
is hoped that this short review provides imagery for residual soil mapping in SW
ration: Application of short-wave infrared
a practical framework for skilled eco- Minas Gerias State, Brazil: A prospecting
(SWIR) spectroscopy: SEG Newsletter, no.
case history in greenstone belt terrain:
nomic geologists to consider and apply 39, p. 1, 16–27. 1
Earth Resource Institute of Michigan,
this plethora of data. Proceedings of the 7th Thematic Conference
on Remote Sensing and Exploration
Geology, Calgary, Alberta, Canada
p. 1171–1187.
ACKNOWLEDGMENTS Drury, S.A., 1987, Image Interpretation in
I would like to thank Mark Cool- Geology, 2nd edition: London, Chapman
baugh for reviewing an early version of and Hall, 283 p.

View publication stats