>> for
from
ng in
cies.
| Ca-
astern
porly-
re of
strict,
els as
Bodie
e was
Were
ed all
1. We
m the
s and
torite,
1atite.
ttered
nount
was
plus a
apped
pond
were
ed as
ter of
scene
dium-
JSGS
t the
ations
rder
ration
e was
ations
vever,
ite in
which
linite.
ted at
t map
lonite.
lation
istrict,
ineral
apped
The
lonite,
yart of
In the VNIR , major differences show up in the comparison
between results obtained by both methods for those minerals
that could be mapped by SAM (hematite, goethite and jarosite).
SAM did not find much hematite in the whole scene, whereas
Tricorder mapped significant clusters of pixels as hematite in
both districts. However, some of the pixels mapped as hematite
by Tricorder were assigned to goethite by SAM. On the other
hand, pixels that Tricorder mapped as goethite were assigned to
jarosite by SAM.
8. SPECTRAL ANALYSIS
In order to assess the accuracy of the results, particularly those
obtained by Tricorder which identified a greater number of
minerals, we selected six sites for ground checking and
collected samples for laboratory spectral analysis. These sites
correspond to areas where one or more alteration minerals were
mapped by Tricorder in the VNIR and SWIR. Rock samples
from each site were analyzed using a Beckman UV-5240
spectrophotometer. The resulting spectra were then convolved
to AVIRIS bandwidths for comparison with pixel spectra.
Table 1 shows these six sites with the sample number, their
location on the scene and a comparison of the minerals assigned
to them by both, Tricorder and SAM. For the location without a
sample number in Table 1, only the pixel spectrum and a library
spectrum were used, since it was not possible to collect a
ground sample.
Sample Pixel Location (row/line) Tricorder Result SAM Result
BO-17 481/435 Na-montmoril. Na-montmoril.
BO-11a 161/352 Ca-montmoril. Na-montmoril.
BO-14b 589/445 kaolinite(pxyl) kaol./smectite
- 524/352 kaolinite kaol./smectite
BO-14a 485/442 halloysite Na-montmoril.
BO-05 78/344 Na-montmoril. & goethite Na-montmoril. & jarosite
Table 1 - Comparison of results between SAM and Tricorder for
selected sites. Spectra for each site is shown in figures 9 to 14, respectively.
An overall coincidence of the most important spectral features
for each mineral was observed. The vibrational absorption
features located at approximately 2.20 mm, characteristic of
hydroxyl-bearing alteration minerals such as those that occur in
Bodie and Paramount, shows a good correlation in both
position and shape between ground sample, AVIRIS pixel and
library spectra for all these samples.
The spectra of the pixel which Tricorder mapped as kaolinite
(sample without number) shows a remarkable coincidence for
the doublet feature centered at approximately 2.20 mm. The
same coincidence was observed for the sample, AVIRIS and
library spectra of the pixel mapped as poorly-crystalline
kaolinite (sample BO-14b), despite the fact that in this case the
feature is more subtle than in well-crystalline kaolinite.
Halloysite (sample BO-14a) was also mapped and differentiated
from well- and poorly-crystalline kaolinite, despite having a
even more subtle doublet in the same region. Apparently
Tricorder can deal with even the most subtle spectral
differences because on its ability to analyze for total shape,
using the three or four most significant SWIR absorption
features and weighting these features differently in the final
steps of the fitting process.
A good correlation between position and total shape can be
seen for the more obvious absorption features related to Na-
montmorillonite and Ca-montmorillonite (samples BO-17 and
BO-11a), particularly the single feature near 2.2 mm. Different
depths of this feature for both minerals was one of the criteria
used by Tricorder to discriminate between them.
165
Sample BO-05 corresponds to a location identified by Tricorder
as containing Na-montmorillonite and goethite. Its spectrum
clearly show the more obvious features due to these two
minerals, the electronic absorption due to Fe” ion of goethite at
0.92 mm and the vibrational feature due to the OH ion of
montmorillonite at approximately 2.20 mm. As these two
spectral regions (VNIR and SWIR) are processed separately by
Tricorder, this pixel has been assigned to both minerals. SAM,
however, mapped this same pixel as jarosite using the VNIR
bands.
9. CONCLUSIONS
We managed to recognize and map a variety of minerals
associated with hydrothermal alteration processes using
AVIRIS data at the Bodie and Paramount mining districts,
California. The method selected for calibrating the data and for
retrieving reflectance from radiance was precise enough to
allow mineral mapping without the use of independent ground
and atmospheric data. This represents a major advantage for
mineral exploration activities in which ground data acquisition
is usually not possible or desirable during the initial phases of
an exploratory program.
In general, Tricorder was able to produce mineral maps with
greater variety of alteration minerals than SAM. Results
obtained for the SWIR region showed a relatively good
coincidence between the two methods for alteration minerals of
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
