HIGH-SPECTRAL RESOLUTION REMOTE SENSING FOR MINERAL MAPPING IN THE BODIE AND PARAMOUNT
MINING DISTRICTS, CALIFORNIA
Alvaro P. Crösta
Geosciences Institute - University of Campinas
Campinas, Säo Paulo, Brazil
Charles Sabine
James V. Taranik
Desert Research Institute
University and Community College System of Nevada
Reno, Nevada, USA
Commission VII, Working Group 4, ISPRS
KEY WORDS: Remote Sensing, Geology, Radiometry, Hyperspectral, Classification, Mapping, Processing, Algorithms.
ABSTRACT:
This paper examines and compares recently developed techniques for hyperspectral data processing in the context of mineral
mapping and exploration for precious metals. Hyperspectral data comprises AVIRIS (Airborne Visible and Infrared Imaging
Spectrometer) imagery acquired by NASA's ER-2 aircraft at an altitude of 25 km in August, 1992. The study area is the Bodie and
Paramount mining districts, in California, USA, both containing hydrothermally altered Tertiary volcanic rocks.
Hyperspectral data were converted from radiance to apparent surface reflectance using a radiative transfer approach, based on
atmospheric modelling using a modified MODTRAN method. The data were then processed for mineral identification using two
techniques: spectral angle mapping (SAM) and the Tricorder algorithm. SAM is a supervised classification technique for mapping
the similarity of image spectra to reference spectra. Tricorder uses an optimized least-square method to compare the spectrum for
each pixel on the scene to library spectra.
The results obtained for the Bodie and Paramount districts show that alteration zones of different mineralogy can be separated
using these methods without any knowledge of field spectra or any a priori field data, thus configuring a "true" remote sensing
method. Applications of this kind of technology are likely to benefit mineral exploration programs for precious metals,
particularly in frontier regions where little geological information is readily available.
1. INTRODUCTION
Hyperspectral remote sensing has been under development
since the first experimental sensor of this type, the Airborne
Imaging Spectrometer (AIS), was first flown in 1983. Its
successor, the Airborne Visible/Infrared Imaging Spectrometer
(AVIRIS) was developed at JPL in 1987 and has continued to
evolve since then. Data acquired by AVIRIS and also by some
commercially operated imaging spectrometers developed during
the 80's allowed a significant number of geological applications
to be developed. Among these applications is the
discrimination of rocks and minerals important to mineral
exploration, in particular, mapping hydrothermal alteration
minerals.
However, the lack of routine availability of hyperspectral data
covering diverse geological environments in different regions
has restricted the full utilization of this type of data. Also,
hyperspectral data required image processing algorithms
specifically designed to take advantage of the high spectral
resolution and to cope with the much larger amount of data than
those used with conventional multispectral data.
With the continued deployment of AVIRIS in different regions
of North America, Europe (1991) and South America (1995),
and the continued operation of other airborne scanners,
progressively more data have become available. New
algorithms for processing hyperspectral data have also been
developed and some of them algorithms have recently been
incorporated into commercial image processing software.
This technology is currently under significant development,
with several new imaging spectrometers being built and
operated in recent years. In addition, spaceborne imaging
spectrometers are planed for the near future. These
developments, summarized by Taranik and Crôsta (1996),
indicate a clear trend toward operational use of hyperspectral
data in geological remote sensing in the near future.
2. OBJECTIVES
This paper examines and compares two different algorithms, the
Spectral Angle Mapper (SAM) and Tricorder, recently
developed for identifying surface materials in imaging
spectrometry data. The objective was to assess the performance
of these algorithms and their ability to map alteration minerals
important to precious metals exploration. Unlike previous
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996