IEEE Geoscience and Remote Sensing Magazine - June 2019 - 52

The focus of this tutorial is on the use of spectral information to derive quantitative information on the absorption features and spatial distribution patterns of minerals
and rocks from spectral information. Hence, the remainder of this introduction focuses on optical remote sensing in the visible to shortwave infrared wavelength region,
and in particular on hyperspectral remote sensing (imaging spectrometry).
Much of modern geologic remote sensing is built on
[4] and [5], the authors of which meticulously analyzed
the spectra of minerals and rocks and designed the first
spectral libraries linking absorption features to mineralogy. In that same era, NASA's Jet Propulsion Laboratory
started to develop and fly airborne imaging spectrometers from the early Airborne Imaging Spectrometer to
the present Airborne Visible Infrared Imaging Spectrometer (AVIRIS) and AVIRIS Next Generation (AVIRIS-NG)
[6], [7]. In 2000, NASA launched the first imaging spectrometer in orbit, the Hyperion EO-1, which was decommissioned in 2017 [8]. There are plans for several more
imaging spectrometers in space, including the German
EnMAP mission [9], the NASA-led HyspIRI mission [10],
and the CHIME mission (the European Space Agency's
hyperspectral imaging satellite planned as Sentinel-10 for
launch in 2024) [11].
Laboratory-based spectroscopy within the visible
to near infrared (VNIR) and SWIR wavelength regions
has been well established since the 1970s for discriminating minerals and mineral groups [12]-[15]. In particular, the work of Hunt and Salisbury showed that key
spectral signatures and absorption features allow identification and mapping of iron minerals (e.g., goethite,
hematite, jarosite, and siderite) using VNIR (0.3-1 µm)
reflectance, while clay minerals (e.g., carbonates, micas,
and sulfates) have diagnostic absorption features in the
SWIR (1-2.5 µm). A wealth of other papers, summarized in [16] and [17], have shown applications of mineral mapping to aid in the exploration of ore deposits.
In the realm of oil and gas discovery, hyperspectral remote sensing has been successfully used to map natural
oil and gas seeping from petroleum reservoirs beneath
Earth's surface (onshore and offshore) [18], and, more
recently, it has been shown that methane plumes associated with seeps can be detected with remote sensing
data [19].
Several techniques and tools have been developed to
extract spectral parameters from the reflectance spectra of
minerals, with the aim of enabling the detection, discrimination, and mapping of ores. A summary is provided in
[20]. In the early days, binary encoding [6] of spectra was
used for mineral mapping. With the increase of computing power, more sophisticated techniques were developed
to derive quantitative information on absorption features.
Roger Clark and his team at the U.S. Geological Survey
(USGS) were instrumental in developing algorithms to isolate absorption features using so-called hull removal and
52

subsequently calculating the absorption depth and position of features [21].
This methodology was subsequently refined by introducing relative-band-depth mapping from radiance data
using local continuum corrections, with the advantage of
also isolating weak absorption features that often were excluded from the analyses using a continuum removal approach [22]. In 2003, the USGS released the Tetracorder
expert system [23] for mineral mapping, which compares
unknown spectra to a set of spectral libraries of minerals, employing diagnostic absorption features isolated by
means of the hull continuum removal approach. Tetracorder uses the diagnostic wavelength regions of mineral
groups to arrive at a spectrum match, expressed as a correlation to different minerals.
Subtle variations in shape and wavelength position of
spectral absorption features indicate changes in the mineralogy and chemistry of the minerals and rocks under investigation. In this tutorial, we present two freeware software packages, chosen from a number of possible tools,
that allow analysts to calculate the shape and wavelength
of spectral absorption features as input for mineralogical
studies. The first tool, IDL DISPEC, is based on [13] and
allows the user to estimate the position, depth, area, and
asymmetry of each spectral absorption feature. The second
program, the Minimum Wavelength Mapper [24], [25], derives the wavelength position and depth of the deepest absorption feature for each image pixel and per-wavelength
region for a hyperspectral image data set. Results are presented from two application cases in geological remote
sensing, one consisting of a proximal hyperspectral sensing of a hand specimen in the laboratory and the other
using an airborne hyperspectral data set from the Cuprite
area in Nevada.
MINERAL SPECTROSCOPY OVERVIEW
The basis of mineral spectroscopy and its ability to measure and interpret composition from VNIR and SWIR reflectance measurements lies in the physical interaction of
such radiation with the electronic and molecular structure
of materials [4], [12]. In particular, VNIR radiation affects
the electronic orbit transitions within ferric iron minerals (e.g., hematite and goethite), which in turn affects the
amount of reflected light (charge transfer absorption) from
their surface. Radiation interacting with minerals, such as
phyllosilicates, affects the energy levels within their molecules, changing the vibrational modes between atoms
(e.g., in terms of stretching versus bending). Minerals
containing hydroxyl groups (OH-), water, and carbonate
exhibit spectral absorption features within the SWIR wavelengths due to the absorption overtones and combinations
of their fundamental vibration energy modes, which occur
at longer wavelengths. The amount of energy absorbed at
such absorption overtones is approximately proportional
to the abundance of such minerals (although not strictly
linear in the VNIR and SWIR ranges), while the wavelength
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

JUNE 2019
IEEE Geoscience and Remote Sensing Magazine - June 2019

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2019
