Hyperspectral Sensors and Applications
39
time and micro- to macro-scale morphology. A further limitation is that the
dominant mineral components of a rock, which generally determine their
petrographic classification, do not necessarily produce the most prominent
spectral features. Even so, dominant spectral patterns produced by minor
mineral components with strong absorption features can significantly assist
in characterising rocks, at least within well constrained geological settings
(Longhi et al. 2001).
A range of multispectral sensors have been used for mapping a variety
of different minerals including the LANDSAT (Carranza and Hale 2002) and
SPOT sensors (Cole 1991; Chica-Olmo et al. 2002). However, given the spatial
resolutions of such orbital sensors, techniques such as SMA are often required.
As imaging spectrometers such as AVIRIS and HyMap provide higher spectral
resolution data, these sensors have frequently been preferred for geological
mapping and mineral exploration although the extent of coverage is often
reduced. As examples, AVIRIS has been used to map mineral assemblages in
areas of the United States (Mackin et al. 1991; Kokaly et al. 1998), including areas
of ammonium in hydrothermal rocks (Baugh et al. 1998), and as a tool in gold
exploration (van der Meer and Bakker 1998; Rowan 2000). Similarly, HyMap
has been utilised to map gold mineralization in Western Australia (Bierwirth
et al. 2002) and for detecting hydrocarbons (Horig et al. 2001). Simulations
have also suggested that different types of metamorphic rocks could be further
discriminated using AVIRIS or MIVIS data. Other applications have included
the measurement of exposed lava and volcanic hotspots (Oppenheimer et al.
1993; Green et al. 1998b), the study of petroleum geology (van der Meer and
de Jong 2001) and the application of AVIRIS to study mineral-induced stress
on vegetation (Merton 1999; Merton and Silver 2000). Recent studies (Kruse
et al. 2003) have also investigated the use of Hyperion for geological mapping
and monitoring the processes that control the occurrence of non-renewable
mineral resources, especially mineral deposits.
The difficulties of applying remote sensing to the mapping of minerals are
illustrated in heavily forested regions, where reflectance based data are less
useful because of the lack of direct observations of the ground surface. Steep
terrain, heavy rainfall and cloud cover as well as high population densities
(and hence changes in land use) in many regions (particularly in the tropics)
also limit exploration using remote sensing data (Cervelle, 1991).
1.6.4
Environmental Hazards and Anthropogenic Activity
Hyperspectral remote sensing offers many opportunities for monitoring natural hazards such as bushfires, volcanic activity and observing anthropogenicinduced activities and impacts such as acidification, land clearing and degradation (de Jong and Epema, 2001), biomass burning (Green et al. 1998a), water
pollution (Bianchi et al. 1995a; Bianchi et al. 1995b), atmospheric fallout of
dust and particulate matter emitted by industry (Ong et al. 2001) and soil
salinity (Metternicht and Zinck 2003). As an example, Chisholm (2001) indicated that spectral indices could be used to detect vegetation water content at
39
time and micro- to macro-scale morphology. A further limitation is that the
dominant mineral components of a rock, which generally determine their
petrographic classification, do not necessarily produce the most prominent
spectral features. Even so, dominant spectral patterns produced by minor
mineral components with strong absorption features can significantly assist
in characterising rocks, at least within well constrained geological settings
(Longhi et al. 2001).
A range of multispectral sensors have been used for mapping a variety
of different minerals including the LANDSAT (Carranza and Hale 2002) and
SPOT sensors (Cole 1991; Chica-Olmo et al. 2002). However, given the spatial
resolutions of such orbital sensors, techniques such as SMA are often required.
As imaging spectrometers such as AVIRIS and HyMap provide higher spectral
resolution data, these sensors have frequently been preferred for geological
mapping and mineral exploration although the extent of coverage is often
reduced. As examples, AVIRIS has been used to map mineral assemblages in
areas of the United States (Mackin et al. 1991; Kokaly et al. 1998), including areas
of ammonium in hydrothermal rocks (Baugh et al. 1998), and as a tool in gold
exploration (van der Meer and Bakker 1998; Rowan 2000). Similarly, HyMap
has been utilised to map gold mineralization in Western Australia (Bierwirth
et al. 2002) and for detecting hydrocarbons (Horig et al. 2001). Simulations
have also suggested that different types of metamorphic rocks could be further
discriminated using AVIRIS or MIVIS data. Other applications have included
the measurement of exposed lava and volcanic hotspots (Oppenheimer et al.
1993; Green et al. 1998b), the study of petroleum geology (van der Meer and
de Jong 2001) and the application of AVIRIS to study mineral-induced stress
on vegetation (Merton 1999; Merton and Silver 2000). Recent studies (Kruse
et al. 2003) have also investigated the use of Hyperion for geological mapping
and monitoring the processes that control the occurrence of non-renewable
mineral resources, especially mineral deposits.
The difficulties of applying remote sensing to the mapping of minerals are
illustrated in heavily forested regions, where reflectance based data are less
useful because of the lack of direct observations of the ground surface. Steep
terrain, heavy rainfall and cloud cover as well as high population densities
(and hence changes in land use) in many regions (particularly in the tropics)
also limit exploration using remote sensing data (Cervelle, 1991).
1.6.4
Environmental Hazards and Anthropogenic Activity
Hyperspectral remote sensing offers many opportunities for monitoring natural hazards such as bushfires, volcanic activity and observing anthropogenicinduced activities and impacts such as acidification, land clearing and degradation (de Jong and Epema, 2001), biomass burning (Green et al. 1998a), water
pollution (Bianchi et al. 1995a; Bianchi et al. 1995b), atmospheric fallout of
dust and particulate matter emitted by industry (Ong et al. 2001) and soil
salinity (Metternicht and Zinck 2003). As an example, Chisholm (2001) indicated that spectral indices could be used to detect vegetation water content at