38
1: Richard lucas, Aled Rowlands, Olaf Niemann, Ray Merton
1.6.3
Soils and Geology
Soils are complex and typically heterogeneous, so their properties therefore
cannot be assessed easily and directly as a function of their composite spectral reflectance profiles, even under controlled laboratory conditions (Ben -Dor
and Banin 1994). Furthermore, the entire vertical soil profile cannot be imaged using multispectral or hyperspectral sensors and inferences have to be
made about the soil based on the surface veneer. In most cases, vegetation
(dry or green) obscures the soil surface either partly or completely. Despite
these limitations, certain variables associated within soils have been quantified using remote sensing data. As examples, hyperspectral remote sensing
(e. g., AVIRIS) has shown promise in retrieving soil moisture content, temperature, texture/surface roughness, iron content and soil organic carbon (SOC)
(Palacios-Orueta and Ustin 1998; Ahn et al. 1999). Using reference reflectance
spectra from soil samples and established procedures, the relationships of SOC
to moisture, exchangeable calcium and magnesium cation exchange capacity
and also soil colour were able to be quantified (Schreier et al. 1988). Many of
these variables actively influence the response across the reflectance region
and, for this reason, only a small selection of spectral channels is required for
their quantification (Ben-Dor et al. 2002).
Colour is also a key attribute as it allows characterisation, differentiation and
ultimately classification of soil types. In general, colour is determined by the
amount and state of iron and/or organic matter (OM) contained, with darker
soils tending to be more indicative of higher OM contents, although manganese
can also produce dark colouration. The colour of soils with low OM is often
dominated by pigmenting agents such as secondary iron oxides. Comparisons
with airborne MIVIS visible data acquired in farmed areas of Italy suggested
that soil colour could be retrieved, allowing estimation of both chroma and hue,
although mapping across the image was considered difficult. As many mineral
components of rocks and soils have distinct spectral signatures, hyperspectral
data have also played a key role in identifying and mapping expansive soils
(Chabrillat et al. 2002).
Geological applications have been the principal driving force behind much
of the early development phases of imaging spectrometry. The development
can be attributed partly to the recognition that a wide range of rock forming minerals have distinct spectral signatures (Curran 1994; Escadafel 1994;
Goetz 1995). Following the work of Hunt (1989), numerous publications have
provided detailed laboratory spectra of rocks and minerals and their mixtures as well as accurate analyses of absorption features, many of which have
been obtained from laboratory measurements. These serve as references for
spectra obtained using ground radiometers or airborne/spaceborne sensors,
with spectroscopic criteria applied to identify and map minerals and alteration
zones (Hunt 1989; Christensen et al. 2000; Longhi et al. 2001).
Despite the availability of suitable reference sources, their use in rock-type
matching has been complicated by several interacting variables including original parent mineral associations, diagenesis, transformation/alteration through
1: Richard lucas, Aled Rowlands, Olaf Niemann, Ray Merton
1.6.3
Soils and Geology
Soils are complex and typically heterogeneous, so their properties therefore
cannot be assessed easily and directly as a function of their composite spectral reflectance profiles, even under controlled laboratory conditions (Ben -Dor
and Banin 1994). Furthermore, the entire vertical soil profile cannot be imaged using multispectral or hyperspectral sensors and inferences have to be
made about the soil based on the surface veneer. In most cases, vegetation
(dry or green) obscures the soil surface either partly or completely. Despite
these limitations, certain variables associated within soils have been quantified using remote sensing data. As examples, hyperspectral remote sensing
(e. g., AVIRIS) has shown promise in retrieving soil moisture content, temperature, texture/surface roughness, iron content and soil organic carbon (SOC)
(Palacios-Orueta and Ustin 1998; Ahn et al. 1999). Using reference reflectance
spectra from soil samples and established procedures, the relationships of SOC
to moisture, exchangeable calcium and magnesium cation exchange capacity
and also soil colour were able to be quantified (Schreier et al. 1988). Many of
these variables actively influence the response across the reflectance region
and, for this reason, only a small selection of spectral channels is required for
their quantification (Ben-Dor et al. 2002).
Colour is also a key attribute as it allows characterisation, differentiation and
ultimately classification of soil types. In general, colour is determined by the
amount and state of iron and/or organic matter (OM) contained, with darker
soils tending to be more indicative of higher OM contents, although manganese
can also produce dark colouration. The colour of soils with low OM is often
dominated by pigmenting agents such as secondary iron oxides. Comparisons
with airborne MIVIS visible data acquired in farmed areas of Italy suggested
that soil colour could be retrieved, allowing estimation of both chroma and hue,
although mapping across the image was considered difficult. As many mineral
components of rocks and soils have distinct spectral signatures, hyperspectral
data have also played a key role in identifying and mapping expansive soils
(Chabrillat et al. 2002).
Geological applications have been the principal driving force behind much
of the early development phases of imaging spectrometry. The development
can be attributed partly to the recognition that a wide range of rock forming minerals have distinct spectral signatures (Curran 1994; Escadafel 1994;
Goetz 1995). Following the work of Hunt (1989), numerous publications have
provided detailed laboratory spectra of rocks and minerals and their mixtures as well as accurate analyses of absorption features, many of which have
been obtained from laboratory measurements. These serve as references for
spectra obtained using ground radiometers or airborne/spaceborne sensors,
with spectroscopic criteria applied to identify and map minerals and alteration
zones (Hunt 1989; Christensen et al. 2000; Longhi et al. 2001).
Despite the availability of suitable reference sources, their use in rock-type
matching has been complicated by several interacting variables including original parent mineral associations, diagenesis, transformation/alteration through