Hyperspectral Sensors and Applications
31
1600 nm but particularly at '" 2080 nm. 02 absorption occurs at 760 nm and
1270 nm. Using this information, estimates of these constituents can be derived using hyperspectral data. As an example, water vapour can be estimated
using the Continuum Interpolated Band Ratio (CIBR) and Narrow/Wide and
Atmospheric Pre-Corrected Differential Absorption (APDA), with these measures utilising differential absorption techniques and applied across specific
atmospheric water vapour absorption features (e. g. 940 and 1130 nm) (Rodger
and Lynch 2001). The EO-l also supported an Atmospheric Corrector for facilitating correction for atmospheric water vapour, thereby optimising retrieval
of surface features. Coarser spatial resolution datasets have also been used to
characterise the atmosphere. For example, MODIS data have been used in the
measurement of cloud optical thickness, cloud top pressures, total precipitable
water and both coarse and fine aerosol contents in the atmosphere (Baum et
al. 2000; Seemann et al. 2003).
For studies of the cryosphere, hyperspectral techniques have been used to
retrieve information relating to the physical and chemical properties of snow
including grain size, fractional cover, impurities, and snow mass liquid water
content (Dozier and Painter 2004). Few airborne studies have been undertaken
in high latitude regions due to the difficulty in acquiring data. Snow products
are also routinely and automatically derived from orbital platforms such as
MODIS, and include snow cover maps from individual scenes to spatial and
temporal composites (Hall et al. 1995; Klein and Barnett 2003).
Within marine and freshwater environments, hyperspectral data have been
used to characterize and map coral reefs and submerged aquatic vegetation
(Kutser et al. 2003; Malthus and Mumby 2003), including sea grasses (Fyfe
2003), and for quantifying sediment loads and water quality (Fraser 1998). At
the land-sea interface, applications have included characterization of aquatic
vegetation (Zacharias et al. 1992), salt marshes (Schmidt and Skidmore 2003;
Silvestri et al. 2003) and mangroves (Held et al. 2003). As an example, Zacharias
et al. (1992) were able to detect submerged kelp beds using airborne hyperspectral data at several spatial resolutions (Fig. 1.2). The benefits of using
hyperspectral CASI data for identifying and mapping different mangrove
species and communities have been highlighted in studies of Kakadu National
Park, northern Australia (Fig. 1.3). When the derived maps were compared in
a time-series with those generated from historical black and white and true
colour stereo aerial photography, changes in both the extent of mangroves
and their contained species and communities were evident (Mitchell 2004).
These changes suggested a long-term problem of saltwater intrusion as a result
of coastal environmental change. The study also emphasized the benefits of
using hyperspectral data at fine « 1) spatial resolution for characterizing and
mapping mangrove communities.
The application of remotely sensed data to the study of coral reefs was proposed in the mid 1980s (Kuchler 1986) although difficulties associated with
wavelength -specific penetration oflight in water, mixed pixels and atmospheric
attenuation caused initial disillusionment amongst some users (Green et al.
1996; Holden and LeDrew 1998). Even so, the science developed dramatically
in the 1990s, due partly to concerns arising from the impacts on coral reefs
31
1600 nm but particularly at '" 2080 nm. 02 absorption occurs at 760 nm and
1270 nm. Using this information, estimates of these constituents can be derived using hyperspectral data. As an example, water vapour can be estimated
using the Continuum Interpolated Band Ratio (CIBR) and Narrow/Wide and
Atmospheric Pre-Corrected Differential Absorption (APDA), with these measures utilising differential absorption techniques and applied across specific
atmospheric water vapour absorption features (e. g. 940 and 1130 nm) (Rodger
and Lynch 2001). The EO-l also supported an Atmospheric Corrector for facilitating correction for atmospheric water vapour, thereby optimising retrieval
of surface features. Coarser spatial resolution datasets have also been used to
characterise the atmosphere. For example, MODIS data have been used in the
measurement of cloud optical thickness, cloud top pressures, total precipitable
water and both coarse and fine aerosol contents in the atmosphere (Baum et
al. 2000; Seemann et al. 2003).
For studies of the cryosphere, hyperspectral techniques have been used to
retrieve information relating to the physical and chemical properties of snow
including grain size, fractional cover, impurities, and snow mass liquid water
content (Dozier and Painter 2004). Few airborne studies have been undertaken
in high latitude regions due to the difficulty in acquiring data. Snow products
are also routinely and automatically derived from orbital platforms such as
MODIS, and include snow cover maps from individual scenes to spatial and
temporal composites (Hall et al. 1995; Klein and Barnett 2003).
Within marine and freshwater environments, hyperspectral data have been
used to characterize and map coral reefs and submerged aquatic vegetation
(Kutser et al. 2003; Malthus and Mumby 2003), including sea grasses (Fyfe
2003), and for quantifying sediment loads and water quality (Fraser 1998). At
the land-sea interface, applications have included characterization of aquatic
vegetation (Zacharias et al. 1992), salt marshes (Schmidt and Skidmore 2003;
Silvestri et al. 2003) and mangroves (Held et al. 2003). As an example, Zacharias
et al. (1992) were able to detect submerged kelp beds using airborne hyperspectral data at several spatial resolutions (Fig. 1.2). The benefits of using
hyperspectral CASI data for identifying and mapping different mangrove
species and communities have been highlighted in studies of Kakadu National
Park, northern Australia (Fig. 1.3). When the derived maps were compared in
a time-series with those generated from historical black and white and true
colour stereo aerial photography, changes in both the extent of mangroves
and their contained species and communities were evident (Mitchell 2004).
These changes suggested a long-term problem of saltwater intrusion as a result
of coastal environmental change. The study also emphasized the benefits of
using hyperspectral data at fine « 1) spatial resolution for characterizing and
mapping mangrove communities.
The application of remotely sensed data to the study of coral reefs was proposed in the mid 1980s (Kuchler 1986) although difficulties associated with
wavelength -specific penetration oflight in water, mixed pixels and atmospheric
attenuation caused initial disillusionment amongst some users (Green et al.
1996; Holden and LeDrew 1998). Even so, the science developed dramatically
in the 1990s, due partly to concerns arising from the impacts on coral reefs