Hyperspectral Sensors and Applications
21
a
b
c
d
Fig.1.la-d. Observations of the same area of mixed species subtropical woodlands near
Injune, central Queensland, Australia, observed using a stereo colour aerial photography,
b CAS!, c HyMap, and d Hyperion data at spatial resolutions of < I m, 1 m, 2.8 m and 30 m
respectively. For a colored version of this figure, see the end of the book
data, tree crowns are still resolved and species can be differentiated although
within-canopy components cannot be isolated and pixels containing a mix of
ground and vegetation spectra are common. Even so, relatively pure spectra
relating to ground (e. g., bare soil) and vegetation (e. g., leaf components) can
be extracted. At ~ 2.5-5 m spatial resolution, which is typical to the HyMap
sensor, only larger tree crowns are discernible and, due to the openness of the
woodland canopy, most pixels contain a mix of ground and vegetation spectra.
At coarser (~ 5- > 20 m) spatial resolutions, tree crowns cannot be differentiated and the averaging of the signal is such that pure spectral signatures
for specific surfaces cannot be extracted from the imagery, and only broad
vegetation and surface categories can be distinguished. Sensors observing at
resolutions from 20 to 30 m include the airborne AVIRIS (Vane et al. 1993),
DAIS-7915 (Ben-Dor et al. 2002) and MODIS-ASTER simulator (MASTER) as
well as the spaceborne Hyperion. These observations illustrate the difficulties associated with obtaining "pure" spectral reflectance data from specific
surfaces and materials, particularly in complex environments.
21
a
b
c
d
Fig.1.la-d. Observations of the same area of mixed species subtropical woodlands near
Injune, central Queensland, Australia, observed using a stereo colour aerial photography,
b CAS!, c HyMap, and d Hyperion data at spatial resolutions of < I m, 1 m, 2.8 m and 30 m
respectively. For a colored version of this figure, see the end of the book
data, tree crowns are still resolved and species can be differentiated although
within-canopy components cannot be isolated and pixels containing a mix of
ground and vegetation spectra are common. Even so, relatively pure spectra
relating to ground (e. g., bare soil) and vegetation (e. g., leaf components) can
be extracted. At ~ 2.5-5 m spatial resolution, which is typical to the HyMap
sensor, only larger tree crowns are discernible and, due to the openness of the
woodland canopy, most pixels contain a mix of ground and vegetation spectra.
At coarser (~ 5- > 20 m) spatial resolutions, tree crowns cannot be differentiated and the averaging of the signal is such that pure spectral signatures
for specific surfaces cannot be extracted from the imagery, and only broad
vegetation and surface categories can be distinguished. Sensors observing at
resolutions from 20 to 30 m include the airborne AVIRIS (Vane et al. 1993),
DAIS-7915 (Ben-Dor et al. 2002) and MODIS-ASTER simulator (MASTER) as
well as the spaceborne Hyperion. These observations illustrate the difficulties associated with obtaining "pure" spectral reflectance data from specific
surfaces and materials, particularly in complex environments.