1. Stand Structure in Terrestrial Ecosystems
Airborne video systems provide a relatively inexpensive source of high-resolution color or multispectral imagery (reviewed in King 1995). Multispectral imagery can be obtained by using
multiple cameras with selective light filters (e.g.,
Everitt et al. 1995) or a single camera with either a
rotating filter wheel or a beam splitter. Spectral
bandwidths as narrow as 10 to 25 nm can be measured at visible and NIR wavelengths. Commercial
airborne systems typically acquire multispectral
imagery with a frame size of 2 X 1.4 Ian and an
effective spatial resolution of 1 to 3 m. In addition
to relatively high spatial resolution, airborne video
systems have the advantage over satellite systems
that the user can select the spectral bands, bandwidths, target location, date, and time. On the other
hand, band-to-band registration and geometrical
rectification"of the imagery can be difficult. In this
decade, videography has been made even more useful by the integration of GPSs, development of radiometric calibration capabilities, and evolution of
digital camera sensors, which provide full-frame
images instead of interlaced paired frames produced by video systems (King 1995).
Satellite-Borne Multispectral Scanners
Passive optical systems exploit the distinctive spectral properties of leaves and nonphotosynthetic
structural components, such as branches and stems,
to derive information related to light absorption and
vegetation structure. The types of materials present,
their relative abundance, and orientation influence
the spectral quality of visible and NIR radiation
scattered absorbed and reflected by vegetation.
Most notably, canopy reflectance is a product of
leaf optical properties, which include strong absorptance at less than 0.7 ).lm and high NIR scattering between 0.7 and 1.3 ).lm (Fig. 1.2). Changes
in the relative proportion of reflected NIR and red
radiation can be used to infer absorbed PAR, canopy cover, and LAI among other structural parameters (see Chapter 3). Leaf orientation and leaf area
distribution modify these relationships by changing
the light-scattering regime and changing the relative proportion of shadowed and unshadowed
leaves. The relative proportion of nonphotosynthetic components and exposed soil modify the reflected signal, in general causing an increase in red
reflectance, decrease in NIR (0.7 to 1.3 ).lm), and
13
increase beyond 1.5 ).lm (see Fig. 1.2). At the scale
of individual plants, crown geometry interacts with
solar energy to modify the types of shadows cast
by plants and the relative proportion of sunlit and
shadowed crown. At the stand level, the density,
types, and variation in crown height interact as a
source of variance in canopy reflectance and shadowing across the landscape. For example, oldgrowth coniferous forest is distinguished from
other forest and landcover types by high withinstand heterogeneity relative to second growth, and
lower reflectance due to extensive shadowing (Fig.
1.3). In general, little floristic (taxonomic) information can be obtained from most systems, but
broad functional vegetation types can be mapped
(Graetz 1990). In contrast to broadband systems,
recent advances using imaging spectrometers have
shown promise for mapping select plant species in
some ecosystems (Roberts et al. 1998a).
Since the advent of the LANDSAT program of
orbiting multispectral scanners in 1972, much research has been devoted to retrieval of canopy
cover, LA!, and canopy condition using digital multispectral imagery acquired from satellite platforms.
Spaceborne remotely sensed data have an advantage over standard aerial photography in that they
sample wavelengths that cannot be measured using
photographic film and provide a standardized digital product, long-term and rapid-repeat coverage,
and regional to global data sets. The primary limiting factor for spaceborne data is spatial resolution,
which ranges from 10 m for SPOT panchromatic
to as large as 8 Ian (AVHRR GAC data). A secondary limitation is the wavelengths that are sampled. For example, A VHRR samples only two
wavelengths in the visiblelNIR, SPOT samples
three, whereas LANDSAT Thematic Mapper (TM)
samples seven channels, three in the visible, one in
the NlR, two in the shortwave infrared, and one in
the thermal. Limitations in spatial and spectral sampling, combined with the spectral and spatial attributes of vegetation, interact to influence the types of
information that can be derived remotely. Common
airborne sensors include NS001, Daedalus,
AVIRIS, CASI, ADAR, ASAR, and MAS, representing a range in spatial resolution to as low as
1 m (ADAR) and spectral resolution up to 224
bands (AVIRIS). Most recent research has focused
on the use of LANDSAT TM data at a spatial resolution of 30 m.
Airborne video systems provide a relatively inexpensive source of high-resolution color or multispectral imagery (reviewed in King 1995). Multispectral imagery can be obtained by using
multiple cameras with selective light filters (e.g.,
Everitt et al. 1995) or a single camera with either a
rotating filter wheel or a beam splitter. Spectral
bandwidths as narrow as 10 to 25 nm can be measured at visible and NIR wavelengths. Commercial
airborne systems typically acquire multispectral
imagery with a frame size of 2 X 1.4 Ian and an
effective spatial resolution of 1 to 3 m. In addition
to relatively high spatial resolution, airborne video
systems have the advantage over satellite systems
that the user can select the spectral bands, bandwidths, target location, date, and time. On the other
hand, band-to-band registration and geometrical
rectification"of the imagery can be difficult. In this
decade, videography has been made even more useful by the integration of GPSs, development of radiometric calibration capabilities, and evolution of
digital camera sensors, which provide full-frame
images instead of interlaced paired frames produced by video systems (King 1995).
Satellite-Borne Multispectral Scanners
Passive optical systems exploit the distinctive spectral properties of leaves and nonphotosynthetic
structural components, such as branches and stems,
to derive information related to light absorption and
vegetation structure. The types of materials present,
their relative abundance, and orientation influence
the spectral quality of visible and NIR radiation
scattered absorbed and reflected by vegetation.
Most notably, canopy reflectance is a product of
leaf optical properties, which include strong absorptance at less than 0.7 ).lm and high NIR scattering between 0.7 and 1.3 ).lm (Fig. 1.2). Changes
in the relative proportion of reflected NIR and red
radiation can be used to infer absorbed PAR, canopy cover, and LAI among other structural parameters (see Chapter 3). Leaf orientation and leaf area
distribution modify these relationships by changing
the light-scattering regime and changing the relative proportion of shadowed and unshadowed
leaves. The relative proportion of nonphotosynthetic components and exposed soil modify the reflected signal, in general causing an increase in red
reflectance, decrease in NIR (0.7 to 1.3 ).lm), and
13
increase beyond 1.5 ).lm (see Fig. 1.2). At the scale
of individual plants, crown geometry interacts with
solar energy to modify the types of shadows cast
by plants and the relative proportion of sunlit and
shadowed crown. At the stand level, the density,
types, and variation in crown height interact as a
source of variance in canopy reflectance and shadowing across the landscape. For example, oldgrowth coniferous forest is distinguished from
other forest and landcover types by high withinstand heterogeneity relative to second growth, and
lower reflectance due to extensive shadowing (Fig.
1.3). In general, little floristic (taxonomic) information can be obtained from most systems, but
broad functional vegetation types can be mapped
(Graetz 1990). In contrast to broadband systems,
recent advances using imaging spectrometers have
shown promise for mapping select plant species in
some ecosystems (Roberts et al. 1998a).
Since the advent of the LANDSAT program of
orbiting multispectral scanners in 1972, much research has been devoted to retrieval of canopy
cover, LA!, and canopy condition using digital multispectral imagery acquired from satellite platforms.
Spaceborne remotely sensed data have an advantage over standard aerial photography in that they
sample wavelengths that cannot be measured using
photographic film and provide a standardized digital product, long-term and rapid-repeat coverage,
and regional to global data sets. The primary limiting factor for spaceborne data is spatial resolution,
which ranges from 10 m for SPOT panchromatic
to as large as 8 Ian (AVHRR GAC data). A secondary limitation is the wavelengths that are sampled. For example, A VHRR samples only two
wavelengths in the visiblelNIR, SPOT samples
three, whereas LANDSAT Thematic Mapper (TM)
samples seven channels, three in the visible, one in
the NlR, two in the shortwave infrared, and one in
the thermal. Limitations in spatial and spectral sampling, combined with the spectral and spatial attributes of vegetation, interact to influence the types of
information that can be derived remotely. Common
airborne sensors include NS001, Daedalus,
AVIRIS, CASI, ADAR, ASAR, and MAS, representing a range in spatial resolution to as low as
1 m (ADAR) and spectral resolution up to 224
bands (AVIRIS). Most recent research has focused
on the use of LANDSAT TM data at a spatial resolution of 30 m.