1. Stand Structure in Terrestrial Ecosystems
Canopy Height
Direct Methods
In herbaceous and shrubland vegetation, canopy
height and stratification can be measured using pin
probes at sample points, which are often arrayed
along line transects or belts (Bonham 1989). Range
scientists have also developed the "swardstick,"
which consists of a plate centered on a graduated
vertical pole that, when properly calibrated, provides rapid estimates of sward height and density
(e.g., Earle and McGowan 1979). In tall forest vegetation, one can access the canopy by ropes or
cranes, trees can be felled to measure height and
foliar profile directly (e.g., Maguire et al. 1996), or
tree height can estimated allometric ally from sapwood area measurements (Husch et al. 1982).
Indirect Methods
In low vegetation, ground cameras have been used
to estimate canopy height both from horizontal profiles and from vertical stereo photographic pairs
(e.g., Ivanov et al. 1994). Active ground sensors
have not seen much use in herbaceous or shrubland
types, although Hutchings et al. (1990) tested the
use of ultrasonic rangefinders to estimate canopy
height in grasslands and found the estimates of
height and mass to be closely related to those obtained by direct swardstick measurements.
Many ground-based methods of tree height estimation have been developed. The time-honored
approach uses hypsometers that measure the angle
to a reference height in the canopy for a known
distance from the base of the tree, and then applies
trigonometric principles to derive tree height
(Husch et al. 1982). Vertical optical rangefinders
can also be used at sample points or along transects
to measure canopy stratification (e.g., Clark et al.
1996). Unfortunately, all of the distance-based approaches depend on obtaining unobstructed views
of canopy elements and are difficult to apply in rugged terrain and dense or multilayered forests.
For over 60 years, aerial photographs have been
used to estimate shrub and forest canopy height
based on either shadow lengths or parallax measurements (Philipson 1997). The use of stereo pairs
depends on obtaining clear views of nearby ground
surface as well as the canopy. Submeter accuracy
of tree heights can be obtained from stereo aerial
19
photographs depending on choice of camera, fiying
height, and equation.
Weltz et al. (1994) tested the use of an airborne
laser profiling system against ground-estimated
height of semiarid vegetation types and found that
the two were in general agreement, although the
waveform analysis method used did not perform as
well in very clumped vegetation or very sparse
shrublands. Nilsson (1996) tested the use of a
helicopter-borne LIDAR system for estimating tree
heights and stand volume in a 12- to 13-m tall pine
plantation and found that mean tree height was sensitive to footprint size and was systematically underestimated by 2.1 to 3.7 m. Means et al. (1997)
have obtained very promising results using aircraft
laser altimetry to estimate tree height, foliar biomass, and total biomass in conifer forests of the
western Cascades of Oregon (Fig. 1.5).
Profiling radar scatterometers have been used
successfully to estimate canopy height. For instance, Hyyppa and Hallikainen (1996) reported
rms errors of 1 m for estimates in 17- to 19-m pine
and spruce stands. Forest height has also been correlated with backscatter data from airborne synthetic aperture radar systems. Working in temperate
forests, Dobson et al. (1995) stratified vegetation
into general structural types based on form, branching pattern, and leaf type, and used calibrated regression equations to estimate height from likepolarized or crosspolarized C-band, L-band, and/or
X-band data, attaining moderate to high accuracy,
depending on the forest type.
In a recent test of the applications of interferometric SAR (INSAR), Treuhaft et al. (1996) estimated ground surface height and canopy depth to
an average accuracy of 5 m over a range of 25 m
using a C-band system. They found that INSAR
systematically overestimated canopy depth and that
the magnitUde of error in canopy depth and ground
surface heights increased with increasing tree
height. Multiple baselines and multifrequency SAR
would be expected to increase accuracy (Hagberg
et al. 1995; Treuhaft et al. 1996).
Vertical Foliar Distribution
Direct Methods
Foliar profiles of herbaceous and shrubland vegetation can be obtained directly by stratified probing
or clipping of the canopy. Ross (1981) and Bonham
Canopy Height
Direct Methods
In herbaceous and shrubland vegetation, canopy
height and stratification can be measured using pin
probes at sample points, which are often arrayed
along line transects or belts (Bonham 1989). Range
scientists have also developed the "swardstick,"
which consists of a plate centered on a graduated
vertical pole that, when properly calibrated, provides rapid estimates of sward height and density
(e.g., Earle and McGowan 1979). In tall forest vegetation, one can access the canopy by ropes or
cranes, trees can be felled to measure height and
foliar profile directly (e.g., Maguire et al. 1996), or
tree height can estimated allometric ally from sapwood area measurements (Husch et al. 1982).
Indirect Methods
In low vegetation, ground cameras have been used
to estimate canopy height both from horizontal profiles and from vertical stereo photographic pairs
(e.g., Ivanov et al. 1994). Active ground sensors
have not seen much use in herbaceous or shrubland
types, although Hutchings et al. (1990) tested the
use of ultrasonic rangefinders to estimate canopy
height in grasslands and found the estimates of
height and mass to be closely related to those obtained by direct swardstick measurements.
Many ground-based methods of tree height estimation have been developed. The time-honored
approach uses hypsometers that measure the angle
to a reference height in the canopy for a known
distance from the base of the tree, and then applies
trigonometric principles to derive tree height
(Husch et al. 1982). Vertical optical rangefinders
can also be used at sample points or along transects
to measure canopy stratification (e.g., Clark et al.
1996). Unfortunately, all of the distance-based approaches depend on obtaining unobstructed views
of canopy elements and are difficult to apply in rugged terrain and dense or multilayered forests.
For over 60 years, aerial photographs have been
used to estimate shrub and forest canopy height
based on either shadow lengths or parallax measurements (Philipson 1997). The use of stereo pairs
depends on obtaining clear views of nearby ground
surface as well as the canopy. Submeter accuracy
of tree heights can be obtained from stereo aerial
19
photographs depending on choice of camera, fiying
height, and equation.
Weltz et al. (1994) tested the use of an airborne
laser profiling system against ground-estimated
height of semiarid vegetation types and found that
the two were in general agreement, although the
waveform analysis method used did not perform as
well in very clumped vegetation or very sparse
shrublands. Nilsson (1996) tested the use of a
helicopter-borne LIDAR system for estimating tree
heights and stand volume in a 12- to 13-m tall pine
plantation and found that mean tree height was sensitive to footprint size and was systematically underestimated by 2.1 to 3.7 m. Means et al. (1997)
have obtained very promising results using aircraft
laser altimetry to estimate tree height, foliar biomass, and total biomass in conifer forests of the
western Cascades of Oregon (Fig. 1.5).
Profiling radar scatterometers have been used
successfully to estimate canopy height. For instance, Hyyppa and Hallikainen (1996) reported
rms errors of 1 m for estimates in 17- to 19-m pine
and spruce stands. Forest height has also been correlated with backscatter data from airborne synthetic aperture radar systems. Working in temperate
forests, Dobson et al. (1995) stratified vegetation
into general structural types based on form, branching pattern, and leaf type, and used calibrated regression equations to estimate height from likepolarized or crosspolarized C-band, L-band, and/or
X-band data, attaining moderate to high accuracy,
depending on the forest type.
In a recent test of the applications of interferometric SAR (INSAR), Treuhaft et al. (1996) estimated ground surface height and canopy depth to
an average accuracy of 5 m over a range of 25 m
using a C-band system. They found that INSAR
systematically overestimated canopy depth and that
the magnitUde of error in canopy depth and ground
surface heights increased with increasing tree
height. Multiple baselines and multifrequency SAR
would be expected to increase accuracy (Hagberg
et al. 1995; Treuhaft et al. 1996).
Vertical Foliar Distribution
Direct Methods
Foliar profiles of herbaceous and shrubland vegetation can be obtained directly by stratified probing
or clipping of the canopy. Ross (1981) and Bonham