NIFER
‘cer
on of crown
ulk density,
havior over
lote sensing
rest. In this
information
State, USA.
d-based fuel
canopy fuel
the intensity
ied region.
backscatter
Jussin et al,
n limits for
s/ha) are too
rests (~300
netric radar
significantly
est. structure
in phase, or
rom slightly
ting to the
is used in
1e the three-
berg et al,
y will allow
ture data -
measure the
length will
eatures. The
-band) and
nopy height
n of canopy
uel density)
shown that
1 phase, arc
) to forest
ige of forest
of IFSAR
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
therefore reduces the effect of radar saturation and provides for
more accurate measurement of forest structure parameters over
a wide range of forest densities.
The use of multifrequency (X-band and P-band) IFSAR
systems for forest mapping has emerged recently, where
research efforts have largely focused on improving forest type
classification (Hofmann et al., 1999; Dutra et al., 2002; Mura et
al., 2001). Previous studies have shown that IFSAR can be used
to estimate biomass and other forest inventory parameters
(Andersen et al., 2004; Mette et al, 2003; Treuhaft and
Siqueira, 2004). In this paper, we present and evaluate an
approach to estimating several critical canopy fuel metrics,
including canopy fuel weight, canopy bulk density, canopy base
height, and canopy height, using polarimetric, dual-frequency
[FSAR data in a Pacific Northwest conifer forest.
2. STUDY AREA AND DATA
2.1 Study area
The study area for this investigation was a 5.2 km? area within
Capitol State Forest, western Washington State, USA. This
forest is primarily composed of coniferous Douglas-fir
(Pseudotsuga menziesii) and western hemlock (Tsuga
heterophylla) and, to a lesser degree, hardwoods such red alder
(Alnus rubra) and maple (Acer spp.). The extent of the study
area is shown in an orthophoto in Figure 1.
1999 orthophoto of Capitol Forest study area
(courtesy Washington State Department of Natural Resources)
Figure 1.
This site is the study area for an ongoing experimental
silvicultural trial, and contains coniferous commercial forest
stands of varying age and density. An extensive topographic
survey was conducted throughout the area to enable rigorous
evaluation of a variety of technologies relevant to precision
forest management, including high-resolution remote sensing
and terrestrial geopositioning systems.
2.2 IFSAR Data
IFSAR data were acquired over the Capitol State Forest study
area in September, 2002, using the TOPOSAR System
(formerly AeS-1, developed by Aerosensing Radarsysteme,
GmbH, and now owned, operated and further developed by
Intermap Technologies Corp.). This two-frequency, multi-
polarization system provides X-band interferometric data in a
single-pass mode and P-band interferometric data in a repeat-
pass mode. X-band and P-band data are not collected
simultaneously owing to recording bandwidth limitations. The
salient parameters of the systems are shown in Table 1.
Table 1. TOPOSAR System Parameters.
Parameter X- P-band
band
Mode (Non-Simultaneous) | Single Repeat-Pass
-Pass
Antenna Baseline (m) 2.4 As Desired (50, 80)
Centre Wavelength (cm) 3 74
Image Resolution (m) Zh 25
0.5
Swath Width (km) 2 4,7 4
Polarization HH HH or
(HH,VV,HV/VH)
Typical Flying Altitude 5,000 5,000
(m)
As shown in Table 1, several operational configurations are
possible. The X-band configuration selected for this project
included 2 km swath, 1 meter resolution imagery and digital
surface model (DSM) sample spacing. There were 4 strips of
X-band acquired, including two partially overlapping swaths
from each of two opposite viewing directions. In the case of the
P-band, the 2.5 meter data were acquired with overlapping
swaths from four orthogonal viewing directions. The vendor
also provided a digital elevation model developed from an
optimized integration of the polarimetric P-band data. A
composite digital terrain model (DTM) was created from the X-
band (in open areas) and P-band (beneath the canopy) data.
The root-mean-square error (RMSE) of this DTM, based upon a
comparison to 350 high-accuracy topographic survey points, (of
which 270 were located in uncut or lightly thinned forest) was
2.59 meters (Mercer et al., 2003). The DTM sample spacing
was 2.5 meters, although a smoothing function reduced the
effective independent spacing width by several meters.
Additional data provided by Intermap Technologies Corp.
included multi-polarization SAR backscatter orthoimages, look-
angle images, coherence images and phase information. The
DTM is shown in Figure 2a, and the X-Band DSM - effectively
a canopy surface model in this project — is shown in Figure 2b.
The X-band elevation strip data were merged into a single file
for the purposes of the analysis. X-band elevation data overlaid
on the P-band digital terrain model in a selected area containing
both mature and young forest stands are shown in Figure 3a.