The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
26
measurements. The exploitation of the theoretical data may lead
to Cab retrieval feasibility maps concerning plausible canopy
structural configurations that may potentially take place during
forest aging.
2. METHODOLOGY
2.1 Radiative transfer models: PROFLIGHT
During development stages in natural forest series two
irreversible processes take place during forest aging: 1) Stands
become woodier. This is particularly the case once reaching the
status of ‘old-growth’ where masses of coarse woody debris
propagate throughout the canopy. And: 2) the spatial
arrangements of woody elements and green foliage become
heterogeneously distributed in vertical and horizontal
dimension (Franklin et al. 2002).
To study the perturbing effects of woody elements (NPV) in the
retrieval of Cab content for any canopy plausibly to occur
during development stages on the basis of synthetic data, a leaf
optical properties model (PROSPECT) was coupled with a 3-D
canopy model (FLIGHT). PROSPECT idealizes the leaf as a
pile of elementary plates composed of absorbing and diffusing
constituents. The version of the model used here (Jacquemoud
et al., 2000) is parameterized by chlorophyll concentration, the
dry matter content, leaf water content, and the effective number
of leaf layers. At canopy level, the bidirectional reflectance
factor (BRF) for coniferous canopies was computed by the
Forest LIGHT interaction model (FLIGHT). FLIGHT allows an
explicit representation of complex canopy structures and a
correct treatment of crown overlapping and multiple scattering
within the scene. In the single crown envelopes foliage is
approximated by statistical foliage properties with optical
properties of both leaf (PV) and woody phytoelements (NPV).
The canopy is (lower) bounded by a soil medium with
anisotropic scattering functioning according to Hapke (1981).
The horizontal exchange of rays with neighboring areas is
arranged by cyclic boundary conditions, meaning that laterally
exiting rays of the bounding box are re-cast from the opposite
plane at the same trajectory angle to extend scattering to an
infinitely extended forest. Subsequently, each generated scene
canopy reflectance is the result of a unique, plausible, stand
configuration.
2.2 Study sites
The models are parameterized based on field data from an old-
growth coniferous forest in Swiss National Park (SNP),
Switzerland (10°13'48"E/46°39'45"N). SNP is one of the few
areas in Western Europe that was not influenced by humans
during most of the 20th century. The forest, characterized by its
advanced age (165-200 yr; Cherubini et al., 2002) of the pine
stands (P. montana and P. cembra), is classified as woodland
associations of Erico-Pinetum mugo (Zoller, 1995). The
overstory canopy is typified by relatively open and
discontinuous stands resulting in a relatively low Leaf Area
Index (LAI; between 1.5 and 4.5, Kotz et al., 2004) and a high
fraction of total woody parts (e.g. trunks, branches, standing
and fallen deadwood).
Further, to exemplify the theoretical results into the context of
real-world forests one young and two mature coniferous stands
were additionally selected:
Young Norway spruce: This Norway spruce (Picea abies (L.)
Karst.) stand is located in the Moravian-Silesian Beskydy
Mountains, in the eastern part of the Czech Republic bordering
with Slovakia (49°50'N, 18°54'E). The trees of the
monoculture plantation are currently 30 years old. Foliage is
concentrated in the dense, uniform overstory, characterized by a
mean canopy LAI of 7.8 and a mean canopy cover of 82%.
(Homolova et al., 2007).
Early mature Lodgepole pine: This Lodgepole pine (Pinus
contorta) stand is located in the central interior of British
Columbia, Canada. The ecosystem in this area is dominated by
the Sub-Boreal Spruce (SBS) biogeoclimatic zone. The forest is
located approximately in the centre of the province (124° 18' N,
53° 39'E) and is between 61-80 years old. The region is
attacked by mountain pine beetle since the mid nineties.
Coarse-resolution LAI maps (Chen et al., 2002) indicate that
LAI of this region ranges between 4 and 5. Crown coverage is
visually assessed to range between 60 and 80%.
Mature forest Norway spruce: This Norway spruce stand is
located in the Sumava Mountains, (Bohemian Forest, southern
Czech Republic; 48° 59' N, 13° 28' E). The region holds some
of the best preserved and least human-influenced spruces
forests in Central Europe. The age of the stand is over 100
(±125) years (Wild et al., 2004). An extensive area of Norway
spruce was affected by a massive bark beetle outbreak since the
mid-1990s whereby vast stands lost all their needles (grey-
attack) (Jonasova & Prach, 2004).
2.3 Optical properties
The investigation of confounding factors affecting the
relationship between reflectance and Cab requires the
simulation of many reflectance spectra, at needle-level and
canopy-level. At needle level, it is well known that needle
chlorophyll concentration depends on age of needles (Jach &
Ceulemans, 2000), stress (Carter & Knapp, 2001) and location
within the canopy. Optical properties of needle leaves were
simulated with PROSPECT whereby a variation of Cab content
was explicitly accounted for. Chosen chlorophyll content
ranged from 15 to 95 pg/cm 2 with steps of 10 pg/cm 2 . This is a
range that may typically occur in a conifer stand.The remaining
model variables, leaf mesophyll structure, dry matter and water
content were empirically acquired during the Fire Spread and
Mitigation (SPREAD) campaign as described in Kotz et al.,
(2004) and were subsequently aggregated to obtain values
generic for the SNP study site (N: 3.80, C d : 0.036 g/cm 2 and C w :
0.044 g/cm 2 respectively). Optical properties of the understory
background of SNP can be essentially conceptualized as a
mixture of PV (shrubs, herbaceous species) and NPV (woody
parts, litter) elements; spots of bare soil are virtually absent
within the forest. Field spectroradiometric measurements of
understory vegetation were aggregated (35 vegetated understory
spectra, 10 bark spectra) to approximate the spectral diversity
characterizing the background spectra (figure 3a). Further,
aggregation of the bark spectra equally led to a generic NPV
signature.
From the set of PROSPECT-generated spectra, the average (p)
and associated variation expressed by the standard deviation (cr),
was calculated and shown in figure la. In a sense, the average
of the reflectance series associated with the Cab range
approximates the reflectance values of a needle with an
intermediate chlorophyll concentration (e.g. around 55 pg/cm 2 ).
The figure clearly demonstrates that the influence of
