parameters, (2) structural parameters, and (3) geometrical and illumination
parameters.
The structure of a vegetation canopy, and the view and solar angles do not
depend on the spectral channel used. At the same time, leaf optical properties,
soil reflectance and the ratio of direct to total irradiance vary significantly
with wavelength. Using the Nilson-Kuusk model or the FCR model for canopy
reflectance calculations we need a common set of structural parameters for all
spectral channels, and a set of optical parameters for every spectral channel. The
same is true for the inversion problem: given the measured canopy reflectance in
several spectral regions we have to determine a set of structural parameters, and
a set of optical parameters corresponding to each spectral channel. To overcome
this complication the leaf optical model PROSPECT by Jacquemoud and Baret (1990),
the soil reflectance presentation with basis functions by Price (1990), and the
wavelength dependence of sky radiation given by McCartney (1978) will be integrated
into Kuusk's fast canopy reflectance model.
2. MODEL COMPONENTS
2.1. Leaf Optical Properties
Jacquemoud and Baret (1990) proposed a model of leaf optical properties spectra
(the PROSPECT model), which has only three input parameters and describes optical
properties of green leaves in the 400 to 2500 nm spectral range. The spectral
resolution of the model is 4 nm in the visible and 17 nm in the NIR spectral
region. The input parameters are a) the pigment concentration C representing
chlorophyll A and B, and carotenoids, b) the water equivalent thickness C^, and c)
a structure parameter N - the effective number of elementary layers inside a leaf.
As an output the PROSPECT model gives leaf diffuse reflectance and transmittance
spectra.
2.2. Soil Spectral Reflectance
The spectral distribution of soil reflectance depends on a very large number of
soil parameters, see (Biehl et al., 1982). However, Price (1990) succeeded in
describing the whole variety of soil spectra using only four basis functions. These
basis functions permit the reduction of the number of soil parameters to four in
a multispectral canopy reflectance model. Even more, in (Price, 1990) we see that
the weights of first two basis functions are significantly correlated, and two
first functions describe 94% of the soil spectrum variance. In the case of dense
homogeneous canopies (ground cover over 70 - 80% ) the soil is screened by above
vegetation and its influence on the canopy reflectance is small. Then we can
approximate the soil reflectance as
Psoil^) s S1
where
S 2 = a + b s-| , (2)
q> l (A) and
The angular distribution of soil reflectance is nonlambertian, similar to
Nilson and Kuusk (1989) and Kuusk (1993).
692
