479
Figures 3 and 4 show the development of leaf re
flectance from the early beginning in May when the
buds are broken and the leaves are light-green.
With the increasing pigment concentrations from May
15 to June 1 the reflectance spectrum changes dra
matically. In the blue region of the spectrum the
flattening of the reflectance spectrum is observed
due to the saturation of the absorption. The chloro-
plasts are no more transparent in the blue absorbing
all incoming light. In the red part of the spectrum a
linear decrease of the reflectance is seen associated
with the shift of the red edge towards longer wave
lengths. When the leaf is dark green at July 1 the
reflectance spectrum does not change remarkable in
the blue. In contrast the green to red domain of the
reflectance decreases. At about 680 nm, the maxi
mum of the chlorophyll a absorption, the reflectance
flattens due to saturation and absorption line broa
dening resulting in an additional shift of the red
edge.
Figure 3: Saisonal cycle of the reflectance spectra of
the model leaf
1: reflectance, when buds are broken, May, 15
2: reflectance at June,l
3 reflectance at September, 1 4
Figure 4: Saisonal cycle of the reflectance spectra of
the model leaf
4. reflectance at September, 15
5: reflectance at October, 1
During summer, the pigment concentration remains
relativ constant. As seen in figure 4, the reflectance
spectrum for September 15 is comparable with the
spectrum from July 1. During the next two weeks,
the chlorophyll concentration breaks down accompa
nied by an increase of a new pigment, the anthocya-
nin. This pigment occurs during the autumn and is
localized in the vacuoles of the epidermis. Due to
the strong absorption of anthocyanin in the 500nm
to 600nm region the green reflectance decreases
while the reflectance in the domain of the red edge
increases leading to the typical red coloring of lea
ves during autumn.
A comparison with the measured reflectance spectra
during autumnal leaf coloring (Boyer et al. 1988)
shows good agreement with our calculated spectra.
The only major difference is seen again in the near-
infrared region due to the constancy of the scatte
ring coefficient used in the model.
4 CONCLUSION
Based on the experimental findings and the different
interpretations of several investigators (Boyer et al.
1988, Buschmann and Lichtenthaler 1988, Collins et
al. 1983, Horler et al. 1980, Horler et al. 1983, Rock
et al. 1988, Schutt et al. 1984, Singhroy et al. 1985),
that the blue shift of the red edge of the reflectan
ce spectrum of vegetation may be correlated with
the vegetative chlorophyll status, the species, the
developmental stage, the leaf surface exposed to the
sensor, the water content and the chlorophyll fluo
rescence, the stochastic model of Tucker and Garratt
(1977) was reinvestigated in order to find a theoreti
cal, bio-optical approach for a comprehensive inter
pretation of all the different explanations.
In a revised version the stochastic leaf model de
monstrated its potential to calculated the reflectan
ce spectrum of vegetation. Optical, geometrical and
physiological parameters as e.g. the specific absorp
tion coefficient of chlorophyll a,, the pigment con
centration and the leaf thickness are used as input
parameters. The revised model takes into account
different mechanisms as absorption, reflection and
Mie scattering at the chloroplasts and the leaf cells.
According to our first results showing good agree
ment with data from the literature, the model pre
sents two ways for interpreting the blue shift. First,
by reducing the light absorption in the leaf, a blue
shift of the red edge can be produced because the
absorption line in the red part of the spectrum
narrows. In turn a reduced light absorption can be
modeled twofold by reducing the pigment concen
tration of the leaf or by reducing the thickness of
the leaf assuming constancy of the pigment concen
tration. Thinner leaves have in general a shorter
light path resulting in an overall less absorption.
