808 
phenomenon which the reflectance phenomenon indeed is. 
Let us now examine some of the major sources of variation in the 
data and look at some trends that were revealed. A surprising finding was the 
large differences between the two results obtained from remeasuring each site. 
The only possible explanation here is that different parts of tree crowns 
were observed in each case, since the measurements were generally done at 
different times of the day, i.e. at different sun's azimuth angles. The 
small changes in sun's elevation should have not affected the results, because 
all measurements were restricted to within 4° fluctuation of the elevation 
angle from the mean. The problem was further enhanced by the fact that the 
instrument had a fairly narrow angle of view ( 20 °) and was at a close distance 
from the crown canopy. Thus crown canopy variation was probably the greatest 
source of fluctuations in the data. This variation is reflected in high 
standard deviations in the reflectance curves. The average standard deviation 
was approximately ±5% of reflectance. From Fig. 1 it can be seen that this is 
too high a value to differentiate the curves in the visible spectrum and even 
in the infrared portion. 
Further scrutiny of the data suggested a trend toward higher 
reflectance values, principally in the IR region, with denser canopies 
(greater crown closure). Using an analysis of variance it was found that this 
indeed is the case: reflectance increases with increase of foliage (biomass) 
and this increase is significant in the IR region. 
Dependence of reflectance on geometric parameters of the observa 
tional set-up was confirmed. Fig. 2 shows an increase of reflectance with the 
increase of azimuth of the instrument w.r.t. the sun ( 0 ° azimuth designates 
instrument looking directly into the sun, i.e. the shadowy parts of the crown). 
Fig. 3 gives the spectral reflectance averages for the two nadir view angles 
($) used, namely 3 = 10° and 3 = 30°. The result is, as intuitively 
expected: as the nadir view angle increases less of the shadowy parts of the 
crown is included and hence higher reflectance. Both of these trends in 
reflectance variation are evidently the result of crown 3 -dimensionality and 
are easily accepted. 
CONCLUSIONS 
The fact that the described study of tree spectral reflectance 
failed to show any real basis for differentiation among the trees indicates 
that the problem of determination of spectral reflectance of tree species is a 
very complex and difficult one, both from the physical point of view 
(difficult access , 3-dimensionality of crowns, etc.) as well as from the 
consideration of the high inherent variability in biological populations such 
as forest trees that depend on variable site conditions. The problem is fur 
ther complicated by changing metereological conditions during observation. To 
complicate the matter further, spectral reflectance is not a unique quantity 
but depends strongly on geometric parameters (direction of illumination and 
observation). This consideration alone severely limits its usefulness in 
practical applications using large sensor field angles such as wide angle