and instrument and the azimuth angle of the instrument w.r.t. the sun. To 
compare signatures of different trees it is thus necessary to keep the above 
parameters the same or, as in our case, to take averages of measurements over 
the same set of parameter values. We chose the second alternative because we 
wanted to study parametric dependence of spectral reflectance. 
INSTRUMENTATION AND FIELD WORK 
Two simple spectroradiometers and a reflectance standard were 
used, namely: An ISCO Spectroradiometer with spectral range from 380nm to 
1350nm and an IL-681 Spectroradiometer with nominal range 400 - llOOnm. The 
ISCO had a manual scan and was used for monitoring daylight irradiance. The 
IL instrument came equipped with automatic scanning and employed an X,Y 
recorder for the data. This instrument was used for the measurement of crown 
reflectances by elevating it above the tree tops with the aid of a mast. The 
field angle of view was 20°. Spectral standard made from KODAK White Paint 
was used to establish cross-calibration between the two instruments by 
measuring simultaneously the 100% reflectance standard with the IL and 
observing daylight irradiance by the ISCO spectroradiometers. In this way a 
transformation curve was derived by means of which the ISCO measurements were 
converted into the equivalent IL measurements. 
On each tree site were taken 14 measurements, consisting of 7 azi 
muthal directions (0°, 30°, 60°, 90°, 120°, 150°, 180°) each taken from two 
nadir view angles (10°, 30°). The average of these 14 measurements led to 
the determination of a set spectral reflectance for the species. Each site was 
visited twice. The average of the two set spectral reflectances determined 
from the two different times of measurement constituted site spectral 
reflectance for the species. Three tree species were measured in two sites 
each in a different locality. The average of the two sites constituted species 
spectral reflectance. For the remaining two species the site reflectance was 
accepted as the species reflectance. Enough statistical character was thus 
built into the data to provide opportunity for statistical comparison of 
signatures. 
RESULTS 
Field spectral traces were digitized in 25 nonometer intervals and 
computer processed to yield, finally, specie spectral reflectances shown in 
Fig. 1. It was found that the traces were reliable only in the range between 
500 - 925nm wavelength because of poor sensitivity of the detector in the IL 
instrument outside of that range. The question arising immediately is, of 
course: Are the curves sufficiently different from each other to enable 
differentiation among tree species on their basis? After a statistical 
analysis that involved examination of variability in redundant data the conclu 
sion is: There is n£ statistical evidence to suggest that the curves are 
indeed different. This is a "negative" result and normally such results are 
not published. However, this author believes that it is valid to publish 
negative results if these reflectan honest attempt to understand a complex