591
VEG facilitates the use of diverse knowledge bases to be incorporated into the inference techniques. In this
study, VEG used additional information to make more accurate view angle extension techniques than the
traditional techniques that only use spectral data from the unknown target in a simplistic manner. VEG used
spectral Hata and a normalized difference technique to infer the percent ground cover of the unknown target.
This estimate of percent ground cover of the unknown target along with information on the sun angle, and
wavelengths were then used to search a historical data base for targets that match the unknown target in
these characteristics. These data captured the general shape of the reflectance distribution of the unknown
target. This historical information was used to estimate the coefficients of the techniques for the conditions at
hand and to test the accuracy of the techniques.
The tests used in this study (Kimes et al. 1994) were difficult ones. For example, techniques were tested
that make long angular extensions using one, two, or four input view angles to predict an unknown nadir
value. Furthermore, a wide variety of unknown targets were tested. The errors ( ^proportional rms) obtained
were on the order of 0.15. In addition techniques were tested that use seven or nine multiple view angles to
predict the directional reflectance distribution over the entire hemisphere of an unknown target The accu
racy of these tests were relatively good considering the relatively dynamic and noisy nature of directional
reflectance distributions. The accuracy of the techniques in this study depends on the smoothness of the his
torical reflectance distributions and the amount of historical data available that closely matches the unknown
target.
-»•CONCLUSION
The VEG system provides a workbench supporting remote sensing scientists doing analysis of directional
optical data. VEG encourages creative investigation by accommodating a variety of repeated tests under
different conditions. The system manages complexity providing an appropriate level of abstraction for
scientific investigation. VEG provides a number of features useful to scientists. A simple interface allows the
scientist to add new techniques to VEG. Other interfaces allow data to be input from external files contain
ing scene information, and the results of analysis to be written to external files. Tools are provided for
managing the interface between VEG and the operating system so that historical databases can be used or
added to during sessions. A tool box provides capabilities to browse the system, dynamically plot data, get
help or add text to the help system and print screen dumps.
Currently VEG has been developed to (1) infer spectral hemispherical reflectance from any combination
of nadir and/or off-nadir view angles (2) infer percent ground cover from any combination of nadir and/or
off-nadir view angles (3) infer unknown view angle(s) from known view angle(s) (known as view angle
extension), and (4) discriminate between user-defined vegetation classes using spectral and directional
reflectance relationships developed from an automated learning algorithm. The errors for these techniques
were generally very good ranging between 2 and 15% (proportional rms).
5-REFERENCES
Deering, D.W., 1989, Field Measurements of Bidirectional Reflectance, in Theory and Applications of Opti
cal Remote Sensing, Wiley, New York, 2: 14-65.
Deering, D.W. and E.M. Middleton, 1990, Spectral Bidirectional Reflectance and Effects on Vegetation
Indices for a Prairie Grassland, in Proc. Symposium on FIFE, American Meteor. Soc., Anaheim CA,
Feb. 7-9.
Dickinson, R.E., 1983, Land Surface Processes and Climate-Surface Albedos and Energy Balances, Adv.
Geophys., 25:305-353.
P.R. Harrison, P.A. Harrison, and D.S. Kimes, 1994, Intelligent Workbench for Studying Earth’s Vegetation,
J. Expert Systems with Applications, (Submitted).
Kimes, D.S., 1983, Dynamics of Directional Reflectance Factor Distributions for Vegetation Canopies,
Applied Optics, 22:1364-1372.
