142 
The classifiers used in this study demonstrated some differences when used in SW and BISE algorithms, 
suggesting the choice of compositing classifier is of concern. But on the other hand, the temporal profiles of different 
classifiers showed some similarities in seasonal variations with vegetation, indicating the choice of the classifier may 
not be critical if the only qualitative studies involved. More differences were found between compositing algorithms 
than between classifiers, which suggested that the choice of classifier is less important than the choice of composing 
algorithms, although a good classifier would certainly increase the liability and the meaning of composited remote 
sensing products. 
In conclusion, the SW showed substantial improvement in composi ting multitemporal AVHRR data by ret aining 
more valuable data while minimizing the high frequency noise. For low vegetation covered earth surfaces, the MSAVI 
and SAVI appeared to be the better classifiers, while for high vegetation densities any of the classifiers (tested in this 
study) can be used in compositing. It is more important, however, to choose the appropriate composing algorithms than 
to choose their classifiers. It should be pointed out, however, that any compositing algorithm can only produces, from 
whatever data it is given, a subset that the algorithm ’thinks’ it is the best. Other errors such as those due to geometric 
registration would most likely remain after compositing. 
Acknowledgements-. The authors are grateful to the USDA ARS Water Conservation Laboratory in Phoenix for financial 
support and Southwest Watershed Research Center in Tucson for providing very convenient working environment. The 
author was also grateful to Cabot F. and LERTS for providing the AVHRR data. This work is also a part of the NASA 
Interdisciplinary Research Program in Earth Sciences (NASA Reference Number IDP-88-086) at the University of 
Arizona (USA) and LERTS (Toulouse, France). 
REFERENCES 
DTorio, M.A., Cihlar, J., and Mullins, D.W. 1990, AVHRR compositing using NDVI and thermal channel data: 
Proceedings of IGARSS’ 90, Washington, D.C., p. 257-260. 
DTorio, M.A.., Cihlar, J., and Morasse. CR., 1991, Effect of the calibration of AVHRR data on the normalized 
difference vegetation index and compositing: Canadian Journal of Remote Sensing, v. 17, no. 3, p. 251 - 262. 
Goward, S.N.. Dye, D.G., Dulaney, W. and Yang, J., 1990, Critical assessment of the NOAA global vegetation 
index data product: Proceeding of the Workshop on the Use of Satellite-Derived Vegetation Index in Weather 
and Climate Prediction Models , Campus Springs, Maryland Feb. 26-27, 1990. p. 34-42. 
Gutman, G., 1991, Vegetation indices from AVHRR: An update and future prospects: Remote Sensing of 
Environment, v.35, p.121-136. 
Holben, B.N., 1986, Characteristics of maximum-value composite images from temporal AVHRR data. 
International Journal of Remote Sensing: v. 7, p.1417-1434. 
Huete, A.R., 1988, A soil-adjusted vegetation index (SAVI): Remote Sensing of Environment: v. 25. p.295-309. 
Huete, AJR., 1989, Soil influences in remotely sensed vegetation-canopy spectra, in (Asrar, G. (ed.)) Theory 
and Applications of Optical Remote Sensing: p. 107-141. 
Kaufman. Y.J., 1989, The atmospheric effect on remote sensing and its correction, in (Asrar, G. (ed.)) Theory 
and Applications of Optical Remote Sensing: p.336-428. 
Pinty B. and M. M. Verstraete,1992, GEMI: a non-linear index to monitor global vegetation from satellites, 
Vegtatio, vol.101, pp. 15-20. 
Qi J., Y. Kerr, and A. Chehbouni, 1994a, External facta - consideration in vegetation index development, this 
conference. 
Qi, J., Chehbouni, A„ Huete, AR., and Kerr, Y.,A, 1994b, A modified soil adjusted vegetation index 
(MSAVI): Remote Sensing of Environment, in press. 
Viovy, N.. Arino, O. and Belward, A.S., 1992, The best index slope extraction (BISE): A method fa - reducing 
noise in NDVI time-series: International Journal of Remote Sensing: v.13, no. 8 , p.1585-1590.