Classification and mapping of ecological zones on a global scale has been a topic of research for many years. This research looks at the development of a global spatial database of ecological zones for the FRA 2000 Report of the United Nations FAO. Besides evaluating the most appropriate type of classification scheme for this purpose, it explores and demonstrates how existing data, for the United States and Canada, can be reclassified to match the FAO classification scheme. Accuracy of mapping is a synergistic function of error, uncertainty, and quality. An assessment of the draft FAO Level D Ecological Zone map was performed which classifies 10-year average, bi-monthly, smoothed AVHRR-NDVI composites of the conterminous United States by applying linear discriminant and decision tree analyses. The results of the linear discriminant analysis were more significantly correlated to the FAO classes, although both approaches suggest that the classification scheme does maximize between-class variance of the NDVI temporal series.

No description available.