ABSTRACT
Traditional approaches used by ecologists to examine, understand and map patterns of natural or semi-natural vegetation have frequently been constrained by spatial, temporal and economic problems. These problems have been particularly serious in studies of successional changes where long time-scales may be involved. Some changes in upland vegetation, for example, may take decades before they are detectable (Ball et al., 1982). However, a variety of remote-sensing techniques offer the ecologist a potential means to overcome such problems. In particular, extensive and frequent use has been made of multi-temporal panchromatic aerial photographs, at a variety of scales, to classify and map vegetation (Fuller, 1981). More recently, satellite data and imagery have been used to identify upland vegetation types (Hume et al., 1986; McMorrow and Hume 1986; Morton, 1986; Jewell and Brown, 1987; Weaver, 1987).
