The Classification Of Semi-natural Vegetation From Landsat Thematic Mapper Imagery A Users Perspective

RESUWE An unsupervised maximum likelihood classification of Landsat Thematic Mapper imagery is found to provide the basis for a thematic map of broad habitat types over an area of complex semi-natural vegetation. The study shows that the cover categories to be mapped must be selected such that they have some spectral homogeneity in addition to ecological significance. The use of photographic prints of satellite imagery in the field is found to be invaluable in this respect. Where cover categories are selected on the basis of ecological divisions alone results are shown to be poor.