Three approaches to computer-aided analysis of LANDSAT-1 MSS data were evaluated utilizing data from a test site in rugged, mountainous terrain. The approaches compared include non-supervised (clustering), modified supervised, and modified clustering. Test field results indicated classification accuracies of 78.5%, 70.0%, and 84.7%, respectively for the three analysis techniques. The modified clustering approach proved to be the optimal computer- aided analysis technique of those tested because of minimal computer time required, highest classification accuracy, and most effective analyst/data interaction. A detailed description of this analysis technique is included.

Date of this Version