Abstract
Surface mining for coal is a major economic activity in east central Ohio. Ohio has strict reclamation laws which require that the mining companies return the mined land to environmentally acceptable conditions. During the decade of the seventies, a particularly conscious effort has been made in Ohio to enforce the reclamation laws. Monitoring the reclamation efforts and progress via traditional means is time consuming, expensive, and often subjective. LANDSAT multispectral data provides a means to eliminate some of the negative aspects of the above.
A nontraditional unsupervised classification procedure has been devised using a clustering algorithm with a NASA modification of the canonical analysis algorithm as implemented on the Pennsylvania State University ORSER system. The algorithms are implemented on the ERRSAC IDIMS/HP 3000 at NASA/Goddard Space Flight Center in Greenbelt, Md. for use in the unsupervised classification approaches. A standard unsupervised clustering/maximum likelihood algorithm sequence is compared to a nontraditional unsupervised clustering/canonical transformation/clustering algorithm sequence in delineation of land cover categories in surface mining areas. This nontraditional unsupervised classification approach demonstrates appreciable improvement in spectral category groupings when compared to the traditional unsupervised classification approach and land cover information.
Date of this Version
1981