Classification of digital terrain models through fuzzy clustering: An application

Experts classifications of spatial data are strongly affected by subjectivity and rigidity of rules. They do not take into account, in a quantitative way, the overlap of classes and as a major consequence, their classifications are often not reproducibles. To overcome this subjectivity, exploratory techniques can suggest a coherent set of rules that will produce suitable polythetic and overlapping classes. The aim of this paper is to validate the unsupervised method of fuzzy clustering applied to classification of raster spatial data