Automated classification of landforms on Mars

We propose a numerical method for classification and characterization of landforms on Mars. The method provides an alternative to manual geomorphic mapping of the Martian surface. Digital elevation data is used to calculate several topographic attributes for each pixel in a landscape. Unsupervised classification, based on the self-organizing map technique, divides all pixels into mutually exclusive and exhaustive landform classes on the basis of similarity between attribute vectors. The results are displayed as a thematic map of landforms and statistics of attributes are used to assign semantic meaning to the classes. This method is used to produce a geomorphic map of the Terra Cimmeria region on Mars. We assess the quality of the automated classification and discuss differences between results of automated and manual mappings. Potential applications of our method, including crater counting, landscape feature search, and large scale quantitative comparisons of Martian surface morphology, are identified and evaluated.