D. M. Mark1
1 NCGIA & Department of Geography, University at Buffalo, Buffalo, NY 14261 USA
Telephone: (1) 716 645 0505
Fax: (1) 716 645 5957
Since the 1960s, geomorphometry has emphasized geometric calculations based on local operators applied to digital elevation data (Evans, 1972; Mark, 1975). Digital elevation data represent discrete approximations to fields of elevations that represent
the objective, measurable form of land surfaces. With the emergence of the World Wide Web and the Semantic Web, there has been increased need for methods to conversion of elevation fields into cognitively meaningful landforms. Some of the research has approached the problem in a ‘bottomup’ way, classifying form at a local level. However, in a somewhat counter-intuitive development, improving quality and resolution of digital elevation has had the effect of broadening the conceptual and computational gap between local geometry and meaningful landforms. The issue of automated feature extraction and classification is further complicated by the fact that people from different cultures, speaking different landscapes, group landforms into categories in different ways. Procedures for the
automated detection, delimitation, and classification of landforms from elevation data may themselves need to be different for different languages, and at the very least will need to have different parameters. This paper expands on the issues outlined above, and then present evidence for cultural and linguistic differences in the classification of landforms. Then some aspects of the solutions will be presented.