Semantic similarity applied to geomorphometric analysis of digital elevation models

Abstract

This paper presents an approach to measure the semantic similarity between digital elevation model (DEMs). We compute a semantic "distance" between concepts in hierarchical structure of geomorphologic application ontology. The method is composed of two stages: analysis and measurement. Analysis stage is focused on performing a geomorphometric analysis: a qualitative value (descriptor) representing a concept in ontology is assigned to each DEM's cell. Measurement stage is based on a comparison between the descriptors of two DEMs. In other words, we measure the semantic "distance" (called herein confusion) between two ontology concepts. The similarity is defined at two levels of measurement: local and global. The first one is defined at cell-level value and the second one considers the entire cells in a DEM. Thus, the similarity between two DEMs at the conceptual level is established. For instance, our methodology can detect that two DEMs share the same or similar geomorphologic features: plateau, downhill, etc. © Springer-Verlag Berlin Heidelberg 2009.