Research on automatic generalization methods of geographical spatial data based on semantic scale

摘要

Scale is an important factor when people acquire laws of geographical phenomena and processes. Generalized scale includes not only spatial scale and time scale but also semantic scale in geographic information science. Semantic scale describes semantic change amplitude and hierarchy of attribute contents of geographic entities. Semantic change amplitude represents attribute character changes in the unit time, the while hierarchy means classification and rank of attribute description. Scale is in inverse proportion to detailed degree of geographic entities when GIS displays multi-scale geographical spatial data. It is difficult that existing GIS display features of different semantic scale. As for the classified or ranked geographical spatial data the optimal solution is the hierarchy or rank of geographic entities displayed is higher when scale becomes small, so the generalization degree of detailed feature is higher. Ontology is a kind of modeling tool of concept model that is able to represent information system at the semantics and knowledge level. Geoontology is a kind of domain ontology and offers glossaries and relationships among concepts in the geographic spatial information domain. As far as the geographical hierarchy and classification system is concerned the relationships among the geographical concepts is hierarchy relationship, namely the relationship between the parent concepts and the child concepts or between hypernyms and hyponyms. Geoontology can represent formally this hierarchy relationship. A geographical concept can be navigated to its parent concept or child concept, and implements the automatic generalization of geographic spatial data by merging the features in the geographical feature classes corresponding to all child concepts of the some geographical concept in geoontology. However the automatic generalization method based on the geoontology cannot smooth the linear features and the boundary of polygon features, which should be implemented by resorting to other automatic generalization algorithms.