Trends and Concerns of Spatial Sciences Managing Storage Structures for Spatial Data in Databases

Data driven, dynamical storage structures are well-suited for fast access on spatial data in databases. In general, spatial objects are stored such that neighbors are physically close to each other; however, these structures do not use information from a semantically richer data model and treat all spatial data similarly. The structure managing access on spatial data reflects the distribution of the data. A collection of spatial data consists of heterogeneously distributed objects. Some of them are equally distributed on the area, some are concentrated at a few locations, others occur less frequently and less systematically. Most storage structures for spatial data are trees or similar structures; the deeper they grow, the less efficient is the search within large regions for widely-spread and less frequently occurring objects. Data models which are part of databases provide details for different kinds of spatial data. This information about the structure of the data will be used adding some knowledge about the density of objects. This leads to a storage structure which is based on both the spatial location and extension of each individual object and the density of object classes. A specific structure, the Field Tree which is part of a database management system (PANDA), will be the centre of concentration. Managing Storage Structures for Spatial Data in Databases Page: 59 Towards an Extended SQL for Treating Spatial Objects M. Egenhofer in: Y.C. Lee (ed.), Trends and Concerns of Spatial Sciences, Proceedings of the Second International Seminar, Fredericton, New Brunswick, Canada, June 1987. TRENDS AND CONCERNS OF SPATIAL SCIENCES