Data Structures and Algorithms for Geographic Information Systems: Selected Topics

For the performance of geographic information systems, it has been widely recognized that efficient data structures and algorithms supporting proximity based access to spatial data are a crucial ingredient. Because of that, in the last years, a lot of data structures for storage of multidimensional non-point objects have been proposed in the literature (for an overview see our tutorial introduction in this volume). Almost all of these approaches are made to support queries based on spatial neighborhood of objects, so-called proximity queries. Common examples of such proximity queries are range queries, returning all objects intersecting a given (rectangular) region, or nearest neighbor queries, determining the object nearest to a given point. But a geographic or cartographic information system should also support queries that do not exclusively refer to the spatial neighborhood of objects. For example, different versions of spatial data must be organized such that proximity queries to any version of the data can be answered efficiently. Or, given a set of object classes, a spatial query that involves an arbitrary combination of these classes should also be supported. Because of the growing size of data sets to be handled in geographic information systems, an appropriate data organisation to support such and several other application-specific queries becomes more and more important. In this paper, we will present two approaches for access structures supporting the two types of queries mentioned above. Handling of versioned, spatial data is the topic of Section 2, while Section 3 discusses queries on some classes selected from the set of object classes stored in a geographic information system.