R-trees for large geographic information systems in a multi-user environment

The authors explore the application of the R-tree hierarchical data structure to a geographic information system (GIS). In addition to demonstrating the R-tree's efficiency in accessing large databases in secondary storage, the authors show how an R-tree-based data structure can facilitate Boolean operations such as unions and intersections as well as area and perimeter calculations on regions. Because of the extra level of abstraction that an R-tree places on data items pointed to by its leaf nodes, R-trees can be implemented on top of most existing GISs that use data structures ranging from cellular representations to polygonal representations and to hierarchical structures such as quadtrees. A set of algorithms for extending the basic R-tree to enable concurrent accesses in a multiuser environment is discussed as is the possibility of parallel processing on R-trees. The potential for a relational-database linkage is also discussed.