Modelling Geographic Phenomena at Multiple Levels ofDetail

Considerable interest remains in capturing oncegeographical information at the finescale, and from this, automatically derivinginformation at various levels of detailand scale via the process of map generalisation. Thisresearch presents a model (database) generalisationapproach able to support the derivation of phenomenatypically present at 1:250,000 scale mappingdirectly from a large scale topographic database.Such transformation involves creationof composite objects, such a settlement, forest,hills and ranges, from component objects, such abuildings, trees, streets, and vegetation in thesource database. In order to perform thistransformation it is important to model the meaningand relationships among source database objectsrather than to consider the object in terms of theirgeometric primitives (points, lines and polygons). The proposed approach involves identification of suchrelationships and then utilising these relationshipsto create higher order objects. The utility of theresults obtained is demonstrated using spatialanalysis techniques and the creation of 'links'between objects at different representations neededfor MRDBs.