Location similarity of regions

This article gives a systematic investigation of location-based similarity between regular regions. Starting from reasonable conditions for such measu res, it is shown that there is only a finite number of location properties to be compared. The complete set of combinations is presented, and their be haviour and interpretability are discussed. Similarity measures are needed for all kinds of matching problems, includin g merging spatial data sets, change detection, and generalization. However, the measures are empirical measures. Therefore, measures found in literatu re seem to be chosen at random. With this synopsis, I show the differences of behaviour for all available choices. (C) 2000 Elsevier Science B.V. All rights reserved.