ABSTRACT
This chapter explores a typology of geographic information systems (GIS) operations in which the concept of overlay is absolutely central and examines the usual implementation of the method. It suggests that ‘cartographic modelling’ by map overlay should be treated as a statistical operation. Map overlay has been used as a basic tool for spatial analysis for almost as many years as thematic maps have been drawn. Most GIS provide an overlay operation that clearly reflects their data model which assumes that the real world can be adequately described by discrete point, line, area or field entities that have exact attributes. A reason for regarding almost all map overlay operations as fundamentally statistical is that some additional prior knowledge of the phenomenon being investigated almost always exists. An obvious stratagem is to use general linear models to screen the possible attributes and to estimate appropriate weights for the overlay.
