ABSTRACT
One of the more obvious issues from this perspective is the existence of two traditions of GIS analysis. The distinction between raster and vector is often seen as a problem of system design, but actually presents a major issue of data interpretation. To emphasize this difference, and to stress the context of data interpretation rather than system design, we will use the terms field and object in this paper, although they are to some extent synonymous with raster and vector respectively. Some spatial databases represent the world as if it were populated by objectsÐpoints, lines and areasÐwith associated attributes, continuing a tradition developed in cartography. Others represent the world as fields, or arrays of pixels, again with associated attributes. The choice between the two representations has variously been seen as depending on the method of data collection (satellites generate fields,cartographers generate objects), the degree of spatial resolution required (objects appear to imply higher levels of spatial resolution, whereas pixels imply a level which is fixed by the pixel size), and the efficiency of algorithms (for example, the widely held perception that overlay is faster in raster). However we will argue in this paper that fields and objects represent fundamentally different forms of abstraction of geographical reality.
