The Prevalence of Objects with Sharp Boundaries in GIS

Abstract

Translate Article

NTRODUCTIONThe debate on vector versus raster based models is nearly as old as the concept of a GIS (Dutton 1979). It wasrestated as a debate between GIS with an object concept (not to be confused with object-oriented as used insoftware engineering; (Worboys 1994) uses the term object based GIS, where objects have sharp boundariesdelimited by vectors and the GIS which model the continuous variation of attributes over space using a regulartessellation - e.g. a raster (Frank 1990). Efforts to merge the two representations were attempted (Peuquet 1983).It has been a very fruitful debate as it has forced us to consider and reconsider the epistemological bases of ourwork and has led to an extensive discussion of fundamental questions (Chrisman 1987), (Mark and Frank 1991),(Mark 1993). The debate has promoted the development of ever more powerful software, achieving a nearlycomplete integration of vector and raster data (Herring 1990).It has been pointed out repeatedly that only few objects in geographic space have natural boundaries whichare sharp and well determined (Couclelis 1992). Most geographic objects seem to be an abstraction of thingswhich have unclear, fuzzy boundaries, if they have boundaries at all. The list includes most natural phenomena,from biotope to mountain range; extensive research efforts center around soil type data (Burrough1993);(Burrough 1986) and often use the techniques of fuzzy logic (Zadeh 1974). Nevertheless, many practicallyused GIS model reality in terms of crisply delimited objects. Cadastral systems, GIS used for facilities’management and automated mapping (AM/FM) and communal information systems all are appropriately orientedtowards distinct objects with well defined boundaries. The same systems, with the same models are also used tomanage soil maps and land use data, where the fiction of sharp boundaries contrasts with our view of reality.Depending on the application area or profession, one or the other spatial concept is more appropriate andallows one to capture the aspects of interest more succinctly. Users are accustomed to see a process described ina particular spatial framework, a legal discussion is most often cast in the framework of spatial objects with crispand determined boundaries whereas a discussion of the application of fertilizer might consider soil types as havingindeterminate boundaries. (Burrough and Frank 1995) and (Couclelis 1995) classify applications or phenomenaaccording to an evolving taxonomy of object types and of boundary types. This leaves two questions open,namely- where do these models of well-defined objects come from and- why are object models with sharp boundaries very often used in applications of GIS despite all obviousproblems with delimiting geographic objects?This paper attempts to address these two questions, starting from the point of view of experiential realism(Lakoff 1987). Experiential realism argues that the natural categories of human cognition are based on theexperience we have of the world and that the way we perceive the world is influenced by our cognitive apparatus(Couclelis 1992) for a practical application to geography (Mark 1993). There are two different environments for ourexperience, namely small scale and large scale (geographic) space. Our daily experience with handling smallobjects in small scale space, e.g., apples, stones, leads to a concept of objects with well defined boundaries. Thiscontrasts with the direct experience of large scale space, which leads to a different conceptual structure of space,mostly without dividing large scale space into delimited objects. These are both fundamental experiences forhuman beings which are deeply embodied in our thinking and give rise to the 'object' and 'field' concept.Experiential realism also assumes that metaphorical mapping (Lakoff and Johnson 1980) can be used toconceive situations in terms of previous experiences in different circumstances. For example, the experience withsmall prototypical objects with well determined boundaries is metaphorically translated to conceive the large scale