Progress reports, GIS: the accuracy of spatial data revisited

Thematic maps have a central and often unquestioned role in geographical information systems (GIS) (Woodcock and Gopal, 2000). They are one of the most common ways of representing spatial data and are unsurprisingly, therefore, a common input and output of analyses undertaken in GIS. They are, however, only a model or simplification and hence a flawed representation of reality. Consequently, it is important that the quality of thematic maps is evaluated and expressed in a meaningful way so that their suitability for use may be assessed. This is important not only in providing a guide to the quality of a map and its fitness for a particular purpose but also in understanding error and its likely implications, especially if allowed to propagate through analyses linking the thematic map to other data sets. Unfortunately, for many geospatial data sets there is commonly a lack of information on data quality and what exists is often poorly communicated (Johnston and Timlin, 2000). The quality of spatial data sets is a very broad issue that may relate to a variety of properties (Worboys, 1998) but with thematic maps the property of interest is typically map accuracy. The accuracy of spatial data sets has long been an important issue in GIS and has been the focus of considerable research, particularly since the influential book by Goodchild and Gopal (1989). As accuracy remains a major research topic and issue of concern to many researchers, it seems pertinent to revisit the topic of accuracy, especially that of thematic maps. For instance, it is apparent that considerable development is required in the methods of accuracy assessment (Scepan, 1999) particularly as the accuracy of thematic maps is one of the greatest limitations to many users (Guisan and Zimmermann, 2000).