The ghost in the machine: spatial data, information and knowledge in GIS

Abstract

It has been a number of years since a paper on geographic information science (GIS) (as opposed to applications) appeared in The Canadian Geographer--a long dry spell or relief from an increasingly technical disciplinary focus, depending on your point of view. For those of you who lean toward the latter perspective this issue serves as an oeuvre to GIS and spatial representation as a way of communicating and integrating data from multiple disciplines. It is also an introduction to a much-changed GIS, one that is increasingly self-reflective. Those readers who have missed reference to GIS in this journal will find a multidimensional, integrative discipline that, perhaps, exceeds their previous expectations for representation and communication. Only a decade ago, circumspection about GIS was rife among geographers (Taylor 1990; Lake 1993; Pickles 1993, 1995). What has changed since then? First, the ability of GIS to incorporate multiple datasets and represent them using multiple epistemologies has extended its utility--and credibility. Second, researchers and academic practitioners have taken a reflexive turn, one permitted by both the sophistication of present-day GIS and a philosophical awakening in the GIS community that was partly galvanized by earlier critiques (Schuurman 2000). What distinguishes this collection of papers as a corpus is their demonstration of GIS--along very different axes--as a form of communication and integration. 'Communication' and 'integration' are encompassing nouns with which to describe GIS, but they may not go far in promoting its dimensionality without corroboration. A short homily serves this purpose. This is a story of GIS, public policy, earth science, representation, integration and communication--all succinctly packaged. Canada does not have a national groundwater policy or program. Rather, the provinces have jurisdiction over water. The Minister for Natural Resources Canada may ask the Geological Survey of Canada (GSC) for estimates of the total volume of groundwater in Canada--and be referred to the provinces, who have no way of answering the question. They cannot answer for all the usual reasons: there is no budget for the independent collection of groundwater data; there is no infrastructure to model the existing data; and there are few guarantees of the quality of existing data. Indeed, groundwater data are provided by private-water well drillers. In some provinces, such as Ontario and Newfoundland, well drillers are required by law to report well-logs that include location and lithological data to the provincial government. In most provinces, this is done on a voluntary basis. These data form the basis for groundwater management in Canada. Not surprisingly, the quality of the data varies widely. In some cases drillers are highly trained and do an excellent job reporting the lithology. The converse is also true, and ground-truthing is rare. Even locational data are questionable, as few drillers have GPSs. Instead, they report location with reference to local landmarks, and '400m from the barn at the intersection' is not an overstated example. The lithological data are just as variable, as few provinces have developed fixed classification systems. Well-log data provide an excellent example of the challenges in creating a geospatial data infrastructure. They are also prime candidates for standardization. In the early 1990s, scientists at the GSC teamed up with the Ministry of Environment in Ontario and initiated a project to improve the quality of well-log data for a portion of southern Ontario in the Greater Toronto Area (GTA) known as the Oak Ridges Moraine. Senior hydrogeologist Dr. David Sharpe headed a group of scientists who developed a local classification system based on twelve categories (which can be further subdivided) to standardize the 80+ classifications found in the well records. They added value to the data by combining it with surficial geology and reference data. …