Geographical Information Systems (GIS) in which geo-coded data sets are manipulated and displayed are a powerful means of managing and analyzing spatial information, and are widely used in environmental planning and monitoring to aid planners to make deci sions and predictions. They are used increasingly by those involved in the preparation and interpretation of Environmental Impact Assessment reports. More effective support would be provided by a computer system which is able to capture and realize the reasoning processes involved in the planning process as well as storing and keeping track of the dynamic spatial data.
A severe limitation of Geographic Information Systems is their lack of support for reasoning about the spatial entities which they represent. Knowledge-Based systems (KBS) are well established and the technology well understood, but there has been little investigation of their amalgamation with GIS, although most modern GIS incorporate analytical models with an interface to powerful query langu ages (Kowalski 1986) and there is a large body of research into topological spatial relations such as (Clementini 1994) The lack of investigation into KBS/GIS integration is because, hitherto, the data representation techniques used in GIS have not been directly compatible with the representations used by existing approaches to spatio-temporal reasoning (Hayes, 1979, Allen, 1983, Kowalski 1986). The integration of GIS with knowledge-based systems would provide a means of supporting experts making decisions and predictions in the domain of dynamic spatial data.
The two types of data which are in common use in GIS are vector data, which originates from cartographic maps and point-referenced spatial data sets, and raster data from the growing number of remote sensors. A technique for the representation of data onto which raster representations map naturally is that of tesserals in which space is divided hierarchically into tiles (Morton 1966, Grunbaum, 1987). It has been suggeste d by (Diaz 1986) and (Bell, 1983) that it could be used as a data representation for GIS.
SPARTA (SPAtial Reasoning with Tesseral Addressing) is a tesseral spatial reasoning system which incorporates a technique, based on the quad tesseral representation of space, which operates a constraint propagation mechanism.
The data representation enables two dimensional (and higher) space to be treated linearly, allowing the reasoning technique to operate using only the constraints isWithin, isSubset and isEqual (and their negations). The technique has been used successfully to resolve two-dimensional reasoning scenarios and its use is illustrated in this paper by its application to the search for suitable building development sites which fall within acceptable noise pollution levels generated by traffi c from a nearby major road.