identification of clusters of leukemia cases. The method was, however, limited by its use of brute force (nonefficient) methods.
In a recent extension of GAM, Openshaw (1995) has implemented a space-time-attribute analysis machine (STAM) linked directly to GIS. The goal of STAM is to determine a set of GIS operators that, when applied to a database query, will identify cases that form a statistically unusual pattern. In its most recent implementation STAM incorporates the use of "genetic" algorithms (i.e., algorithms that adapt to their environment, for example, by defining the appropriate spatial scale and resolution for a particular problem context).
Another significant development in GIA is the incorporation of object-oriented programming (OOP) concepts to extract the data needed for an analysis from a GIS, solve spatial analytic problems, and subsequently link the solutions back to the GIS for display of results and further analysis. One example of this approach involves development of custom routing software for use by the U.S. Agency for International Development (USAID) in decision making related to food aid transport in southern Africa (Ralston, 1994). Southern Africa imports large quantities of food grains, and the state of the economy dictates that much of this is in the form of food aid. The software developed for USAID is based on an OOP approach linked to the commercial GIS, ArcInfo. The combination provides a flexible tool for prepositioning food in storage facilities, setting prices for acquisition of more carrying capacity, determining where to add more storage, making distribution decisions, choosing best modes and routes for transport, and determining the location and cost of bottlenecks. The OOP approach was found to be particularly useful in quickly adapting the decision model to deal with changes in obstacles to distribution.
Geographic visualization (GVis) can be defined as "the use of concrete visual representations . . . to make spatial contexts and problems visible, so as to engage the most powerful human information processing abilities, those associated with vision" (MacEachren et al., 1992, p. 101). The dramatic increase in volume of georeferenced data being collected and generated today is exceeding our capacity to analyze and digest it. Using the power of human vision to recognize patterns and synthesize spatial information increases the capacity of geographic researchers to cope with this data volume. For example, a simple 48 × 48 matrix of fiscal transfers for the United States generates 2,256 pieces of information for each time period considered. Such information can be concisely summarized in a simple yet effective visualization (e.g., see Figure 4.6).
GVis combines display with analysis capabilities to enable the search for patterns and relationships; the identification of anomalies; the analysis of directions and flows; the delineation of regions; and the integration of local, regional, and global information (see Figure 4.6). The development of flexible GVis tools