Spatial information systems were created manually for many years. For example, historically, the source of cholera in London in 1854 was found by John Snow, by mapping where victims of the disease lived. The graph clearly showed them to be close to the Broad Street Pump,1 one of the city’s water wells. Another example is Zipf’s Law,2 which predicted that a number of sociological phenomena. His regression model followed a straight line on log-log paper. For example, Zipf’s law describes the rank of metropolitan statistical areas (SMSAs) in the census of cities over 2,500 plotted against their population, or, for that matter, the number of copies of a St. Louis newspaper bought in suburbs out to 150 miles.
What is new in our computer age is that storage capacity, computing speed, and technology all grew to the point where large volumes of geographic and spatial information can be used for understanding business and other phenomena. Spatial information covers all sorts of data (e.g., demographics, customer locations, real estate locations and values, and asset location). Sears, for example, uses spatial information to find the optimal routing for its 1,000 delivery trucks that cover 70% of the U.S. population. To use the system, they hand-coded millions of customers’ addresses, put them in their data base, and then use the results to determine routes each day.
A spatial information system is more complex than a conventional back-office database system. In many respects, it is like a CAD/CAM system where data is kept on each of many layers, and the layers can be superimposed on one another as the user chooses. The spatial information system accesses spatial and attribute information, analyzes it, and produces outputs with mapping and visual displays. It needs to keep data on spatial boundaries and on attributes of the data. It includes tools and models to manipulate the data and boundary information. Furthermore, rather than just adding up columns of numbers, it requires understanding and using numerical algorithms, statistics, and operations research optimization and simulation techniques. In addition, spatial tools act on boundary layers, including union, overlay, buffering, nearest neighbor, and spatial attraction. You need much more expertise and capability to deal with spatial information systems data than with ordinary databases.
Despite the complexity, or perhaps because of its complexity, spatial information systems provide capabilities that offer competitive advantage. Of course, determining the extent of that advantage is a difficult task. Pick3 points out that the up-front data costs tend to be much higher (more data needs to be acquired) and some costs, such as training needed for users with no experience, are difficult to estimate accurately. Furthermore, many more of the benefits are intangibles compared to conventional systems. For example, spatial information system applications tend to move up the value chain of a firm as they are used for planning or decision support (see Section III of this book). The value of the visualization aspects of spatial information systems is hard to quantify because we know little about how visual models improve decision-making.