side, these paths constrain the options available to users to those that, in the judgments of system designers, make the most cartographic sense in practice. It is, therefore, difficult or even impossible for users to do things to the data that are inappropriate cartographically. By contrast, information visualization systems do not have the geospatial option constraints of GIS. Typically, systems such as Advanced Visualization System or Data Explorer do not enforce projection and registration constraints, thus providing flexibility but also allowing for the problems of misplaced data in the display.
A GIS has the capability to produce high-quality graphical representations, especially maps, making it a potentially valuable support system for the process of spatial thinking in the K–12 context. However, users must be aware of and understand the importance of data quality. Professional-looking final products may conceal data errors. These errors may be referential (i.e., an error in specifying something such as a street address), topological (i.e., a linkage error in spatial data such as an unclosed polygon), relative (i.e., an error in the position of two objects relative to each other), or absolute (i.e., an error in the true position of something such as a floodplain boundary not aligned with property boundaries) (Tomlinson, 2003). Currently, no GIS can automatically handle data error problems in a satisfactory manner. Moreover, products may be graphically misleading. No GIS can guide K–12 operators in the choice of map symbols and other graphic effects.
The process of exploring data on a GIS-produced map could be enhanced if users had real-time control over the visual display. Most information visualization systems provide user interface controls that remain “live” after the display is constructed. This enables users to change the appearance of features in the display interactively (e.g., a color ramp, a size control for point symbols, a transparency control for an image layer). Currently, GIS lack such a capability.
GIS provides poor support for the modeling of time (Peuquet, 2002) and related presentations via animation (MacEachren, 1994). Unlike animation systems such as Director and Flash that explicitly represent time (t) values, existing versions of GIS have no temporal “coordinate” as in x,y,t. Although there are ways to work around this problem, achieved by stacking map layers in a temporal sequence of cross sections that can be refreshed several times per second (Goodchild, 1988), they lead to a noncontinuous sense of time for users. Many important aspects of science and geography revolve around processes occurring through time (e.g., carbon and water cycles, glacial change, migration, urban expansion).
Although GIS lacks the capability to examine processes that occur continuously through time, technology exists for large-scale geospatial virtual representations of the entire Earth over time and in three dimensions. Keyhole Inc. Images (http://www.earthviewer.com) provides users, even those with legacy computers, with access to terabytes of imagery and GIS files to view Earth as a three-dimensional object. Figure 8.2 shows screenshots of Earthviewer (http://www.earthviewer.com), which allows users to zoom smoothly from a whole-Earth view to resolutions as detailed as 1 m and to “fly” over a realistic rendering of Earth’s topography. The data to support these views are fed over the Internet, so a broadband connection is required for adequate performance. Earthviewer and similar developments come close to the vision of “Digital Earth” outlined by former Vice President Al Gore in Earth in the Balance (Gore, 1992). Earthviewer accommodates varying spatial resolutions, building its views dynamically from a patchwork of data obtained from various sources.
As an engine for performing transformations, operations, and analyses, GIS displays its full power for supporting spatial thinking. The earliest GIS was developed in response to the need to make accurate measurements of the size, shape, and characteristics of areas from large numbers of paper maps (Foresman, 1998), a task that is inaccurate, tedious, and expensive when performed by hand. This vision of a GIS as a calculating machine dominated thinking well into the 1990s, and