GIS Predictive Modeling in Archaeology
The development and use of spatial models which reflect human settlement on the landscape are of both academic and practical planning interest. Archaeologically, the analysis of settlement patterns can provide information on environmental strategies, social organization, and other valuable scientific information. From a management perspective, the development of models of previous human settlement can improve the planning process and management of growth within existing regulatory constraints (such as NEPA compliance). This is possible because the potential for archaeological sites can be compared with various other environmental and cultural information in the GIS environment. Many local, state, and federal governmental entities are currently using GIS for management and planning purposes, with many smaller, local government agencies (such as counties and municipalities) successfully adopting the technologies. Access to archaeological predictive data allows the incorporation of cultural resource information into the overall planning process. The development of models of prehistoric and historic landuse provides the best chance for areas with a higher potential of cultural resources to be avoided, if possible, or for sites to be located and documented before development, if needed.
Recently, a number of states in the U.S. and provinces in Canada have begun projects to incorporate archaeological sensitivity information into their statewide GIS databases. Recent statewide GIS examples include the Mn/Model of Minnesota (BRW, Inc. 1996), Ontario, Canada (Dalla Bona, 2000), Kentucky (Mink et al. 2000), and North Carolina (Madry, et al. 2000).
The overall framework for this work is that of regional archaeological settlement analysis. The analysis of regional archaeological settlement patterns dates from the early work of Phillips, Ford and Griffin (1951) in the lower Mississippi Valley, and Gordon Willey's (1953) important work in the Viru Valley of Peru. In these early works the theoretical and methodological issues of regional settlement pattern analysis were delineated. Later work in Mesoamerica (Adams, 1961) and Mesopotamia (Sanders 1960, 1965), integrated these earlier approaches with ecological and systems theory. The application of systemic paradigms to settlement archeology by David Clarke (Clarke 1972) was another major early development.
The development of cultural resource management legislation in the 1970's significantly altered the way that archaeological data were both discovered and utilized. This change from site specific excavations to regional survey and testing occurred at the same time that GIS was being developed as a practical planning and research tool.
Regional archaeological site predictive models developed within the context of GIS are now an established and relatively mature subfield of archaeology, having been conducted around the world since the 1970s (Sebastian and Judge 1988, Thomas, 1988, Lock and Stancic 1995, Westcott and Brandon, 2000). While the most appropriate techniques and methods are still being debated (Ebert, 2000), the ability to create useful site location models for management purposes is now generally accepted within both the archaeological and planning communities. Several reviews of such models have been published (Thomas, 1988, Allen, Green, and Zubrow 1990, Warren, 1990, Maschner, 1996, Westcott and Brandon 2000).
The majority of GIS predictive modeling projects seek to model the spatial distribution of archaeological resources in the landscape using environmental (and less often cultural) variables. Many studies have shown a positive correlation between environmental variables (soils, elevation, slope, distance to water, etc.) and changing patterns of archaeological settlement over time (Padgett and Heisler 1979). Ward showed a relationship with soil types that are preferable for corn production with Mississippian site locations well before the advent of GIS (Ward 1965). Medium to coarse, well drained soils were found to correlate with habitation sites in Mississippi (Jenkins, Currin, and DeLeon 1975:65-68).
GIS provides a powerful, integrative environment for the mapping and analysis of cultural resources over a given landscape. The generally accepted method in use today is an inductive modeling procedure, where the location of a sufficiently large sample of known archaeological sites are compared with various other values in the GIS (usually, but not always, environmental variables) at the same location. Patterns of co-occurrence are statistically determined, and areas with similar patters of occurrence are then "predicted' to be of higher likelihood to contain similar archaeological remains. The opposing paradigm would be a purely theoretical and deductive model, where a theory of human occupation of the landscape is developed and tested against a known area. This is rarely conducted for a variety of practical reasons. A major factor is that anthropological theory has not developed to the point that it can be used to develop site-specific quantitative models that would be useful in planning activities.
