A Predictive Site Location Model on the High Plains: An Example with an Independent Test

Models of archaeological site location that have a predictive capacity formed an integral part of the Pinon Canyon Archaeological Project, located in Las Animas County, southeastern Colorado. The models were developed through a focus on settlement theory and previous archaeological work, and through a combination of techniques and technologies including multivariate statistics, decision theory, principles of pattern recognition, image classification, and the computer science domain known as Geographic Information Systems. One previously undescribed model for a portion of the Pinon Canyon study region is examined. This model, although it performs at only a moderate level, is noteworthy be cause a completely independent survey of that region recently has been conducted, allowing a iCblind test" of its performance. We may therefore compare the predicted performance of this model against ac tual circumstances allowing an objective assessment of the utility of predictive modeling at Pinon Canyon. Over the past decade there have been numerous studies devoted to the development of archaeological site location models that have a predictive capacity (e.g. Kohler and Parker 1986, Judge and Sebastian 1988; Scholtz 1981). A variety of such models were developed from 1983-1985 at the Pinon Canyon Maneuver Site, a large U.S. Army base encompassing nearly 1,000 square kilometers in Las Animas County, southeastern Colorado. The purpose of these models was to contribute to a description and understanding of prehistoric locational be havior, to guide subsequent survey efforts to those regions where sites were most probable in order that a significant portion of the limited amount of survey to be performed would be con ducted in the most archaeologically sensitive areas, and to provide a planning mechanism that would aid in the future management of cultural resources by indicating where sites were likely to be located in unsurveyed areas. Thus, predic tive site location modeling constituted an impor tant aspect of the Pinon Canyon Archaeological Project and, consequently, a rigorous, no-non sense modeling approach was required that would provide a useful, defensible, and testable outcome. Saddled with this grave responsibility as director of the modeling efforts, I decided to employ a proven methodology developed by scientists at NASA, the Jet Propulsion Laboratory in Pasadena, and elsewhere: tech niques of pattern recognition and classification of features in remotely sensed satellite images (Moik 1980; Schowengerdt 1983). By treating various characteristics of the Pinon Canyon landscape as "images", and known samples of archaeological site-present and site-absent locations discovered by our survey crews as the features of interest, these methods could be employed to develop a feature classification model that could be "mapped" over the entire Pinon Canyon landscape to indicate site-likely and site-unlikely regions, thereby yielding a predictive capacity (Kvamme 1988). The path toward achieving this outcome was long and tedious and involved not only knowledge of geographical and anthropological settlement theory and the local archaeological situation, but also multivariate statistical methods, decision theory, principles of pattern recogni

[1]  F. Massey,et al.  Introduction to Statistical Analysis , 1970 .

[2]  W. J. Judge Paleoindian occupation of the central Rio Grande Valley in New Mexico , 1973 .

[3]  Ernestene L. Green Location Analysis of Prehistoric Maya Sites in Northern British Honduras , 1973, American Antiquity.

[4]  Julius T. Tou,et al.  Pattern Recognition Principles , 1974 .

[5]  Robert L. Bettinger,et al.  Prehistoric Pinon Ecotone Settlements of the Upper Reese River Valley, Central Nevada , 1976 .

[6]  Michael A. Jochim,et al.  Hunter-gatherer subsistence and settlement: A predictive model , 1976 .

[7]  Gregory A. Johnson Aspects of Regional Analysis in Archaeology , 1977 .

[8]  L. Grossman MAN-ENVIRONMENT RELATIONSHIPS IN ANTHROPOLOGY AND GEOGRAPHY∗ , 1977 .

[9]  D. Roper Archaeological survey and settlement pattern models in central Illinois , 1979 .

[10]  Robert L. Bettinger,et al.  Explanatory/Predictive Models of Hunter–Gatherer Adaptation , 1980 .

[11]  Robert A. Schowengerdt CHAPTER 2 – Digital Image Processing , 1983 .

[12]  Robert A. Schowengerdt,et al.  Techniques for image processing and classification in remote sensing , 1983 .

[13]  Sandra C. Parker,et al.  Predictive Models for Archaeological Resource Location , 1984 .

[14]  P. Burrough Principles of Geographical Information Systems for Land Resources Assessment , 1986 .

[15]  Lynne Sebastian,et al.  Quantifying the Present and Predicting the Past: Theory, Method, and Application of Archaeological Predictive Modeling , 1988 .