Geographic Information System modeling to predict potential Great Oasis site locations in Iowa: an evaluation of three models using logistic regression
This study utilized Geographic Information Systems to create predictive models for archaeological sites in northwestern and central Iowa. The modeling focused on the group known as the Great Oasis culture that lived in Iowa between 950 and 1100 AD. A total of 139 Great Oasis sites are currently identified. Iowa has approximately 65 archaeological sites identified with a Great Oasis component that are concentrated in three major clusters near Sioux City, along the Des Moines River and along the Raccoon Rivers. One hundred and thirty non-sites were randomly selected for comparison with the known sites. The Great Oasis sites were modeled using a Geographic Information Systems database to create descriptive and predictive models of the environmental factors that may have influenced prehistoric site locations. The purpose of this study was to improve the understanding of Great Oasis sites in Iowa and determine the predictive power of archaeological modeling. Environmental variables were chosen based upon current literature about the Great Oasis and on Chi-Square analysis of variables. The variables found to be most appropriate for Great Oasis modeling were landscape position, distance to water, drainage class, parent material, and GLO historic vegetation. Three models were generated using ArcView 3.2. Statistical methods were used to determine the power of each model for predicting site locations. All models were analyzed using Chi-Square analysis, logistic regression, and map calculations. The models displayed a 43% to 59% improvement over chance for selecting areas that might contain Great Oasis sites. Archaeologists may use the findings to identify Great Oasis habitation sites and could be beneficial to Cultural Resource Management field surveys. Geographic Information Systems modeling of Great Oasis sites offers another tool for the analysis of site data.