Susan Mulholland, George Rapp, Stephen Mulholland, Russell Rothe, Jame Huber
Mn/Model, a GIS-based model for site prediction in Minnesota, performed adequately for northeastern Minnesota; but the more numerous Woodland tradition sites used to derive the models overwhelmed the older Paleoindian/Archiac sites. This project was designed to enhance the Mn/Model databases to better represent the landscapes of northeastern Minnesota with respect to these early sites. Researchers conducted a logistic regression analysis of 108 known sites for common environmental parameters with four variables: Distance to water, stream flow, pollen average, and height above mean elevation within 90m. Field survey by shovel testing was restricted to 50 random points. Although locations of high, middle, and low probability for archaeological resources were tested, none of the locations had such resources. The lack of new sites in this survey is not considered an indication that the model is invalid, but more a result of the use of only randomly chosen points. The non-site data generated are useful for enhancing Mn/Model but the logistic regression model developed from the known sites still requires testing.
Download or order