A predictive model of the Wiyot Cultural Area in northwest coastal California
The focus of this thesis is a Geographic Information System (GIS) based predictive modeling process with results as applied to the Wiyot Cultural Area in Humboldt County, California. Predictive models aim to identify patterns that demonstrate a range of human activities, especially economic maximization and cognitive cultural decisions. This model relied on secondary document analysis in order to recreate the prehistoric environ and identify site type and use areas including historic maps, oral history, ethnographic documents and archaeological site records. Logistic regression was employed to discriminate significance values among the variables derived from the document analysis, selecting the strongest environmental and cultural variables for the modeling process. Topographic setting, distance to prairies, and distance to navigable waterways were selected as variables significant at the 0.05 level. Distance to freshwater, elevation, and vegetation setting did not meet the criteria for significance and were excluded. A relative probability map was produced which indicated both site and non-site locations within the study area. Internal and statistical testing showed that although only three variables were employed, the model exhibited a 66% predictive success rate with a 71% increase in performance over random guessing. The goal is for this model to be employed as a management, planning, and research tool for the Humboldt area. It is to be used as a voice of empowerment for indigenous peoples in their efforts to preserve prehistoric sites and sacred spaces on lands that encompass their traditional cultural area, but are officially owned or managed by non-tribal entities.