Mapping the Landscape for Archaeological Detection, Preservation, and Interpretation: A Case Study in High Resolution Location Modeling from the Blue Mountains of Northeastern Oregon
- Publication Year:
- Repository URL:
- https://digitalscholarship.unlv.edu/thesesdissertations/3170; https://digitalscholarship.unlv.edu/cgi/viewcontent.cgi?article=4173&context=thesesdissertations
- ecology; GIS; landscape; lithic; Modeling; predictive; ecology; GIS; landscape; lithic; Modeling; predictive; Archaeological Anthropology
thesis / dissertation description
Archaeological location modeling (ALM) is an important tool in most survey strategies, and has contributed substantially to economizing efforts to locate and characterize the archaeological record. The increasing availability of high resolution (<3m) airborne light detection and ranging (lidar) data has the potential to refine the application and ultimately the role of ALM. This research tests the precision and accuracy gained by incorporating lidar derived data into an ALM. The site records and other environmental data used in this study were all generated over the last four decades by the resource specialists of the Malheur National Forest. The Weights-of-Evidence (WofE) probability method (Bonham-Carter 1994) was used to produce two ALMs; one based on a 10m digital elevation model (DEM) created from satellite imaging, and the second from a 3m resolution lidar derived DEM. Independent variables (e.g., slope, aspect, distance to water, etc.) commonly used in ALM were largely replaced by index variables (e.g., slope position classification, topographic wetness index, etc.). The final models were classified into areas of high, medium, and low archaeological potential, then cross-validated against a reserved random dataset. Models were then compared using the Kvamme gain statistic and site to area frequency ratio. The 3m model demonstrated a significant improvement over the results obtained from the 10m model and the current probability model used in the study area. A number of factors including model resolution, statistical methodology, and the character of the independent and dependent variables all contributed to the increase in precision and accuracy. The incremental improvement in modeling efficiency demonstrated here will create time and cost saving in the management and preservation of cultural resources, and ultimately contribute to a better understanding of patterns of past human land use.