A DEEPER DIVE INTO THE WATER: A COMPARISON OF HYDROLOGIC FEATURES AS VARIABLES IN PRECONTACT SITE LOCATION PREDICTIVE MODELS FOR THE VIRGINIA PIEDMONT

No Thumbnail Available

Publication or External Link

Date

2024

Citation

Abstract

The use of predictive modeling in Cultural Heritage Resource Management (CHRM) archaeology has become commonplace since its foundational principals were established in the 1980s, but criticisms of the practice persist, often centered around their lack of theory and dehumanization of the archaeological record. Proximity to water, typically expressed in the United States as distance to streamline data from the National Hydrography Dataset (NHD), is one of the most utilized variables when creating predictive models for Precontact period sites, but how does the variable “distance to streamline” compare to other hydrologic variables? In this thesis I seek to answer the question “how do distance to stream confluences and distance to wetlands compare to distance to streamline when attempting to predict Precontact site locations in the Virginia Piedmont?”The publication Quantifying the Present and Predicting the Past: Theory, Method, and Application of Archaeological Predictive Modeling (Altschul et al. 1988) is considered foundational to the practice of predictive modeling in archaeology; it is referenced frequently in modern theoretical works and throughout this thesis. The approaches to creating archaeological predictive models are typically divided into two camps: models that utilize an inductive, or correlative, approach and models that utilize a deductive, or theory driven, approach. Rather than establishing distance correlations between wetlands and stream confluences with previously recorded site data, I utilize a deductive approach where I establish the importance of those variables through archaeological theory pertaining to subsistence and settlement patterns and test their value with site data. Inductive associational models are very good at showing that archaeological site distribution is strongly patterned, but they often lack the explanatory framework that would be useful for management decisions based on their findings. The Study Area the models are tested on is located within Orange County, Virginia near the town of Locust Grove, and encompasses about 686 acres. The Study Area contains two main streams, named Cormack Run and Mine Run, the confluence of those streams and other lower order streams, as well as wetlands located adjacent to the streams. Precontact occupations have likely occurred in this region for the past 12,000 years, if not longer. The test results demonstrate that models created using deductively derived variables perform well enough to justify their use in CHRM contexts, but also include the added benefit of an explanatory framework. The guidelines for archaeological investigations in Virginia allow for the use of predictive models when conducting inventory surveys, meaning the archaeological predictive models (APM) created for this thesis could be utilized in a real-world context. The primary focus of this thesis was to determine if using hydrologic features other than streams, specifically stream confluences and wetlands, to express the distance to water variable would improve the performance of an APM. I demonstrated that, yes, other hydrologic features may be better predictors of Precontact site locations in the Virginia Piedmont. Secondarily, I hoped to show that an APM created using a deductive approach would perform well enough to be considered appropriate for use in CHRM contexts. The high probability areas of all three of the APMs I created yielded Kg values high enough to be considered as having predictive utility. This demonstrates that the use of all three of the APMs I created could be considered appropriate to guide survey efforts in a CHRM context.

Notes

Rights