UUnniivveerrssiittyy ooff AArrkkaannssaass,, FFaayyeetttteevviillllee SScchhoollaarrWWoorrkkss@@UUAARRKK Graduate Theses and Dissertations 5-2015 LLaannddssccaappee VViissiibbiilliittyy aanndd PPrreehhiissttoorriicc AArrttiiffaacctt DDiissttrriibbuuttiioonn aatt PPeeaa RRiiddggee NNaattiioonnaall MMiilliittaarryy PPaarrkk Jake Lee Mitchael University of Arkansas, Fayetteville Follow this and additional works at: https://scholarworks.uark.edu/etd Part of the Archaeological Anthropology Commons, and the United States History Commons CCiittaattiioonn Mitchael, J. L. (2015). Landscape Visibility and Prehistoric Artifact Distribution at Pea Ridge National Military Park. Graduate Theses and Dissertations Retrieved from https://scholarworks.uark.edu/etd/1106 This Thesis is brought to you for free and open access by ScholarWorks@UARK. It has been accepted for inclusion in Graduate Theses and Dissertations by an authorized administrator of ScholarWorks@UARK. For more information, please contact [email protected]. Landscape Visibility and Prehistoric Artifact Distribution at Pea Ridge National Military Park Landscape Visibility and Prehistoric Artifact Distribution at Pea Ridge National Military Park A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts in Anthropology By Jake Mitchael Hendrix College Bachelor of Arts in Anthropology, 2011 May 2015 University of Arkansas This thesis is approved for recommendation to the Graduate Council ______________________________________ Dr. Marvin Kay Thesis Director ______________________________________ Dr. Kenneth L. Kvamme Committee Member ______________________________________ Dr. Jami J. Lockhart Committee Member Abstract Pea Ridge National Military Park, in the north east corner of Benton County, Arkansas, is the 4,300 acre site of a crucial Civil War Battle. Human occupation of the Ozark Highland landscape, however, extends far into pre-history. A 2005 report to the National Park Service details the findings of a four year cultural resource survey of the park. The sampling strategy employed in the research design (random sample site selection and 2.5% park coverage) provides an excellent dataset to assess prehistoric land use. This dataset is not dependent on artificially defined sites, representing singular activity in a limited geographical space. Instead it allows for interpretation of patterns of land use; while artifacts may not be spatially or temporally associated, their provenience on the landscape can be assessed in relationship to various landscape elements and environmental variables. Trends in artifact location can be seen with this representative sample distribution. The 2005 report examines artifact distribution with respect to permanent and intermittent streams. The predictive models produced from the analysis closely relate the availability of water and caloric expenditure required to travel across the landscape to a majority of the prehistoric material at the park. The report also explores seasonal expressions of land use at Pea Ridge. The goal of this project is to explore the relationship between another landscape variable, visibility, and prehistoric locations that do not conform to the models of the original study, those with higher travel costs to water. Economic models like cost-to-water are meaningful interpretations of land use, but I feel that such models preclude other elements of landscape experience. By comparing the distributions of conforming and aberrant prehistoric artifact groups against three different measurements of visibility, I hope to show that landscape perception could be a reliable predictor of prehistoric material in high cost areas. Acknowledgements Thanks are in order to all who contributed time, information, thoughts, support, feedback, data, or recommendation towards my completion of this project. Thank you especially to Dr. Marvin Kay, who provided access the PERI survey data and reports and a backboard for me to bounce ideas off of. I appreciate your time and dedication to helping me develop a worthwhile product. Thanks to Dr. Ken Kvamme for his frequent aid with the spatial/ analytical component of this project. Thanks to Dr. Jami Lockhart for his time and suggestions for dealing with the manipulation and extent of the elevation data. For dredging up information regarding decade old projects, thanks to Dr. Jason Tullis and Jason Herrmann. Thanks to Cassidy Van Beek and Julie McGilvray of the National Park Service for the time and data. Additional thanks to Rachel Opitz and Kevin Eads. Thanks to all those involved in the initial 2001-2004 survey. Without that legwork, examination of artifact distribution on such a large scale would not have been possible. Finally, a most loving thanks to my family, Betsy and Elsie, for the support at home. Table of Contents I. Introducton 1 II. Project Context 4 Pea Ridge setting 4 The PERI survey as a distributional dataset 4 III. New Data 10 LIDAR derived DEM 10 Data considerations 13 IV. Assessment 17 Landscape visibility as a factor of land use 17 Cost surfaces 19 Prehistoric location groups 22 Variables of landscape visibility 24 Landscape openness 25 Visible sky 27 Assessment of the landscape visibility variables 28 Landscape visibility regression model 33 Seasonality and Viewshed 35 V. Discussion and Conclusions 44 Post assessment considerations 44 Conclusion 47 VI. References Cited 49 VII. Appendix A 54 VIII. Appendix B 59 I. Introduction A 2005 survey report of the Pea Ridge National Military Park (PERI) details a modern strategy for statistical archaeological sampling, the analysis of the collected data, and the conclusions drawn from the analysis (Kay and Herrmann 2005a, 2005b, 2005c, 2005d). The sampling strategy is versatile, suitable for both historic and pre-historic survey in most every geographic region, surface visibility condition, and vegetation. The impacts are minimal, involving as much surface survey as possible and a specific regime of shovel tests in each sample unit without extensive, costly, and destructive excavation. The coverage is substantial and flexible; as much could be done as can be afforded. Such a survey returns a truly distributional dataset, not reliant on high density deposits or bounded site areas. The records of material remains discovered, recorded, and recovered from the survey were digitized and assessed using Geographic Information Systems, specifically IDRISI and ArcGIS, and produced a number of intriguing interpretations. One decade removed, the sampling strategy, survey technique, and models produced stand firm. What has improved is the spatial resolution of the digital model of the space assessed in the survey. The original study was structured and conducted on a 30 meter Digital Elevation Model (DEM). Light Detection and Ranging (LIDAR) scans of Benton and Washington Counties, Arkansas, including Pea Ridge National Military Park, now exist that improve the resolution to 25 feet. Such improved resolution drastically affects the perception of the space in such an analysis (Chase et al. 2012; Harmon et al. 2006). My goal is to utilize the DEM from the LIDAR dataset to contribute to the assessment of land use at Pea Ridge. As an extension of the original assessment I would like to test quantifiable aspects of 1 space and landscape morphology as variables in a phenomenological assessment of land use. The debate in archaeological theory about what we can truly “know” has always baffled me. I understand the merits of the two extremes: on the one hand, what can be measured can be understood, and on the other, personal experience and individuality defy summarization (Hole, 1980; Zubrow, 2006). I, however, do not find the positivist and post-processual perspectives to be mutually exclusive. The objective/ subjective dichotomy so frequently contested in theoretical discussion is not wholly in opposition, I think. Typically, “scientific” archaeology attempts to generalize cultural manifestations into “laws,” glossing over any subjective or experiential interpretation. I argue that the problem arises from generalizing specific instances of human activity, not that generalization itself precludes experiential interpretation. Rather than assessing a site of cultural activity and developing generalities about specific behavior, I would like to assess the nature of a distribution across a landscape as a range of human activity. The synthesis of deductive and inductive reasoning in model construction is not new (Kay and Allen, 2000; Kay and Herrmann, 2005a, 2005d; Redman, 1973; Schiffer and Gummerman, 1977b; Wheatley and Gillings, 2004), but the recognition of patterns in archaeological data typically revolves around physiological/ economical aspects of human land use. Measurable aspects of landscape and the location of artifacts in relation to them should allow for both “scientific” and “phenomenological” interpretations. Both physiological needs for survival and a metaphysical experience of the world can both be represented using techniques of quantified spatial analysis (Knapp and Ashmore, 1999; Llobera, 2003, 2006; Lock and Harris, 2000; Tilley, 1994; Wheately and Gillings, 2000). Specifically, I aim to assess elements of landscape as they pertain to visibility. Perception of the environment is an integral part of land use (Gibson, 1979; Knapp and Ashmore 2 1999, Llobera, 2003, 2006; Tilley, 1994). Visual components of a landscape can serve both functional/ economic and experiential/ metaphysical purposes. This report is the result of an attempt to validate such a claim using variables that represent landscape visibility rather than a traditional viewshed analysis. Visibility measurements commonly used in archaeological assessments involve intervisibility of sites or the portion of the landscape visible or invisible from an observation point. Such calculations pertain to the particulars of the observation points, quite frequently monumental architecture or relatively large, dense sites (Fisher et al. 1997; Jones, 2006; Lake et al. 1998; Llobera 2006; others). While the value of assessing such visual interplay between such sites and their surroundings is apparent, there is little to suggest that intervisibility of positive prehistoric locations at PERI is significant whatsoever. Instead, I want to represent the landscape in a visual sense (as an extension of Marcos Llobera’s concepts of visualscape and visual property: landscape elements of a visual nature [2003, 2006]) and assess the distribution of artifacts in this context. As a secondary goal, I would like to explore visibility as a factor in land use strategy for upland areas during summer, when water is least abundant. The broad distribution of prehistoric locations beyond the caloric-cost predicted zones appears to coincide with areas of increased solar illumination during the summer months. During a hot and dry Ozark summer, what factors would encourage land use well away from limited available water? I argue that the broad distribution in upland areas confers a more expansive viewshed, making upland exploitation a valuable land use strategy in summer. 3 II. Project Context Setting Pea Ridge National Military Park is a National Park Service (NPS) landholding located in the Northeast corner of Benton County, Arkansas. The park was the site of a large American Civil War battle, but the cultural resources on the premises are not limited to the conflict. Both prehistoric and historic material remains are abundant within the park boundaries, as evidenced by this survey (Kay and Herrmann, 2005a, 2005b, 2005c, 2005d). Frequent and repeated, if not continuous human occupation at PERI is suggested with even a cursory assessment of the cultural resources present (Kay and Herrmann, 2005b-c). Situated near the southern edge of the Springfield Plateau, a sub-region of the Ozark geologic formation, PERI is characterized by low rolling mountains with plains between. The landforms are moderately dissected by ephemeral, seasonal, and permanent streams. Typical Ozark vegetation populates the landscape: oak forest and low growth understory dominate some areas (Cozzens, 1940) while tree stands are interspersed with prairie grasses elsewhere. The topographical diversity within the park, as well as the landscape redundancy throughout the Southern Ozark region make it an ideal setting for an exploration of land use decisions (Kay and Herrmann, 2005a). The PERI survey as a distributional dataset The initial PERI survey was carefully crafted as a truly distributional study of the cultural resources, both historic and prehistoric, for the National Park Service. Rather than copy/pasting the research design from typical resource inventory and salvage operations, the project implemented a sampling strategy structured to address the resource management needs of 4