Archaeological Probabilistic Modeling of the Charleston Naval Weapons Station, Berkeley and Charleston Counties, South Carolina

Summary

As part of the Phase II testing of 29 archaeological sites and the Phase I archaeological survey of approximately 4,000 acres of the Charleston Naval Weapons Station (NWS), this study uses previous archaeological research to develop a GIS-based probabilistic model (set of equations) that can be employed as a land management tool. The model identifies land units most likely to contain archaeological sites from specific time periods. Fifteen GIS data layers were generated representing the known or potential relationships between archaeological site locations and 11 independent environmental variables. Fifteen different equations were employed to create unique GIS data layers that represent the sum of all predictor variables ( with varying weights) across the four USGS 7.5 minute quadrangles containing the NWS. Each formula was based on evaluation of previous probabilistic models in the region and the background archaeological literature. The completed GIS data layers were divided into site probability categories of High, Moderate, Low, and Very Low, and then evaluated against all of the known archaeological sites from the study area.

Three principal characteristics are expected in the best formulas for each of the relevant time periods: (1) the equation displays a significant positive relationship between sites and high and moderate potential zones; (2) the equation displays a significant negative relationship between sites and low and very low potential zones; and (3) all or most sites from each time period are located within high or moderate potential zones. These characteristics were determined by a special fit equation that used values from a chi-square analysis. Using multiple formulas ensures that a broad-based look at potential settlement indicators is evaluated and the best equations for each time period can be identified. New formulas, variables or land areas can also be added to the database and incorporated in future studies with little or no alteration of the model.

Although the equations were tested against known archaeological sites and the most promising ones identified, additional testing of the results is required. Since the distribution of previously recorded archaeological sites is dependent on past survey strategies and project areas ( only some of which are available as digital data), evaluation of the probabilistic categories against those data may be highly biased. Therefore, a small scale test of the model will be carried out through the Phase I survey on the NWS and the inclusion of field results from Brockington et al. (1995). The same survey methodology will be employed for a set percentage of the survey acreage, and evaluation of the field results will be compared to the probabilistic assessments by time period ( using all 15 formulas). No test may be possible for zones that do not contain statistically significant surveyable acreage within the NWS. The results may indicate variables that should be dropped from the model or suggest additional variables which may be important.

This Resource is Part of the Following Collections

• Joint Base Charleston Cultural Resources Archive

Keywords

Spatial Coverage

Individual & Institutional Roles

Contact(s): Joint Base Charleston

Prepared By(s): Brockington and Associates, Inc.; Ecology and Environment, Inc.

Submitted To(s): US Navy Facilities Engineering Command

Record Identifiers

Contract No.(s): N62467-97-D-0860

File Information