This project will develop and implement a three stage dynamic inductive archaeological predictive model. The model will use a GIS approach to integrate available environmental and cultural variables in order to rank proposed project areas for the likelihood of containing archaeological resources. Field verification in the form of archaeological survey of a proposed project area will aid in the refinement of the initial models.

Project Tasks:

1. Development of digitized environmental and cultural information for the three physiographic provinces of North Carolina, including computerizing existing OSA site files.
2. Development of GIS archaeological predictive models using the digitized information for the three provinces of North Carolina.
3. Creation of a WWW compatible GIS model Graphical User Interface (GUI) for use by NCDOT and SHPO/OSA staff.
4. Application of the GIS archaeological predictive models to multiple corridor/alternative TIP projects to aid in the selection of a preferred corridor/alternative.
5. Field testing of the model using GPS survey transects in upcoming NCDOT projects to quantitatively test and refine the model.

The initial work will be done in the Piedmont physiographic province. The Piedmont is an area consisting of all or portions of some forty-five (out of 100) counties, a total of 38% of the state. Over 360 USGS 7.5 minute quad sheets cover this region. The initial work areas will be based upon the locations of upcoming NCDOT highway projects where multiple alternatives are present. This will include TIP projects projected as Environmental Assessment (EA) and Environmental Impact Statement (EIS) undertakings. The initial model development will cover the following counties: Cabarrus, Granville, Guilford, Randolph, Chatham, and Wake (Figure 5).

The following table shows the size of each of the six counties in square miles, and the total number of prehistoric archaeological sites currently recorded in each county. Total site numbers and the total number of sites with data available in computerized format are also shown.

County Size (sq. miles) Prehistoric Archaeology Sites
Cabarrus 364.42 245 total, 171 in computer
Granville 531.17 278 total, 139 in computer
Guilford 650.12 426 total, 323 in computer
Randolf 787.47 1,324 total, 1,014 in computer
Chatham 437.00 784 total, 621 in computer
Wake 833.92 1,483 total, 1,483 in computer

Total 6 counties 3,884.57 4,370 total, 3,751 in computer
Total Piedmont 20,015.36 (38% of NC)
Total NC 52,672 sq. mi. 35,000 total, 21,000 in computer

This shows that we will be conducting a 19.4% sample of the Piedmont physiographic province, and 7.3% of the total area of North Carolina. This area also contains 12.4% of the recorded prehistoric sites in the state. This should be a sufficient representative sample for the Piedmont region. The counties also provide a good geographical sample of the Piedmont (North to South and East to West).

Existing digitized environmental data will be used in the initial model development. This will include all available data on county soils, relevant drainage system information, geological base mapping, USGS 7.5 and 15 minute quadrangles, Digital Elevation Models, geomorphological data, digital ortho quarter quads and other aerial photography and remote sensing data, and other available relevant information.

Previously recorded archaeological site information, including the North Carolina archaeological site forms, will be reviewed and the information will be digitized for use in the development of the model. These counties currently have a total of 4,370 recorded prehistoric archaeological sites, of which 3,751 are already available in computerized format. An assessment of the current data condition of OSA's database will be required to determine the anticipated data entry, capture, and conversion to standard compatible metadata. All GIS and database output will be compatible with NCDOT's Arc/Info and ArcView software environments.

These data will be used to establish a model to assess the probability of the presence or absence of archaeological sites within a user defined area, in this case TIP project areas. Isolated finds will not be used in the development of the model, and clear definitions of archaeological site presence/absence will be expressly defined. Previously documented sites will be assessed as the dependant variables used in the initial model development. Independent variables will consist of the environmental (and some cultural) constraints including (but not limited to) the following: site distance to water, degree and direction of landform slope, existing biotic communities, river basin, landform, elevation, and soil properties. Additional variables will be incorporated in consultation with NCDOT and OSA.

The modeling process and results will utilize a 30 meter grid cell resolution in order to coincide with existing 7.5 minute USGS data. This should have sufficient spatial resolution to allow for quantitative field testing of a user-defined area to further refine the initial model, while being manageable computationally. Spatial statistical modeling and analysis will be conducted using appropriate univariate and multivariate techniques. Multiple logistic regression analysis will be used to determine the appropriate model components.

Subsequent to the development of the initial model, it will be applied to NCDOT project areas as directed by NCDOT. The results will consist of both graphical and numerical representations of the probability of a given user defined area for containing archaeological sites. Probability will be expressed as low potential, moderate potential, or high potential. Refinements to these basic categories are expected as the modeling process in developed over time. Graphic representations will include color gradient-based GIS layers delineating a TIP project area's potential for containing archaeological sites. Numerical representations will include total acreage values, for each corridor/alternative by probability zone for containing archaeological sites, as well as percentage values for each probability zone.

A Graphical User Interface (GUI) will be created for NCDOT and OSA use. The GUI will be compatible with NCDOT Arc/Info and ArcView software, and will be designed to allow for the generation of TIP project area data on a routine basis by non-GIS specialists. Training, documentation, and written instruction will be provided to NCDOT and OSA staff at a level to allow proficient use of the application. In addition, this application will be accessible via the internet by the creation of a WWW site for the purpose. An important focus of the website design and creation will be to protect sensitive specific archaeological site location information, while at the same time allowing access to the predictive model by appropriate state and federal agencies for planning purposes.

After the application of the predictive model to specific TIP projects, portions of the TIP projects Area of Potential Effects (APE) will be subjected to intensive archaeological survey and site evaluation. Methods used during the field investigations to accurately locate archaeological resources will include Global Positioning Systems (GPS) data collection. This will allow for quantitative site location information to be compiled and exported to coordinate systems used by the NCDOT in TIP project development. Both real time and post processed code differential correction methods will be employed to assure the accuracy of the data and the integration of the field information into the GIS database. These field archaeological survey and site evaluation data will be used to test and refine the accuracy of the original GIS predictive model.

Upon successful development and refinement of the GIS model for the initial six counties of the Piedmont phyisographic province, the same approach will be extrapolated out to the remaining areas of the Piedmont.

The same overall process will then be conducted for the other two physiographic provinces (Coastal plains and Mountains) using the same process as outlined above. The basic assumptions used to develop the initial model will aid in the development of the Mountain and Coastal plain GIS based models, and it is assumed that these later models will benefit from lessons learned in the initial work in the Piedmont.