| Home | Portfolio | Resume | Photos |
Jim Kompanek
Introduction
The purpose of this project was the examine the environmental factors that may play a role in the distribution of Middle Woodland archaeological sites in Licking County, Ohio. Licking County is located in central Ohio, just east of Columbus, and is the location of numerous significant archaeological sites, including some of the largest geometric earthworks in the United States. This period dated from approximately 0 to 500 AD in central Ohio and corresponded with the rise of the Hopewell culture through much of the mid-west. This period is generally associated with nucleated villages located along major drainages, as well as associated geometric earthworks.
Objectives
The goal of this project was to examine the distribution of known Middle Woodland sites based on a series of environmental variables and to then develop a series of weighted parameters to predict the location of potential Middle Woodland archaeological sites in Licking County. These environmental factors were then combined using map algebra to create a final map of potential archaeological site locations.
Data Sources
The primary data sourced utilized for analysis was based on a shape file provided by the Ohio State Historic Preservation Office (SHPO). This file contained the location of every archaeological site in the county, as well as associated attributes. Although attribute information was available, only about 30 percent of sites (out of 102 recorded) contained information outside of site name and location (Figure 1). As a result, it was necessary to generate appropriate environmental attributes based upon data provided by other sources. This additional environmental data was obtained from the USGS seamless site, as well as the NRCS Geospatial Gateway. This included high resolution digital elevation model, hydrographic datasets, and soil maps for the county.
Utilized Data Sources
SiteData.shp ---> Shape file containing site centroids and associated attributes which was obtained from the Ohio SHPO office;
1/3rd Arc Second DEM ---> Used to determine elevation and generate slope, aspect, and relief raster datasets for the county, and obtained from the USGS seamless site;
Hydrographic dataset ---> Obtained from the USGS seamless site and used for the basis of determining the distance to water variable;
NRCS Official Soil maps ---> Obtained from the NRCS Geospatial Gateway and provided soil information for the county.
Figure 1. Middle Woodland archaeological sites recorded in Licking County, Ohio.
Results
Nearest Neighbor Distance
Before examining environmental variables, it was necessary to determine if the distribution of known archaeological sites was due to actual clustering of sites (events) or random chance. This was accomplished through a Nearest Neighbor Distance analysis. This analysis measures the distance between nearest points and compares it against the expected values from a random sample of points. According to this analysis, there is an observed mean distance index of 0.51 and a Z score of -9.4 standard deviations (Figure 2). As a result, it can be presumed that clustering of events is evident with a "less than 1% chance likelihood that this dispersed pattern could be the result of random chance." According to this analysis, there is a significant clustering of archaeological sites and the site locations are not simply an independent phenomena.
Though the above analysis appears promising, it is unclear to what degree sampling bias plays a role here. Because complete site attribute information is unavailable, it is unclear how many and which of the sites were identified as a result of a systematic survey or were identified in a haphazard manner by local informants. In addition, for the sites identified during a systematic survey, the survey boundaries were unavailable, so it is unclear whether certain areas may have lacked sites because of an actual lack of sites, or if they simply never undergone a systematic cultural resource survey.
There is also another fundamental question in regards to site distribution. Because of differential preservation, erosion, decomposition, limitations in testing, etc., it is simply not possible to identify every past human occupation, and even known sites are just a small sample of all past activities. It is unclear how the application of statistical methods are impacted by the examination of a sample-of-a-sample-of-a-sample. Regardless, for the purpose of this project, it will be assumed that the placement of documented sites is representative of all sites in the county.
Figure 2. Average Nearest Neighbor Distance Analysis of site distribution.
Kernel Density
Although sampling bias is an issue, the previous section indicates there is a likely significant clustering of archaeological sites. Visually, there appears to be a clustering of archaeological sites along the major waterways, and especially at the confluences of major rivers in the county. The next step involved generating a kernel density map of Middle Woodland archaeological sites (Figure 3). It was necessary to experiment with a variety of search parameters before coming up with a search radius of 5,000 m and a cell size of 100 m. A higher search radius resulted in most of the map containing a high density of archaeological sites and a lower search radius simply resulted in a small high density area surrounding each individual event. The radius set to 5,000 m provided a good balance between the two. The cell size of 100 m was utilized due to the scale of the map and was the smallest cell size that my computer could handle in an efficient manner. Figure 3 depicts appears to confirm the pattern observed in Figure 1, as sites are predominately distributed along major rivers and especially at their confluences.
Figure 3. Kernel density distribution of Middle Woodland archaeological sites.
Soil Series Analysis
The first environmental variable examined was that of the soil type. Because individual soil series represent a variety of intertwined environmental variables, it potentially plays an important role in site placement. As previously mentioned, the soil data was obtained from the NRCS Geospatial Gateway. Because the complete attribute data was not available for the provided site data, it was necessary to generate the soil count within ArcMap. Figure 4 depicts the number of archaeological site present within each soil type and Figure 5 contains this data as plotted within a county map. The next step involved classifying each soil type as "High, Medium, or Low" based upon potential for containing archaeological sites and to assign a number value for each classification. This breakdown was determined by Natural Breaks (Jenks) and the following weight value was assigned: High - 30, Medium - 20, and Low - 10. This weighing scheme was used for each of examined environmental variables for this project.
A potential limitation of examining soil series is that each soil series ranged in size from a few acres to tens of thousands of acres. The pattern observed in Figure 5 may simply be the result of certain soil types being larger than others. Figure 6 seems to indicate that the alluvial soils surrounding the major rivers, as well as the western-most soil types, are the most likely to contain archaeological deposits. Again, it is unclear whether this observation is due to an actual difference in distribution, or simply the result of more homogenous soil types in the western portion of the county. If more time was available, it may have been beneficial to "lump" similar soil series (e.g., all alluvial, upland, etc.) together to get a better picture of the local environment.
Figure 4. Bar graph indicating the number of sites (Y) for each soil series (X).
Figure 5. Soil types likely to contain archaeological sites.
Elevation
The elevation of each site was the next environmental variable examined (Figure 6). When the site distribution was examined based upon elevation. The break down of number of sites per elevation range was examined and the data suggests lower elevations are more likely to contain archaeological sites. Based upon this distribution, the elevation model was reclassified into the same classified scheme used above (Figure 7).
Figure 6. Elevation data for Licking County.
Figure 7. Elevation ranges likely to contain archaeological sites.
Slope
The slope of each landform was also examined (Figure 8). The slope was generated based upon the high resolution DEM for Licking County (see Figure 6). The number of sites per range of slope suggested that sites were more likely to be located on or along flatter landforms. Based upon Natural Breaks (Jenks) of the data, the slope raster was reclassified based upon the same weight scheme (Figure 9). The predicted distribution of sites based upon slope appears consistent with the previous analyses.
Figure 8. Terrain slope in Licking County.
Figure 9. Slope ranges likely to contain archaeological sites.
Aspect
The aspect of each landform was also examined (Figure 10). This was generated based upon the high resolution DEM for the county (see Figure 6). This break down of sites by aspect was categorized by Natural Breaks (Jenks) and the aspect raster was reclassified appropriately (Figure 11).
Figure 10. Aspect map generated for Licking County
Figure 11. Aspect ranges likely to contain archaeological sites.
Distance from Major Body of Water
The last environmental variable examined was the distance each archaeological site is from a major body of water. A Straight Line Distance Analysis was used to create a distance raster (Figure 12), which in turn was used to generate the distance each site was from a major river. This data suggested that sites were likely to be located close to a source of water, as the number of sites decreased with increasing distance from a river. Based upon this data, the subsequent raster was reclassified and weighed appropriately (Figure 13).
Figure 12. Straight Line Distance Analysis for major bodies of water.
Figure 13. Distances from water likely to contain archaeological sites.
Likely Locations for Middle Woodland Sites in Licking County
The final step for this analysis involved using Raster Calculator to combine each of the reclassified rasters using map algebra (Figures 14 and 15). The subsequent result appears to be somewhat similar to the previously generated kernel density map (see Figure 3).
Figure 14. Likely Locations of Middle Woodland Archaeological Sites (stretched) in Licking County, Ohio.
Figure 15. Likely Locations of Middle Woodland Archaeological Sites (classified) in Licking County, Ohio.
Conclusions
The examination of environmental variables did not lead to any surprising results and were quite similar to the kernel density map generated in Figure 3. Based upon the existing data, it appears that Middle Woodland sites are likely to be near major rivers, at lower elevations, on relatively level terrain, and facing a southern direction. If time permitted, a complete examination of the site distribution may reveal more interesting results. A statistical analysis to determine the significance of the relationship between each variable and site location may provide somewhat more meaningful results than simply classifying each site by natural breaks. It would also be interesting to examine how the different environmental variables related to each other. If significantly more time was available, a complete examination of the paper county site files would also have been useful; as different site types may potentially be located within significantly different environments. By combining all Middle Woodland sites together (as done in this project), a somewhat skewed perception of factors may have resulted. For example, a resource procurement site would likely be where the particular resource was plentiful, whether it be raw materials, fish, game, etc., irregardless of the other variables.
If I were to conduct this project over again, I would have focused on less variables, but examined them more intensively. I believe the relationship between site placement and soil types is probably the most significant, as each soil type is essentially a conglomeration of all the environmental variables examined in this project, as well as numerous others. By closely examining each soil series, it would be possible to combine similar soils, which in turn may paint a better description of the local environment.
This document is published in fulfillment of an assignment by a student enrolled in an educational offering of The Pennsylvania State University. The student, named above, retains all rights to the document and responsibility for its accuracy and originality.
