Jessica Mellar

University of Wisconsin, Milwaukee

Geospatial Distribution of Archaeological Features in the Ka’a’awa Valley

Introduction

After conducting a ground based survey to identify and locate sites of past anthropogenic land use within the Ka’a’awa Valley, I became interested in the relationship between the location of archaeological features and the environment. Recent historic changes to the landscape as a result of World War II land use, cattle ranching, and tourism have increased the potential for disturbance to archaeological sites along the valley floor and thus has decreased the ability to identify features within this setting.

This distribution of locatable features has resulted in the following research questions: What is the relationship between the location of archaeological features and the environment within the Ka’a’awa Valley? Do the physical and environmental factors of slope, aspect, and distance to water act as a geospatial control affecting the location of features?

Methodology

In order to answer these questions, it was necessary to maintain consistency in data collection. Data was collected using a Trimble GPS unit in order to provide an accurate account of the distribution of features. Additionally, a Pentax Optio WG-14 was used to obtain geotagged photos of the sites for reference.

GeoTagged Photos to Points in ArcMap 10.1 for Reference

I worked with Shelby Eagleburger in order to create a data dictionary to be used for archaeological research within the context of the Ka’a’awa Valley. The data dictionary was used to identify features based on shape, composition, and soil color.

Data Dictionary Used for Data Collection

The features were then classified based on the above attributes in an attempt to create a discrete typology for the identified features. For example, rock walls are features that are linear in shape, composed of unworked rock, and exist in an agglomerated structure. A terrace is also linear in shape and composed of unworked rocks but exists in parallel lines embedded into the ground.

Example of Classification Rock Wall (left) and Terrace (right)

Platforms are features that are rectilinear or circular in shape, composed of unworked rocks, and are paved with small cobbles. Rock scatter is the term used to describe a seemingly unnatural placement of rocks on the hillside. This feature is commonly located directly downhill from other rock features.  The term rock feature is used to describe an agglomeration of unworked rocks that appear anthropogenic in nature, but do not have a single distinctive form.

Example of Classification Platform (left), Rock Scatter (center), and Rock Feature (right)

Analysis

In order to provide an analysis of the relationship between archaeological features and the environment, it was necessary to apply a Digital Elevation Model (DEM) to the World View 2 imagery. I used the slope process in ArcMap 10.1 to provide the basis for further analysis. Hillshade was then applied to the slope raster output in order to calculate aspect. It was imperative to convert all archaeological features to points in order to extract raster values to the points and append those values to the attribute table.

Once aspect was determined, I used the Extract to Point process to identify the aspect of each known archaeological feature. The features were then symbolized accordingly for visualization purposes and the attribute table was exported to Excel. This allowed for a comparison of aspect by feature class which resulted in providing a percentage of each feature class and its relationship to aspect.

Distribution of Features by Aspect

I was surprised to learn that there are no features that fall within a Southwest or Flat aspect. When looking at the distribution of features by aspect, it is significant that more than 50% of rock walls occur on an East aspect while two-thirds of platforms occur on a Northeast aspect.  Additionally, I compared the location of archaeological features and proximity to water. In ArcMap 10.1, I merged all water features into one linear feature. The resulting hydrology shapefile was then used as to run the Near tool for analysis.

The resulting map provides a visual symbolization of each feature categorized by equal interval with the darkest blue representing close proximity to water and yellow representing furthest distance to water. The attribute table from this function was then exported to Excel to produce a graphic representation of proximity to water by feature class.

Bar Graph Showing Proximity to Water by Feature Class.

The feature class Rock Scatter only occurs within a short distance of water. Because of this relationship, it is possible that rock scatter is not anthropogenic in nature and is a result of natural erosion processes. Additionally, the class Rock Feature decreases in distribution as distance to water increases. This relationship may help to identify the original purpose of these otherwise unknown features.

Conclusion

Further analysis is needed to distinguish the purpose and chronology of the archaeological features located within the Ka’a’awa Valley. Although the goal of my research was to identify and compare anthropogenic features dating to pre-contact Hawai’i, it is likely that some of these features are the result of recent historical land use. It is important to note that the data used for this analysis is not representative of all archaeological features within the Ka’a’awa Valley. Rather, these features represent only archaeological sites that were identifiable through ground-based survey conducted June 10 – June 17.

There appears to be a relationship between feature class and environmental factors, however, the small sample size has resulted in inconclusive findings.  Future archaeological research in the Ka’a’awa Valley should include a valley-wide ground-based survey in order to locate and identify all sites of anthropogenic land use. Using this data, a statistical analysis of the geospatial distribution of archaeological features within the context of slope, aspect, hydrology, and other factors should be conducted to determine what factors most strongly affect the location of features.