Toward an Automated Model for Archaeological Site Detection in Eastern Botswana, a Clustering Method

Summary

This paper is an effort to create a predictive model for archaeological sites in an area of Eastern Botswana. With a rather arid climate, much of Botswana’s ground surface (and archaeology) is easily visible to airborne and spaceborne sensors. Without sufficient training data for supervised classification, an iterative spectral clustering method was used to group spectrally similar pixels from multispectral imagery into a large number of spectrally distinct but unknown classes. By visually assessing and removing classes that do not correlate with known sites in the region, the remaining classes provide a map for prospective site locations. This work illustrates how satellite imagery and digital remote sensing methods enable the inspection of large areas with little processing time, something that would be impossible from the ground in a single field-season. Also illustrated by this work is the need for on-the-ground inspection of the prospective sites to confirm their existence and to improve the model.

This Resource is Part of the Following Collections

• Society for American Archaeology 83rd Annual Meeting, Washington, DC (2018)

Keywords

Spatial Coverage

Record Identifiers

Abstract Id(s): 22583