Abstract
The research presented in this article tests a methodology for estimating ancient Maya populations through the use of an ISODATA unsupervised classification of QuickBird imagery. The aim of this research was to expand the results of ground surveys in jungle environments in a cost-effective manner. The lowland jungle of Guatemala is composed of a mosaic of vegetation classes and geomorphological catenas. The ancient Maya exploited the ecological niches present in the landscape but chose to build their residences predominantly on well-drained uplands. Upland terrain and vegetation can be accurately isolated from the rest of the landscape using an unsupervised classification of aggregated multispectral QuickBird data. By testing a 25 km2 research area near San Bartolo, Guatemala, the study found that this methodology presented similar results to archaeological surveys conducted elsewhere in the Maya area at a much greater cost and allowed for comparable population estimates.
Acknowledgements
I would like to thank William Saturno of Boston University for his invitation to participate in archaeological research around San Bartolo and for the use of remote sensing data acquired by the project. Thomas Sever, Burgess Howell and Daniel Irwin of MSFC/NASA provided constant support in the interpretation and manipulation of remote sensing data for archaeological investigations. Nicholas Dunning conducted environmental investigations that greatly enhanced my understanding of the intersite area. Numerous archaeologists are to be thanked for contributing to the intersite research, including: Damaris Menéndez, Julio Cotom, Robert Griffin, Joshua Kwoka, Jose Garrido and Patricia Rivera. Field research was supported by a John G. Owens Fellowship awarded by William Fash at Harvard University, while support for writing came from a postdoctoral fellowship at Brown University. I would like to thank two anonymous reviewers and Stephen Houston, who reviewed an earlier draft of this article.