Integrating Multi‐element Geochemical and Magnetic Survey at Ancient Sagalassos (Southwest Turkey): Anthropogenic Versus Natural Anomalies

Abstract

ABSTRACTSeveral studies in the field of archaeological prospection have suggested that multi‐element soil geochemical survey holds potential as an archaeological survey method, complementing more common techniques such as magnetometry and fieldwalking. Despite this, only limited research regarding this topic has been carried out. In this study, a geochemical survey was conducted in the Eastern Suburbium of the Roman to Byzantine city of Sagalassos, with the aim of investigating the possibilities and limitations of the technique in facilitating the interpretation of magnetic survey data. To explore the multivariate patterns present within the chemical dataset, the data were statistically analysed using a contiguity‐constrained spatial clustering algorithm. The resulting spatial clusters revealed that an area characterized by high‐frequency magnetic anomalies was spatially associated with a chemical cluster rich in Cu, K, P, Pb and Zn. This is interpreted as resulting from the accumulation and decomposition of occupational debris. Enhancements of Al, As and Ba helped to interpret an area with a low magnetic signal as a region where limestone bedrock was located close to the soil surface. Finally, two zones of enhanced magnetization were shown to spatially overlap with two clusters of soils containing elevated levels of Co, Cr, Mg, Mn, Ni and (Fe) and Ti, V and (Fe), respectively, indicating the presence of two different types of mafic to ultramafic ophiolitic bedrock near the soil surface. This study confirms that multi‐element geochemistry has potential as a survey technique, because it offers direct information on soil bedrock or ancient human disturbance, thereby adding an extra dimension to the interpretation of geophysical survey data. It is shown that soil chemical data are particularly valuable for distinguishing anthropogenic and natural impacts creating geophysical anomalies. Contiguity‐constrained clustering was found to be a promising data analysis technique for multivariate geochemical survey data. Copyright © 2013 John Wiley & Sons, Ltd.