CONICET | Buscador de Institutos y Recursos Humanos

Mapsrepresenting the distribution of chemical residues over anthropogenic floorsare the main diagnostic tools used by archaeologists for addressing theidentification of geochemical signatures of past actions. Geostatistics allowsproducing these maps from a sample of locations by modelling the spatialautocorrelation structure of these kind of phenomena. However, the homogeneityof the prediction regions is a strong assumption in the model. The presence ofbarriers, such as the inner walls of domestic units, introduces discontinuitiesin prediction areas. In this paper, we investigate how to incorporateinformation of a geographical nature into the process of geostatisticalprediction. We propose the use of cost-based distances to quantify thecorrelation between locations, a solution which has proved to be a practicalalternative approach for archaeological intrasite analysis. The cost-basedapproach produces more reliable results avoiding the unrealistic assumption ofthe homogeneity of the study area. As a working example, a case study of thedistribution of two specific chemical signatures in domestic floors ispresented within a controlled ethnographical context in Northern Gujarat(India). On a broad disciplinary scale, the benefits of using our approachinclude improved estimates in regions with complex geometry and loweruncertainty in the kriging predictions.