Comparative cluster analysis to localize emergencies in archaeology

In the Northwestern Peru, near the city of Trujillo, an Italian Archaeological Mission (MIPE), in close collaboration with the National Institute of Culture (INC) of Peru, is working on the site of Chan Chan, the widest archaeological complex in the world for the constructions in raw earth. The mission is carrying out a multidisciplinary survey with the purpose to study a methodology for the site maintenance and the creation of a local center of documentation. Inside this project we want to explain a step of the research by means of a remote sensing analysis based on a Quickbird high-resolution image. By applying some image classification techniques, it has been possible to investigate the whole archaeological site, with the aim of a better definition of its general features and of a contribution to the exploitation of the surroundings. This analysis was carried out in open source environment implementing and performing two different classification algorithms and comparing the final results. As we had at our disposal no set of reliable ground truth of the study area it was necessary to follow an unsupervised classification approach. The off-line clustering techniques have been: K-means and Fuzzy K-means. Both approaches rely on minimizing a cost function of dissimilarity (distance) measure. We have overcome the problem of data group overlapping in a hard classification testing a fuzzy approach where each observation belongs to a cluster with a fuzzy membership degree. The resulted classification clearly separates the central archaeological site from the areas intensively cultivated and besides puts in evidence the coastal band, characterized by a reduced parceling and by the presence of damp zones. Another purpose of this research has been also to individualize the principal features of the Chan Chan site, possibly, by putting in evidence some archaeological emergencies and discovering new mark and signs. We performed every phase by the software Integrated Land and Water Information System (ILWIS) 3.4 developed by the ITC of Enschede and open source from July 2007. All collected information are useful to organize a master plan in a GIS, dedicated to the conservation of the archaeological complex.