High resolution 3D ERT to help GPR data interpretation for researching archaeological items in a geologically complex subsurface

Muro Leccese (Lecce) contains one the most important Messapian archaeological sites in southern Italy. The archaeological interest of the site arises from the discovery of the remains of Messapian walls, tombs, roads, etc. (4th-2nd centuries BC) in the neighbourhood. The archaeological remains were found at about 0.3 in depth. At present the site belongs to the municipality, which intends to build a new sewer network through it. The risk of destroying potentially interesting ancient archaeological structures during the works prompted an archaeological survey of the area. The relatively large dimensions of the area (almost 10,000 m(2)), together with time and cost constraints, made it necessary to use geophysical investigations as a faster means to ascertain the presence of archaeological items. Since the most important targets were expected to be located at a soil depth of about 0.3 m, a ground-penetrating radar (GPR) survey was carried out in an area located near the archaeological excavations. Unfortunately the geological complexity did not allow an easy interpretation of the GPR data. Therefore a 3D electrical resistivity tomography (ERT) scan was conducted in order to resolve these interpretation problems. A three-way comparison of the results of the dense ERT measurements parallel to the x axis, the results of the measurements parallel to they axis and the combined results was performed. Subsequently the synthetic model approach was used to provide a better characterization of the resistivity anomalies visible on the ERT field data. The 3D inversion results clearly illustrate the capability to resolve in view of quality 3D structures of archaeological interest. According to the presented data the inversion models along one direction (x or y) seems to be adequate in reconstructing the subsurface structures. Naturally field data produce good quality reconstructions of the archaeological features only if the x-line and y-line measurements are considered together. Despite the increased computational time required by the 3D acquisition and 3D inversion schemes, good quality results can be produced. (C) 2008 Elsevier BY. All rights reserved.