GNGTS 2015 - Atti del 34° Convegno Nazionale

as architectural elements, rotary querns in volcanic stone, mosaic tiles, slugs frommetalworking processes, and an imperial coin. Aerial photographs and survey data document the existence of a complex vicus during the Roman Imperial period. Otherwise, a less wide area was occupied by the Early Medieval settlement. At the same zone the fortuitous finding on the ground surface of a fibula and a bronze buckle allow to hypothesize to be related to the presence of a cemetery area. The site was still inhabited in the Late Middle Ages: to this period should be attributed some pottery kilns and perhaps an abbey, mentioned in historical documents of the sixteenth century, and then replaced by a farmhouse in the modern period. It is clear that the site at Badia plays a central role for the historical investigation of rural settlements in southern Apulia. In this work we focused our attention on the area where Roman Imperial and Byzantine evidences are mostly noticeable. Some rectangular transects have been identified as possibly promising zones for near surface geophysical prospecting (Fig. 1). In these areas Ground-penetrating Radar (GPR) and magnetometric (in gradiometry configuration) were undertaken. Due to the complexity of most archaeological environments, the integration of different methods can help to overcome the limitations related to a single method, such as low reflection amplitude in the case of scarce dielectric contrast between archaeological structures and hosting materials or a high level of clutter in strongly heterogeneous terrains for example in GPR. Amultimethodological approach, involving GPR and magnetometry, generally reduces the uncertainty in the interpretation of the results, helping to define better the boundaries of individual archaeological buildings or the extent and layout of large settlements (Cammarano et al. , 1997; Piro et al. , 2000, 2003; Gaffney et al. , 2004; Seren et al. , 2004; Cardarelli et al. , 2008; Giannotta et al. , 2014; Leucci, 2015). A good spatial overlap of anomalous zones in different geophysical datasets is generally a powerful indication of the presence of underground bodies with physical properties strongly different from the embedding soil, such as anthropogenic structures. Therefore, the integration of different geophysical surveys can help to better define the location, depth and geometry of archaeological bodies. Being sensitive to different physical parameters, different geophysical methods show anomaly patterns which, in most cases, do not overlap, pointing to features of different origin (Orlando, 2007; Leucci, 2015). As different methods give complementary information, their integration is essential to identify features that otherwise could be undetected and to provide a more comprehensive description of an archaeological site. GPR survey. Ground-penetrating radar (GPR) is one of the faster near surface non-invasive tools, and it provides to best available resolution among the geophysical techniques. Depending on the external conditions and on the central frequency of the antennas, it can allow on average a penetration depth from 1 to 3 m. It is based on the propagation and scattering of the electromagnetic waves, and consequently it is sensitive to variations of the electromagnetic parameters in the subsoil, especially the dielectric constant and the electric conductivity (Davis and Annan, 1989). The resolution achievable from a GPR ranges about from 40 to 10 cm in archaeological applications, again depending on the kind of soil and on the central frequency of the antennas. The investigated areas are geo-referenced in Fig. 1. The soil was composed of slimy and quite moist sediments over a calcareous basis. The prospecting was performed with pulsed a RIS H-Mode GPR system equipped with a double antenna at 200 and 600 MHz, in the case at hand the best results were provided by the antennas at 600 MHz, being the depth of the targets of interest smaller than 2 m. In the following, GPR slices and a three dimensional visualization are proposed (Conyers and Goodman, 1997; Leucci, 2002; Goodman and Piro, 2013). The profiles were 0.5 m spaced from each other, for each trace 512 time domain samples were chosen, the bottom scale of the times was 80 ns with regard to the 600 MHz antennas. We adopted a manual gain function to compensate the losses in the soil, and performed a 2D linear processing making use of the GPR-Slice Version 7.0 software ( www.gpr-survey.com/ ). 90 GNGTS 2015 S essione 3.2