GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 3.3 517 BAYESIAN APPROACH TO THE INVERSION OF MAGNETIC DATA FOR NEAR-SURFACE STUDIES. APPLICATION TO THE ARCHAEOLOGICAL SITE OF SAN PIETRO INFINE (SOUTHERN ITALY) R. Salone 1 , S. Tarantino 2,3 , L. Spagnuolo 2 , A. Frisetti 4 , M. La Manna 1 , A. Emolo 2 , R. Di Maio 1 1 Dipartimento di Scienze della Terra, dell’Ambiente e delle Risorse, Università degli Studi di Napoli Federico II, Napoli, Italy 2 Dipartimento di Fisica “Ettore Pancini”, Università degli Studi di Napoli Federico II, Napoli, Italy 3 Istituto Nazionale Di Geofisica E Vulcanologia, Aquila, Italy 4 Dipartimento di Scienze Umanistiche, Università degli Studi Suor Orsola Benincasa, Napoli, Italy Introduction. As it is well known, the geophysical prospecting is a key tool for addressing the archaeological research, due to its capability to identify buried archaeological features, thus allowing to plan a proper excavation work (e.g., Di Maio et al. , 2016; 2018). To this purpose, the inversion process of the geophysical data has a fundamental role in the accurate modelling of possible anomalies resulting from any geophysical survey. In this note, we propose a new algorithm for the 2D inversion of magnetic data, which is based on a Bayesian probabilistic approach. The effectiveness of the proposed algorithm is tested on synthetic and field magnetic data for retrieving the parameters of the anomaly causative sources. In particular, the field data come from a high-resolution magnetic survey performed in the archaeological area of San Pietro Infine (Caserta, Southern Italy) (Fig. 1a), which is considered by the archaeologists as relevant for the reconstruction of the ancient Roman settlement. In fact, this site is known since pre-classical times, as it represents the junction of a road axis controlled by the fortified walls of the neighboring areas (Zambardi, 2009). During the Roman colonization, in the 3rd century BC, the pre-existing road axis was crossed by the Via Latina (Zambardi, 2007), as shown by the paving stones found near Santa Maria del Piano (Fig. 1a). The church of Santa Maria probably is identifiable with the medieval building of San Pietro in Flea (name derived from the roman toponym “ Ad Flexum ”), mentioned by written sources as a property of Montecassino (Fabiani, 1981). A new algorithm for 2D inversion of magnetic data. The proposed algorithm has been developed in a Bayesian probabilistic framework and validated on synthetic and real magnetic data. The basic concept is the assumption that a volume of magneticmaterial can be considered as a set of magnetic dipoles (Telford et al. , 1990), each of which generates its own anomaly. If these dipoles are aligned with each other, for example, because of induction due to an external magnetizing field, then the body may exhibit residual magnetism. The volume under consideration can therefore be considered as a continuous distribution of dipoles resulting in a dipole moment vector per unit of volume, M → , of intensity M, also called Magnetization Intensity. The inversion algorithm is, in fact, based on the superposition principle : the overall anomaly associated with the volume of magnetic material is due to the sum of the anomalies of the various dipoles present in that volume. For the computation of the forward problem, firstly the investigated volume has been discretized into cells to each of which a susceptibility contrast, ∆ χ , has been associated, which represents the parameter to be explored in the inversion procedure. Since the magnetic measurements in the archaeological test area were acquired in gradiometric mode, the gradient associated with each explored section along profiles extracted by the observed 2D magnetic map was calculated and subsequently compared with the acquired data along the selected profiles. For each possible 2D map of susceptibility, Δ χ x P , z P , the probability density function is as below: