High-resolution, 3-D multidisciplinary imaging across the 1980 (Ms 6.9) Southern Italy earthquake fault scarp at Pantano di S. Gregorio Magno (SA)

We present in detail the preliminary results of a 3-D high-resolution seismic dataset acquired at Pantano di San Gregorio Magno (SA) across a surface scarp caused by the Ms 6.9, 1980 Irpinia earthquake, i.e., the 4th largest Italian earthquake of last century (Pantosti & Valensise, 1990). Some previous successful attempts of 2-D seismic imaging performed by Bruno et al. (2010) in the same area, clearly demonstrate that the near-surface throw of the Irpinia Fault in this area is modest and, therefore, subsurface constraints may be complementary with paleoseismology and classical analysis of short-term morpho-tectonic indicators (e.g., Pantosti et al., 1993; Pantosti &Valensise, 1990) for active fault detection and characterization. The seismic acquisition, made in collaboration with INGV and CNR, is the key part of the scientific project TESIRA (TEst Site IRpinia fAult), funded in 2021 by the University of Naples "Federico II". The project aims at acquiring multidisciplinary geophysical data and to achieve, through the integration of this multivariate dataset, an accurate 3-D imaging of the shallow structure of the fault zone and to understand the link between the shallow faulting and the petrophysical changes, which affect rock permeability and surface degassing. A microgravimetric and a 3-D Electrical Resistivity survey have been already acquired between September 2021 and January 2022. The gravity measurements were carried out using a Scintrex CG-5 Autograv gravimeter, with a resolution of 1 ?Gal. 3-D electric data were acquired using 216 nodes, with both Wenner-Schlumberger and Dipole-Dipole configuration. The surveyed area is 240x230 m2. The 3-D seismic data were acquired in July 2022, using two overlapping arrays with a dense geophone distribution: 5 m inline spacing and 10 m cross-line spacing. The first array being made of 12 lines for a total of 504 receivers, the second array consisting of 11 lines, with 525 receivers. The two arrays overlap in the central part (5 profiles). This device setup investigated at high resolution an area of about four hectares, with a total of 18 NNE-SSW, profiles, and 789 geophone positions and a total number of 1,115 vibrating points acquired with a high-resolution vibratory source. The preliminary results will consist of a 3-D tomography illuminating the first 50 m of the investigated area and a preliminary stack. Trace editing, in particular first arrival picking, is particularly challenging, due to the larger numbers of waveforms to be picked (more than 550,000) and the use of a vibratory source which generates artificial events before the first arrivals. To facilitate the picking, we use and discuss the advantages and the limits of a machine learning algorithm (i.e., neural networks) for automatic event picking.