GNGTS 2017 - 36° Convegno Nazionale

570 GNGTS 2017 S essione 3.1 components of the electric and magnetic fields on surface were measured with a L-shaped configuration of 100 meters electrical dipoles and two Ant/4 magnetometers. Considering the technical problems that forced to repeat some soundings, 22 MT sites were successfully acquired, and added to those already available in the area (Fig. 1). For each site, at least 17 hours were recorded using the sampling rate of 256 Hz. In addition over one hour with a sampling rate of 4096 and 1024 Hz was also acquired. We processed the data and the directionality and dimensionality analyses have been performed. The results indicate a wide 3D structure at depth with a clear role of a 2D structure mainly at shallow level in the proximity of the Cornia basin. We have selected 10 MT sites, with static shift effects, for TDEM data acquisition. The TDEM soundings were acquired by using a TEM-FAST 48 and laying out a rectangular loop of wire 50x50 and 100x100 m and pulsing it with a controlled current; the configuration was a coincident loop. Methods. The results of a geophysical inversion can be influenced by the reliability of the starting model. Based on this concept, for each of the 4 MT profiles, three resistivity models were implemented and tested as starting model for the 2D inversion (Rodi and Mackie, 2001): i) homogeneous halfspace; ii) resistivity distribution by assigning resistivity values to units of the geological model; iii) interpolation of 1D models obtained with Particle Swarm Optimization (PSO). The homogeneous a-priori model is a halfspace with an electrical resistivity of 100 Ωm. The implementation of the a-priori models from geological information required some effort. First of all, the detailed 3D geological models for the Lago Boracifero area was built in Petrel environment. Hundreds of deep geothermal wells, seismic and geological data have been integrated. The study of six resistivity well logs allowed the definition of resistivity values to be assigned to each geological unit. In addition in the frame of the IMAGE project a deep surface to borehole ERT was acquired (Capozzoli et al. , 2016). Electrodes were located in a very high temperature borehole, the Venelle 2 well, that is going to be deepened for the exploration of supercritical resources. The results have confirmed the low resistivity anomalies that were recognized by previous magnetotellurics surveys, which results were debated for a long time. Considering the large variability of the ground resistivity and the scarcity of direct data in the deep level of the system, the geological model could not be completely reliable as starting model for MT inversion. As a way to constrain the inversion using only MT information we also tested the use of a-priori models based only on the MT data and not on geology. We interpolated 1D models to obtain pseudo-2D resistivity models to be used as a-priori models for the deterministic 2D inversion. We present our attempt to treat the 1D magnetotelluric inverse problem with a probabilistic approach, by adopting the Particle Swarm Optimization (PSO), a heuristic method based on the concept of the adaptive behaviour to solve complex problems (Godio et al. , 2016; Santilano, 2017). This approach allowed us also to perform a joint optimization of MT and TDEM data with an innovative scheme. Fig. 1 - Location of the MT and TDEM soundings and profiles: 1) Quaternary; 2) neoautochthonous (M-P); 3) Ligurids (J-Eo); 4) Tuscan Nappe (T-M); 5) Calcare Cavernoso and anhydrites (T); 6) metamorphic Units (Paleozoic); 7) remote site; 8) Lago Boracifero sector.