New insights into the High Agri Valley deep structure revealed by magnetotelluric imaging and seismic tomography (Southern Apennine, Italy)

Highlights

• Electrical resistivity model obtained from a 2D Magnetotelluric survey in High Agri Valley (Southern Apennine, Italy);

• 3D seismic tomographic model obtained from the inversion of passive seismic data;

• Joint interpretation of electrical resistivity values and P-wave seismic velocities using a cluster analysis.

Abstract

We present an electrical resistivity model obtained from a 2D Magnetotelluric survey across a large sector of the Southern Apennine in the High Agri Valley (HAV), a NW-SE trending intra-mountain basin, with a high seismogenic potential. The intensive hydrocarbon exploitation (Val d'Agri oilfield) makes this area also affected by induced seismicity. In this HAV sector, the injection of salt-water in an unproductive disposal well (Costa Molina 2) causes localized swarms of microearthquakes; a second cluster of continuous induced seismicity is also observed SW of the Pertusillo Lake and it is associated to the seasonal fluctuations of the reservoir's water level.

The major insight inferred from this study concerns a better understanding of the geological and tectonic framework in the HAV. The electrical resistivity model images the subsurface as conductive sedimentary sequences (Allochthonous Units) upon the carbonate Apulian Platform Unit characterized by higher resistivity values. Both these units appear composed of thrust-and-fold system deepening with larger wavelength anticlines N-E toward. Most of the structures identified in the magnetotelluric model are rather superficial and confined within the Allochthonous Units. A sudden break of the Apulian platform under the central part of the MT profile defines a conductive zone possibly associated to a major SW-dipping reverse fault or to several branches, as closely spaced thrust-sheets cutting eastern flanks of the Agri Valley.

Additional information on the HAV deep structures comes from the joint interpretation of the resistivity model and a 3D seismic tomographic model obtained from the inversion of passive seismic data collected in the period 2002–2018. The availability of this elastic representation of the subsurface allowed us to perform a cluster analysis on the electrical resistivity and seismic P-wave velocity distribution within the subsoil. This joint quantitative interpretation unveiled new insights, otherwise hidden by individual models, on the subsurface structure distinguishing some rheological zones in terms of barriers and asperities.

Introduction

The High Agri Valley, (hereinafter HAV), a NW-SE trending intra-mountain basin of Southern Apennines, is one of the areas of Italy with the highest seismogenic potential as demonstrated by the historical Mw 7.0 strong earthquake occurred in 1857 (Mallet, 1862). Moreover, the impoundment of an artificial water reservoir (Pertusillo Lake) and the intensive hydrocarbon exploitation of the Val d'Agri oilfield makes this area affected by induced seismic events as well. Since 2006, the coproduced saline water formation by oil exploitation has been disposed, in the interval between 2890 and 3096 m depth (b.s.l.), through its injection into an unproductive well (Costa Molina 2, hereinafter CM2, at 1045 m a.s.l.), located in the marginal sector of the Val d'Agri oilfield. This activity is causing fluid-induced microseismicity (Stabile et al., 2014a; Improta et al., 2017), well confirmed by the fact that the onset of the seismicity occurred only after June 2006 when the injection operations began (Stabile et al., 2014a). In addition, a second cluster of continuous induced seismicity has been observed SW of the Pertusillo Lake and it has been associated to the seasonal fluctuations of the reservoir's water level (Valoroso et al., 2009; Stabile et al., 2014b; Stabile et al., 2015; Telesca et al., 2015).

Since early Pleistocene, brittle tectonics have strongly controlled the formation and evolution of the HAV, which is bounded to the eastern flank by SW-dipping transtensional faults (Eastern Agri Fault System, EAFS) and to the western flank by NW-SE trending and NE-dipping Monti della Maddalena Fault System (MMFS), (Maschio et al., 2005). The seismogenic fault system causative of the 1857 earthquake is still debated. According to Benedetti et al. (1998) and Cello et al. (2000) the active fault system should be the EAFS, whereas Maschio et al. (2005) attributed the main seismogenic potential of the area to the MMFS.

While in the Northern Apennines the good quality of seismic profiling (e.g., Barchi et al., 1998; Boncio et al., 2000) provides a clear geometry of deep active faults, in the Southern Apennines less data are available. Indeed, this area is characterized by a low magnitude/poor instrumental seismicity (Mazzotti et al., 2000; Mazzoli et al., 2005) and the seismogenic sources for large historical earthquakes are unknown (Ercoli et al., 2020). The assessment of the active structures is very complicated but, in some cases, important results concerning the deep structure of subsoil are kept hidden by oil enterprise for strategic reasons.

Despite several tens of years of oil and gas exploration activity, most of the subsurface geophysical data, specially coming from seismic lines (Butler et al., 2004; Shiner et. al., 2004; Nicolai and Gambini, 2007; Mazzoli et al., 2013; Candela et al., 2015), magnetotelluric surveys (Dell'Aversana and Morandi, 2002), and data wells (D'Andrea et al., 1993; Monaco et al., 1998; Finetti et al., 2005), are not public. Then, the deep crustal structure of the Southern Apennine in the sector of the HAV is still a matter of discussion.

Therefore, this orogenic segment is well suited for applying non-invasive investigation techniques such as magnetotellurics, in order to contribute to a more holistic framework.

Generally, magnetotelluric (MT) exploration can provide valuable information on the structures and processes at lithospheric level thanks to high capability of penetration depths and its sensitivity to image lateral and vertical electrical conductivity variations. As known, fault zones are characterized by significant permeability contrast due to their fracturing level. Deep crustal fluids could strongly affect conductivity specially if interconnected over kilometer scale distances (Wannamaker et al., 2002; Bedrosian et al., 2004). High electrical conductivity could be indicators of rheological weak zones in the crust within which fracturing are concentrated, against low electrical conductivity values typical of the undeformed zones (Tank et al., 2005).

Unfortunately, most of geophysical inverse problems must face a twofold order of problems: the spatially varying resolution of obtained images and the non-uniqueness of the mathematical solution. These factors constitute a huge limitation for the right analysis of a geophysical model, and it is especially true when the interpretation is based on images describing a single physical parameter, such as resistivity. These problems are very often overcome by integrating the information provided by separated geophysical approaches and by their joint interpretation, which may be carried out in both a qualitative (Stanley et al., 1990) and a quantitative way (Bedrosian et al., 2007; Shaharabi et al., 2016). To this purpose, several works in literature combined seismic and electrical methods for the subsoil characterization and description (e.g. Eberhart-Phillips et al., 1995; Bedrosian et al., 2004; Tank et al., 2005).

In this paper, we present the results of a 35 km-long MT survey, crossing a large sector of the Lucanian Apennine and traversing the HAV in correspondence of the two clusters of induced seismic events. The MT profile, oriented at N40 direction and with an investigation depth down to 12 km, has the objective to obtain a resistivity model of the deep structure of our investigated area.

The MT model results have been integrated with a 3-D elastic crustal model, well resolved down to 6–8 km depth and retrieved by also inverting unprecedented seismic data belonging to a very dense seismic network installed in the area (Stabile et al., 2020). The obtained elastic model is described in terms of images of P-wave velocity (V_P) and of the ratio between P-and S-wave velocities (V_P/V_S).