Seismic‐electromagnetic signals from two monitoring stations in Southern Italy: Electromagnetic time series release

Abstract

The seismic‐electromagnetic phenomenon entails the generation of transient electromagnetic signals, which can be observed both simultaneously (co‐seismic) and preceding (pre‐seismic) a seismic wave arrival. Following the most accredited hypothesis, these signals are mainly due to electrokinetic effects, generated on microscopic scale in porous media containing electrolytic fluids. Thus, the seismic‐electromagnetic signals are expected to be suitable for the detection and tracking of crustal fluids. Despite the growing interest in this phenomenon, there is a lack of freely available observational database of earthquake‐related electromagnetic signals recorded at co‐located seismic and magnetotelluric stations. To fill this gap, we set up two multicomponent monitoring stations in two seismically active areas of Southern Italy: the Gargano Promontory and the High Agri Valley. This work is both aimed to systematically analyse earthquake‐generated seismic‐electromagnetic recordings and to make the collected database accessible to the scientific community.

INTRODUCTION

The passage of seismic waves in the subsoil can be accompanied by transient electromagnetic signals, resulting in the so-called seismic-electromagnetic (SEM) effect. The investigation of this phenomenon began in the late 1930s with the first evidence of the coupling between the electromagnetic field and the seismic waves (e.g. Thompson, 1936). Currently, the most accredited hypothesis about the origin of the SEM coupling is explained by the electrokinetic theory rigorously formulated by Pride, 1994 through the coupling of Biot's poro-elastodynamics and Maxwell's electrodynamics equations in partially or fully-saturated porous media. The theory holds that when solid grains are in contact with an electrolytic fluid, a so-called electric double layer (EDL) can be established, consisting of two parallel layers of charge. The first layer is composed of ions (either positive or negative) bound to the grain surface. The second one is a diffuse charge layer composed of freely moving counterpart-ions balancing immobile ions. When a seismic wave propagates in such media, the counter-ions in the fluid start moving relatively to the immobile ions resulting in an electrical streaming current (Haartsen & Pride, 1997).

Three kinds of SEM responses are predicted by the theory.

The first and most common observed is the ’co-seismic field’, travelling along with the seismic wave at the same velocity. In this case, the electric potential generated in the EDL under the oscillating pressure source, changes linearly with it, in amplitude and sign. The electromagnetic effect travels together with the seismic wave and, since the total current is zero, no electric or magnetic field is observed outside the seismic disturbance.

The second effect is known as ‘Interface Responses’ (IRs). It occurs when the seismic wave passes through an interface that separates two porous media with different electrical, hydromechanical and petrophysical properties. In this case, an imbalance in the electric current acts as a source for an electromagnetic independent signal, travelling at high speed and reaching the surface first.

The third is called ‘direct field’, a pure electromagnetic signal directly generated by the seismic source, and it is linked to the relative motion between the fluid and the solid phase in the focal area.

Successful field (e.g. Butler et al., 1996; Dupuis et al., 2007; Strahser et al., 2011) and laboratory experiments on sand or core samples employing active seismic sources (e.g. Bordes et al., 2015; Holzhauer et al., 2017) as well as numerical investigation for layered isotropic media (e.g. Gao et al., 2013; Gao & Hu, 2010; Garambois & Dietrich, 2002; Huang & Ikeya, 1998), have highlighted that the electrokinetic effect is particularly sensitive to the fluid conductivity, to the fluid-pH and to the water-saturation state. Hence the potential application of these signals can be twofold: firstly, they could contribute to the characterization of the fluids hosted in sediments or in reservoir rocks; secondly, they could provide, in a broader scenario, useful tools for early warning strategies (in the case of IRs).

Earthquake-related electromagnetic signals were successfully recorded (e.g. Huang, 2011; Sun et al., 2019) in several seismically active areas: fortuitously during magnetotelluric (MT) surveys (e.g. Honkura et al., 2000; Matsushima, 2002) or during MT monitoring (Balasco et al., 2014), mainly without co-located seismic and MT stations.

Thus, a systematic observational database deriving from co-located seismic and MT stations is necessary to recognize the SEM responses (co-seismic or IRs) and to set constraints between source/site-dependent characteristics and electromagnetic signal detectability. Moreover, a systematic analysis of the multicomponent database allows to define the most appropriate acquisition station set-up.

In the framework of the FLUIDS project, aimed to build up and test the next generation of crustal fluid monitoring systems, two multicomponent monitoring stations were deployed in 2021 in Southern Italy in two seismically active areas, the High Agri Valley and the Gargano Promontory (Figure 1). Hereafter, we will refer to the station belonging to the High Agri Valley Geophysical Observatory (HAVO; ) as TRAMC and to the Gargano Promontory Observatory as GPO. Here we present the experimental set-up of the two stations mentioned above and describe the release of electromagnetic time series containing earthquake-related SEM detected so far.

[IMAGE OMITTED. SEE PDF]

INVESTIGATED AREAS

TRAMC station and GPO are both located in Southern Italy but in different geological contexts, the Gargano Promontory and the High Agri Valley (Table 1 and Figure 1). The Gargano Promontory is the northern part of the Apulia platform, the foreland of the Apennine thrust belt (Bosellini & Morsilli, 1997). It has been hit by strong earthquakes in the past, such as the 1627 (Mw = 6.7 ± 0.1), the 31 May 1646 (Mw = 6.7 ± 0.3) and the 20 March 1731 (Mw = 6.3 ± 0.1) (Del Gaudio et al., 2007). Currently, the area is characterized by micro-seismicity—with a rate of about 400 events/year (Miccolis et al., 2021), by a positive local anomaly of heat flow and hosts a large karst system (Cotecchia, 2014). The High Agri Valley is an intermontane basin of the southern Apennine chain located in the Basilicata region. It is characterized by complex geological setting and by active tectonics, as testified by destructive earthquakes that occurred in this area, such as the Mw 7.0 in 1857 (D'Addezio et al., 2009). The High Agri Valley is also a natural laboratory for fluid investigation, hosting the largest onshore western European oil field (Beaubien et al., 2023) and where fluid-induced seismicity is observed (Stabile et al., 2020).

TABLE 1 General information on the location of GPO and TRAMC stations.

General stations info
Name	Acronym	Location	Latitude (°)	Longitude (°)	Altitude (b.s.l. -m)
Gargano Promontory Observatory	GPO	Stignano	41.71	15.58	279
TRAMutola MultiComponent station	TRAMC	Tramutola	40.29	15.80	894

EXPERIMENTAL SET-UP

The simultaneous measurement of seismic and electromagnetic fields is made possible by the coexistence of seismic and MT stations at the same acquisition site (Figure 2). Specifically, the GPO consists of the seismic station OT04 and the MT station GARG while the TRAMC station consists of the seismic station MARCO and the MT station TRAM (see Table 2 for equipment-specific information).

[IMAGE OMITTED. SEE PDF]

TABLE 2 Detailed information on the equipment of the Gargano Promontory Observatory station (GPO) and TRAMutola MultiComponent station (TRAMC). Both are equipped with seismic and magnetotelluric (MT) stations, for the simultaneous measurement of the two fields (seismic and electromagnetic).

Stations equipment
Station	Seismic	Magnetotelluric
GPO	Name	Name
OT04	GARG
Installation year	Installation year
2013	2021
Seismic Network	Magnetometers
OTRIONS (OT)	MFS-06e (Metronix)
Velocimeter	Electrodes
Lennartz 3D-V	EFP-06e (Metronix)
Data-logger	Data-logger
SL06/SARA	ADU-10e (Metronix)
Sample frequency	Sample frequency
100 Hz	256 Hz
TRAMC	Name	Name
MARCO	TRAM
Installation year	Installation year
2018	2021
Seismic Network	Magnetometers
GEOFON (GE)	MFS-06e (Metronix)
Velocimeter and accelerometer	Electrodes
STS-2.5 and FBA ES-T	EFP-06e (Metronix)
Data-logger	Data-logger
Quanterra330	ADU-08e (Metronix)
Sample frequency	Sample frequency
100 Hz	128 Hz

GPO

The seismicity of the Gargano Promontory is monitored by the seismic network OTRIONS (), installed in 2013 and managed by the University of Bari (Italy). The seismic station OT04 is equipped with a short-period Lennartz 3D-V seismometer and a 24-bit SL06/SARA data-logger. It is powered by a solar panel and data are acquired with a sampling rate of 100 Hz. Seismic waveforms in miniSEED format and station metadata are accessible through the European Integrated Data Archive (EIDA, ) since 2018, thanks to the collaboration between the seismological group of the University of Bari, the ReCaS-Bari datacenter (computing centre of the University of Bari and the National Instituite of Nuclear Physics) and the INGV (for more details, refer to Filippucci et al., 2021).

The MT station was installed in September 2021. The set-up (Figure 3) consists of two orthogonal induction coils MFS-06e (Metronix Geophysics) that acquire the time-varying horizontal magnetic field components (H_x-NS and H_y-EW) and two 50 m electrical dipoles to measure the electric field components (E_x-NS and E_y-EW). The coils are buried in 0.5 m deep trenches while the EFP-06e unpolarizable electrodes are placed in 0.5 m deep drilled holes. Regarding the data-logger, the stations were initially equipped with an ADU-08e from Metronix Geophysics (DE), later replaced by an ADU-10e from the same company for reasons related to power consumption. It is important to point out that the real challenge in monitoring the seismic-electromagnetic phenomenon, besides finding places as far as possible from noise sources, it is the station's energy autonomy. The magnetotelluric stations have higher consumption than the seismic ones, especially for magnetometers, equipped with internal amplifiers and filters. This power consumption is usually minimized by a low sampling rate for long-term acquisitions (e.g. 1–16 Hz). However, the need to use sampling frequencies similar to seismic ones (e.g. 100 Hz) may lead to power issues when using solar panel power supplies, especially from October to April. These problems arose in the initial stages of monitoring when the ADU-08e system was used and were solved with the installation of the ADU-10e system. Indeed, the ADU-10e has a far more lower power consumption than the ADU-08e (2.5 ± 0.98 W, as a peak power consumption, for the ADU-10e and 15.5 ± 0.98 W for the ADU-08e) since it is equipped with neither a display nor an internal GPS and allows the acquisition at only two sampling frequencies (256 Hz or 512 Hz).

[IMAGE OMITTED. SEE PDF]

Both the seismic and the MT stations are controlled remotely, which allows at any time to check the status of the station and recordings as well as to schedule acquisitions. The remote control is guaranteed by a MikroTik SXT R LTE router with a generic data sim and access was provided via specific gateway coinciding with the VPN address of the ReCaS-Bari datacenter.

TRAMC

The seismic station MARCO in the High Agri Valley was placed in 2018 and it is controlled by the German Research for Centre for Geosciences (GFZ—Germany) in collaboration with the Institute of Methodologies for Environmental Analysis (IMAA) of the Italian National Research Council (CNR) (Italy). It is equipped with a Strekeisen velocimeter (STS-2.5) and a three-axial EpiSensor Force Balance accelerometer (FBA ES-T) thus resulting in a six-channel seismic station, all managed by the 24-bit Quanterra330 data-logger with a sampling rate of 100 Hz. This station is powered by a solar panel and batteries and is also equipped with a weather station (Vaisala WXT530) and a GNSS sensor. Waveforms in miniSEED format and station metadata can be downloaded via EIDA platform (as for the OT04 station) or through the Geofon FDSN Web Services ().

The MT station has a much longer monitoring history. It was installed in 2005 for different monitoring purposes (Romano et al., 2014) and then used as a seismic-electromagnetic phenomenon monitoring station (Balasco et al., 2014). Presently, it has a set-up similar to the one of GPO (Figure 3). Up to 2016, the station was equipped with an EMI-Schlumberger EMI-24 MT station and EMI BF-4 induction coils which did not allow any remote control of the system. From 2016 to 2021, a system failure prevented the monitoring activities which restarted in September 2021 when a new Metronix Geophysics magnetotelluric system (ADU-08e equipped with MFS-06e induction coils) replaced the old system. The use of a greater number of solar panels (compared to the GPO) limited, but not solved, the power issues foreseen in 2024 the substitution of the ADU-08e system with an ADU-10e one.

Both the seismic and the MT stations are remotely controlled by using sim cards with static iPs. In this case, the gateway coincides with the VPN address of the router.

DATA DESCRIPTION

Making the TRAMC and GPO stations suitable for MT continuous monitoring has required a long testing phase that has made our acquisitions irregular in time and often mutually asynchronous (Table 3). The SEM expected transient amplitude associated with seismic events dependson earthquake magnitude and hypocentral distance (see Balasco et al., 2014). The transient adds to the natural EM field in the range of 0.1–10 Hz and is characterized by a mean value fluctuation varying between 0.05–0.1 μV/m in the electric field components in both sites. These values act as a threshold for the detectability of SEM transients. All the events that are expected to exceed the threshold were successfully detected by the TRAMC station during its operating period. Conversely, the GPO has not yet recorded any seismic-electromagnetic signal during its operative period and is therefore subject to further analysis.

3

TABLE

Temporal working diagram of the GPO and TRAMC stations, indicating the stations activity per years, months and weeks; colours accord to the legend beside.

Here we describe the released TRAM dataset that consists of segments of electromagnetic time series recorded in correspondence of seismic events. The released dataset () is organized in compressed folders (*.zip, Figure 4a) dedicated to individual events and named with the event date (in ddmmyy format), a rich text format file (*.rtf) with the station coordinates and an Excel file (*.xlsx) file containing information about source earthquakes such as origin time, magnitude, latitude, longitude, depth. In each folder (Figure 4b), the subfolders contain the MT waveforms sampled at different frequencies: 128 Hz (original frequency), 32 Hz, 16 Hz and 8 Hz. The directory structure is the same for all (Figure 4c): for each sampling frequency, there is a control file (*.xml) containing a variety of acquisition information, such as time series duration, date and time, electrode and magnetometer specifications; the time series in original format (*.ats) for each component of the electromagnetic field (E_x,E_y,H_x,H_y) and the corresponding in (*.txt) extension. The released MT signals were recorded during earthquakes that occurred within 70 km radius from the station (2.0 < Mw < 3.5; Figure 5—on the right), with the exception for two events from Greece (Mw = 6.1, 6.3) and one from Bosnia-Herzegovina (Mw = 4.8), which successfully generated SEM transients despite the huge distances (Figure 5—on the left). In the case of the earthquakes within 70 km radius, the time series were cut 10 s before the origin time and had a length of 110 s. In the case of Greek and Bosnian events, the cut is always referred to 10 s before the origin time, but the time series have a duration of 600 s.

[IMAGE OMITTED. SEE PDF]

[IMAGE OMITTED. SEE PDF]

Data analysis

Since electromagnetic time series encompasses a multitude of contributions, the detection of SEM transients cannot be presumed as a given.

This holds particularly true at the sampling frequency used (128 Hz) which enables the acquisition of the transient in its entirety but entails a multitude of electromagnetic sources, such as anthropogenic ones.

For this reason, the optimal workflow we propose consists of two primary steps: applying the Continuous Wavelet Transform (CWT; Torrence & Compo, 1998) and the signal filtering to the time series.

The CWT is suitable for analysing localized power variations in the signal spectrum, which is what the SEM represent. On the one hand, it allows to make detection where it is not possible to observe any signal in the time domain (Figure 6); on the other hand, it allows to establish the spectral content of the transient, useful to choose the most suitable filter for it.

[IMAGE OMITTED. SEE PDF]

It is important to note that the application of the (CWT) is not always necessary for detection, as depicted in the Figure 7: if the SEM amplitude exceeds the variance of the background natural electromagnetic field, it becomes distinctly visible.

[IMAGE OMITTED. SEE PDF]

To perform the CWT, we commonly utilize the ‘wavelets’ package released by Torrence and Compo (). As for signal filtering, our approach involves applying a Finite Impulse Response (FIR) filter (Shenoi, 2005) using the MatLab© ‘filtfilt’ function.

A more detailed explanation and an example of application of this workflow can be found in Romano et al., 2018.

CONCLUSIONS

Two monitoring stations (GPO and TRAMC) in Southern Italy were set up to detect seismic-electromagnetic signature. These two stations are currently operating in the Gargano Promontory and in High Agri Valley, respectively. The dataset here described and released refers to the electromagnetic time series recorded by the TRAM station. This phenomenology in which seismic and electromagnetic fields are coupled is closely linked to the presence of fluids in the subsoil and is therefore a potentially useful tool for the detection and tracking of crustal fluids. The development of analytical and numerical models on the phenomenon, however, is not accompanied by an adequate sample of real data that can clarify the detection characteristics, the dependence on the measuring site and the possible use of these signals. We decided to publish this dataset in order to start filling this gap.

ACKNOWLEDGEMENTS

We want to thank the reviewers for their helpful and constructive comments.

FUNDING INFORMATION

This research has been partially supported by the project ‘Detection and tracking of crustal fluid by multi-parametric methodologies and technologies’ of the Italian PRIN-MIUR programme (grant no. 20174X3P29) and also carried out within the RETURN Extended Partnership and received funding from the European Union Next-GenerationEU (National Recovery and Resilience Plan—NRRP, Mission 4, Component 2, Investment 1.3—D.D. 1243, 2/8/2022, PE0000005).

OPEN RESEARCH BADGES

This article has earned Open Data, Open Materials and Preregistered Research Design badges. Data, materials and the preregistered design and analysis plan are available at [].

DATA AVAILABILITY STATEMENT

Publicly available datasets were analyzed in this study. Electromagnetic time-series can be found here: , while the corresponding seismic waveforms can be found at .

References

Balasco, M., Lapenna, V., Romano, G., Siniscalchi, A., Stabile, T.A. & Telesca, L. (2014) Electric and magnetic field changes observed during a seismic swarm in Pollino area (southern Italy). Bulletin of the Seismological Society of America, 104(3), 1289–1298. Available from: [DOI: https://dx.doi.org/10.1785/0120130183]