Swarm Investigation of Ultra-Low-Frequency (ULF) Pulsation and Plasma Irregularity Signatures Potentially Associated with Geophysical Activity

Abstract

Launched on 22 November 2013, Swarm is the fourth in a series of pioneering Earth Explorer missions and also the European Space Agency’s (ESA’s) first constellation to advance our understanding of the Earth’s magnetic field and the near-Earth electromagnetic environment. Swarm provides an ideal platform in the topside ionosphere for observing ultra-low-frequency (ULF) waves, as well as equatorial spread-F (ESF) events or plasma bubbles, and, thus, offers an excellent opportunity for space weather studies. For this purpose, a specialized time–frequency analysis (TFA) toolbox has been developed for deriving continuous pulsations (Pc), namely Pc1 (0.2–5 Hz) and Pc3 (22–100 mHz), as well as ionospheric plasma irregularity distribution maps. In this methodological paper, we focus on the ULF pulsation and ESF activity observed by Swarm satellites during a time interval centered around the occurrence of the 24 August 2016 Central Italy M6 earthquake. Due to the Swarm orbit’s proximity to the earthquake epicenter, i.e., a few hours before the earthquake occurred, data from the mission may offer a variety of interesting observations around the time of the earthquake event. These observations could be associated with the occurrence of this geophysical event. Most notably, we observed an electron density perturbation occurring 6 h prior to the earthquake. This perturbation was detected when the satellites were flying above Italy.

1. Introduction

The ongoing Swarm mission of the European Space Agency (ESA) provides an opportunity for more detailed knowledge of the Earth’s magnetic field and the near-Earth electromagnetic environment. Swarm is the first constellation mission of ESA’s Earth Explorers [1]. The constellation consists of three identical satellites, a lower pair (A and C), which are positioned “side by side” with a nominal 150 km separation at the equator and a 4–10 s along-track separation to avoid collision at the poles, and an upper satellite (B) in a nominally different local time orbit. The final configuration of the three-satellite mission, with two spacecraft (Swarm A and C) flying side-by-side at a low altitude (∼460 km) and one spacecraft (Swarm B) flying at a slightly higher altitude (∼510 km), was achieved on 17 April 2014. For our analysis, we used magnetic field data from the vector fluxgate magnetometer (VFM) on board the three Swarm spacecraft, at 1 Hz cadence. We also processed 2 Hz electron density data from the electric field instruments (EFIs) on board the three satellites. This offers the possibility of investigating ultra-low-frequency (ULF) wave events [2,3]. One of the tools used to analyze these events is time–frequency analysis (TFA). This is a processing tool established for deriving Pc1 (0.2–5 Hz) and Pc3 (22–100 mHz) wave indices [4], but it is also able to detect geophysical signals, due to, for example, plasma instabilities, indicating phenomena such as post-sunset equatorial spread-F (ESF) events or plasma bubbles (see, for instance, ref. [5,6]) and artificial disturbances (anomalies), e.g., spikes and jumps. From a space weather-based point of view, plasma bubbles play a critical role as they can affect global navigation satellite system (GNSS) signals (see, for instance, ref. [7,8]).

ESF events are recognized by sudden depletions in high-electron-density environments. They display a clear seasonal and longitudinal variation in occurrence rates, peaking symmetrically around the dip equator at approximately ±9.5° magnetic latitude, and extending to ±20° N [5,9], or possibly even more, depending on factors like solar activity and geomagnetic conditions. ESF events demonstrate a linear relationship with solar extreme ultraviolet (EUV) flux, due to the direct influence of EUV on ionization, thermospheric dynamics, electric fields, and seed perturbations, all of which are critical drivers of ESF activity, and a weak correlation with geomagnetic activity, as indicated by the Kp geomagnetic index (https://www.gfz-potsdam.de/en/section/geomagnetism/data-products-services/geomagnetic-kp-index, accessed on 6 September 2024). Their distribution along longitudes varies with absolute electron density and geomagnetic conditions, with increased occurrence during equinoxes (March and September) and variation during solstices (June and December). High occurrence rates are observed over African and Pacific regions during June, but there is suppression in the Atlantic–American sector and minimal depletions over India [5,9]. In December, the highest ESF rate is over the Atlantic–American region, decreasing westward and eastward. Despite the seasonal variations, ESF events can occur throughout the year, with fluctuations attributed to solar cycle variability. Post-sunset equatorial plasma bubbles (EPBs) contribute to ionospheric irregularities, affecting GNSS signals and other communication systems which rely on quiet ionospheric conditions [7,8]. These irregularities are primarily triggered by the Rayleigh–Taylor instability, which enhances the gradient at the F-layer’s bottom side. However, despite the well-established climatological patterns of EPBs, their predictability remains challenging due to day-to-day variability and storm-time behavior.

A few previous studies have suggested that ULF magnetic fields may be associated with earthquakes [10,11]. The majority of these studies refer to the detection of these signals using ground-based magnetometer measurements ([12,13] and references therein). Ground-based electromagnetic observations at several different bands have provided multiple pieces of evidence of possible earthquake-related anomalies at various lead times. Among them, the most intensively investigated are as follows: (a) seismic electric signals (SESs) (<1 Hz), which are geoelectric signals ([14] and references therein); (b) ULF magnetic field anomalies, some of which are attributed to direct emissions from the lithosphere, whereas others are likely to be related to seismogenic perturbations in the lower ionosphere (depression in the horizontal ULF magnetic field), e.g., [12,13,15,16]; (c) ULF/extremely low- frequency (ELF) (3 mHz–40 Hz) atmospheric electromagnetic radiation ([16,17] and references therein); (d) fracture-induced electromagnetic emissions (FEME) (also known as fracture-induced electromagnetic radiation, FEMR), successively emerging in the MHz and kHz bands ([18,19,20] and references therein); (e) very low frequency/low frequency (VLF/LF) sub-ionospheric propagation anomalies [21,22]; and (f) ionospheric anomalies using ionosonde sounding, e.g., [23,24].

On the other hand, there had only been a handful of studies that attempted to correlate ULF pulsations with seismic activity from spaceborne measurements before the Swarm era, e.g., [25] for the CHAMP satellite and [26] for the DEMETER satellite. Following the launch of the Swarm satellites, there has been an increasing number of studies (e.g., [27,28] and references therein) on the near-Earth electromagnetic environment associated with the lithosphere–atmosphere–ionosphere coupling (LAIC). For instance, De Santis et al. [27] analyzed Swarm data around the time of the Nepal earthquake (M7.8), which occurred on 25 April 2015. The comparison of the Pc3 wave power levels during the same two-month period over three years (the earthquake year, together with the previous year and the following year) showed clear differences in the region around the earthquake’s epicenter. In [28], the authors performed an intensive and systematic analysis of Swarm satellite magnetic field and electron density measurements all over the world from 1 January 2014 to 31 August 2018, and compared the results with M5+ earthquakes occurring in the same period. Applying the superposed epoch and space approach, they found a robust statistical correlation between pre-earthquake anomalies and earthquakes. The occurrence of anomalies were in almost all pre-earthquake period of 90-day investigation, and some occurred even on the same day of the earthquake. This paper also confirmed the Rikitake law [29], relating the earthquake’s magnitude and the precursor time, a law that was initially found to apply to ground data, including electromagnetic data. De Santis et al. [30] applied a multiparametric and multilayer approach to the study of the 6 July 2019 M7.1 Ridgecrest earthquake. Many parameters, from the lithosphere to the atmosphere and the ionosphere, were used. Regarding the magnetic field measured by the Swarm satellites, some groups of anomalies were detected, in particular 150, 70–65, and 35–25 days before the earthquake. Interestingly, there was an astonishing agreement in the cumulative number of lithospheric anomalies and of atmospheric and ionospheric anomalies, supporting the hypothesis that the lithospheric activity was the main driver of atmospheric and ionospheric anomalies. Both [29,30] are very good examples that the Swarm satellite magnetic field and electron density measurements can complement ground-based observations, underlying the great importance of this satellite data analysis.

In this study, we focus on the ULF pulsation and ESF activity observed by Swarm satellites during a time interval centered around the occurrence of the August 2016 Central Italy earthquake, which had a 6.0 moment magnitude and hit Central Italy on 24 August 2016. Because the Swarm’s orbit was in close proximity to the earthquake epicenter, with Swarm crossing Italy approximately 6 h before the earthquake occurred (see Figure 1), data from the mission may offer valuable observations from that time. These observations could be associated with the occurrence of this geophysical event.

This paper is organized as follows. Section 2.1 presents the data from and the geomagnetic activity during the time interval considered, while Section 2.2 describes the characteristics of the tool set applied to analyze these data. The rest of the paper presents the obtained results from applying the TFA toolbox to data around the time of the Central Italy Earthquake (Section 3) and finishes with some discussions and conclusions (Section 4).

2. Materials and Methods

2.1. Data Description and Geomagnetic Activity Conditions

The August 2016 Central Italy earthquake (epicenter: 42.71° latitude, 13.17° longitude) had a 6.0 moment magnitude and hit Central Italy on 24 August 2016 at 03:36:32 central European summer time (CEST) (01:36 UTC), displaying a focal mechanism consistent with extensional kinematics and a minimum stress axis oriented perpendicular to the Apenninic chain’s axis [31]. Its epicenter was close to Accumoli, with its hypocenter at a depth of 4 ± 1 km, approximately 75 km southeast of Perugia and 45 km north of L’Aquila, in an area near the borders of the Umbria, Lazio, Abruzzo, and Marche regions.

In this study, we focus on the ULF pulsation and ESF activity observed by Swarm satellites. The temporal window we have chosen is centered around the seismic event, with a fortuitous alignment of Swarm satellites passing over the geographical location of the earthquake epicenter a few hours prior to the earthquake. Such synchronous positioning of the Swarm constellation above the affected region is an infrequent circumstance and the analysis of the specific datasets could provide valuable insight regarding irregular signal signatures that precede strong geophysical activity.

Regarding Swarm magnetic field measurements, the mission offers 1 Hz absolute scalar magnetometer (ASM) data, as well as 1 Hz and 50 Hz vector field magnetometer (VFM) data. Furthermore, there is an ASM burst mode that offers 250 Hz ASM data for selected time intervals. Regarding Swarm in situ ionospheric electric field and plasma measurements [32], the mission offers 2 Hz electron density data collected from the Langmuir probe component of the electric field instrument (EFI). Regarding wave investigation with Swarm, the mission can detect ULF waves (at the Pc1 and Pc3 frequency ranges), but also the ELF portion of whistler waves (10–120 Hz band) detected during ASM burst mode sessions [33]. The Swarm TFA toolbox could, in principle, investigate all the above-mentioned types of waves detected with instruments onboard the mission (see also Section 2.2 for more details). Therefore, an inherent limitation of the tool may be considered that it is bound to analyze specific frequency ranges due to the cadence constraints imposed by the mission payload.

Figure 1 has been produced using the VirES web interface (https://vires.services, accessed on 6 September 2024). VirES for Swarm is an advanced platform tailored for the effective manipulation and retrieval of data pertinent to Swarm. VirES includes tools for studying various geomagnetic models by comparing them to real-time measurements obtained from the Swarm satellites, providing a comprehensive understanding of Earth’s magnetic field dynamics under specific space weather and ionospheric conditions. In this specific 3D visualization of the globe, the tracks of the three Swarm satellites are shown, with the lower pair of satellites flying above Italy right before (i.e., approximately 6 h before) the occurrence of the August 2016 earthquake. Specifically, the track expands from approximately 19:15 UTC to 20:00 UTC on 23 August 2016. In Appendix A, supplementary to Figure 1, we present additional satellite tracks observed before and after the depicted trajectory of Figure 1. These supplementary tracks serve to delineate the longitudinal positions of the satellites relative to the Earth’s surface and to evidence the uniqueness of the track over the epicentral region.

Regarding the geomagnetic activity conditions around the time of the earthquake, we have examined the hourly disturbance storm-time (Dst) index (https://wdc.kugi.kyoto-u.ac.jp/dstdir/, accessed on 6 September 2024). The Dst index denotes the mean variation in the horizontal component of Earth’s magnetic field, as observed at four mid-latitude magnetic observatories. This variation is attributed to enhancements in the magnetospheric ring current encircling the Earth, thereby serving as a proxy of magnetic storm intensity. In Figure 2, we show the Dst index for the month of August 2016. On 23 August 2016 at 22:00 UTC, the Dst index reached a minimum value of −73 nT, indicating a moderate magnetic storm [34]. This event was recorded a few hours (i.e., approximately 3.5 h) before the occurrence of the earthquake in Central Italy.

2.2. The Swarm Time–Frequency Analysis (TFA) Toolbox

For the self-consistency of this paper, let us provide a few details about TFA. The TFA tool is a MATLAB code that includes a graphical interface, as well as a dedicated back end, that can be used to perform wavelet analysis and visualize the results for both Pc1 and Pc3 waves, using both Swarm magnetic field Level 1b 50 Hz and 1 Hz data [4]. Following recommendations from the Advisory Board of the Swarm Data, Innovation and Science Cluster (Swarm DISC), the tool has been further developed. It has also been generalized to accommodate the analysis of other types of time series from both satellite and ground station measurements. In particular, the resulting TFA toolbox provides users with the ability to study different wave types (e.g., compressional waves, Alfvén waves, etc.), various magnetic field components (e.g., mean field aligned (MFA) coordinates), and other geophysical measurements (e.g., electric field and plasma parameters). The TFA toolbox is also able to detect ESF signature events and artificial noise. It is also possible to analyze data from ground stations in a consistent format using TFA, e.g., 1 Hz magnetic observatory data, as available in the virtual research service, VirES for Swarm. Moreover, parts of the TFA toolbox have been integrated into the Python SwarmPAL module (https://swarmpal.readthedocs.io/, accessed on 6 September 2024) and are continuously being developed and upgraded.

Figure 3 shows the graphical interface of the MATLAB version of the tool, with the input parameters displayed as drop-down menus on the left panel of the screen and the results of the analysis presented in the right panel. The typical outputs are composed of three rows of plots; the top one comprises the time series of the (processed and filtered) magnetic field, the middle comprises the image of the wavelet power spectrum for the given range of frequencies, and the bottom row plots the electron density (green line) and the satellite’s magnetic latitude in the same graph. All panels correspond to the same time interval. Additional columns can be used to plot the analysis results of other components of the magnetic field from the same satellite, or the same component (or total field) from another satellite of the Swarm constellation.

Examples of the four types of events identified by the TFA tool are shown in Figure 4. A true ULF event will resemble the image in the top left panel, where strong wavelet power is displayed continuously for a long time interval, while also being bounded from top and bottom, i.e., having the entirety of its activity within the frequency range that was indicated. ESF events have a similar signature in the wavelet images, but they are typically composed of two parts, which are symmetric around the magnetic equator and often look like mirror images of one another. Their detection usually requires additional information from the electron density time series that exhibits rough variations during such events. One example is shown in the top right panel of Figure 4. Other types of signals are the ones produced due to spikes or discontinuities in the data. These produce short, broadband signatures in the data that are easily identified as such and are classified as artificial signals that should not be mixed with other signals of physics-based origins (bottom left panel). Lastly, when no wave activity of any kind is present, the wavelet spectra are characterized as background noise (bottom right panel of Figure 4).

The user provides as inputs a start and end date/time, a chosen satellite, and a selected component of the magnetic field (or its total vector magnitude). The magnetic field data are retrieved automatically from ESA’s Swarm File Transfer Protocol (FTP) repository, not only for the specified interval, but also for a few hours prior and after it. This is necessary because the wavelet analysis, like most spectral methods, requires the analyzed signal to be periodic. If it is not, then edge effects emerge at the beginning and end of the analyzed interval that need to be removed, which in this case is achieved by introducing a few extra hours of signal that, in the end, can be safely removed from the process. After the data retrieval, the CHAMP, Ørsted, and Swarm field model (CHAOS) internal part [35] is subtracted from the data, to keep only the magnetic field component that is caused by space weather phenomena, and the new residual time series is high-pass filtered to remove the longer time-scale variations and bring into focus the requested frequency range (Pc3, Pc2 or Pc1). It is on this filtered signal that the wavelet transform is applied [36] and the wavelet power (squared magnitude) is plotted in the time–frequency domain. Ancillary data like the satellite’s magnetic latitude and the electron density are also retrieved, for the same time period, and plotted along with the wavelet spectrum. After the analysis is complete, the tool provides an additional option to view the results in a track-by-track fashion, i.e., to see the plots that correspond to each half orbit of the satellite, navigate to the next/previous one and save all such figures on disk.

3. Results

In Figure 5, Figure 6, Figure 7 and Figure 8 we present the wavelet spectra of the Swarm magnetic field recordings, along with Swarm electron density variations analyzed using the TFA toolbox. Figure 5, Figure 6, Figure 7 and Figure 8 represent four ascending satellite tracks covering the time interval from 17:41 to 23:06 UTC on 23 August 2016. The considered time interval includes a track when Swarm A and C satellites were flying above Italy, i.e., 6 h before the occurrence of the 24 August 2016 Central Italy earthquake. Six more Swarm ascending tracks, analyzed using the TFA toolbox, are presented in Appendix B: three tracks before and three tracks after this time window.

Figure 5 presents a Swarm TFA plot for a full satellite track on 23 August 2016 from 17:41 to 18:27 UTC, before the Central Italy earthquake that occurred on 24 August 2016 at 1:36 UTC. In this track, Swarm A and C fly over Turkey (c.f. Figure A1). Swarm B, A, and C (from left to right) show the filtered series of the magnetic field magnitude (top panels), their corresponding wavelet spectra for the joined Pc3 and Pc4 range (middle panels), and a composite plot of the measured electron density (green line) and their location at magnetic latitude (blue line) from −90° to +90° (bottom panels). Apart from the elevated wavelet power over the north and south poles, no other activity is observed for this track (please note that, for this and the following tracks, electron density data for Swarm B are unavailable).

Figure 6 is similar to Figure 5, but for a Swarm track on 23 August 2016 from 19:14 to 20:00 UTC, approximately 6 h before the earthquake. This is the track wherein Swarm A and C satellites fly over Italy (c.f. Figure 1). Elevated power in the wavelet spectra of the magnetic field is observed in (at least) four distinct areas (including the poles), with more prominent activity for Swarm A and C. Simultaneously to the magnetic measurements, perturbations of electron density are observed at low latitudes only for Swarm C, which is flying closer to the area of earthquake epicenter in comparison to Swarm A.

Figure 6 depicts the trajectory followed by Swarm A and C satellites over Italy. Perturbations in electron density are discernible at lower latitudes (i.e., two peaks roughly at −10° and +15° around the equator) however are evident solely in Swarm C data. These observations potentially indicate an ESF event, typically occurring within ±20° of the geomagnetic equator. The geographic longitudes associated with this event are approximately 11.85° for Swarm A and 13.27° for Swarm C, whereas the geographic longitude of the earthquake epicenter stands at 13.17°. This likely ESF event transpired in late August, nearly a month preceding the autumnal solstice. As mentioned before, during equinoxes, characterized by maximal solar heating at the equator, ionospheric dynamics are intensified, amplifying the likelihood of ESF occurrences. Remarkably, solar cycle 24 peaked in April 2014, marking the current phase as its declining period. Although periods of heightened geomagnetic activity, such as geomagnetic storms, can perturb the ionosphere and potentially enhance ESF activity, the relationship between geomagnetic activity and ESF events is intricate and not strictly linear.

Figure 7 is similar to Figure 5, but for a Swarm track on 23 August 2016 from 20:46 to 21:32 UTC, approximately 4.5 h before the earthquake. However, Swarm A and C satellites fly over the Atlantic ocean (c.f. Figure A2). For Swarm A and C, we observe simultaneous post-sunset perturbations both for the magnetic field and the electron density data at low latitudes. The latter indicates ESF activity. Please note that this track ends approximately half an hour before the Dst index reaches its peak value for the magnetic storm shown in Figure 2.

In Figure 7, simultaneous post-sunset perturbations in both magnetic field and electron density data are observed at lower latitudes for both Swarm A and C satellites, clearly indicating ESF activity. Analogous to the observations in Figure 6, deviations in electron density are observed within the 0° to +20° range, characteristic of ESF occurrences. The geographic longitudes traversed by Swarm A and C during this event are approximately 348.2° and 349.6°, respectively. As previously noted, this track is close temporally to the time of the Dst peak value of the magnetic storm, thus reflecting a heightened geomagnetic activity as well.

Figure 8 is similar to Figure 5, but for a Swarm track on 23 August 2016 from 22:19 to 23:06 UTC. This track is closer to the time of the earthquake; however, the satellites fly at different distances from the epicenter (c.f. Figure A3). There is an increased wavelet power spectral density; thus, elevated ionospheric activity is observed only at the poles. This track is also very close temporally (approximately 20 min after) to the time of the minimum Dst index value of the moderate magnetic storm that occurred.

4. Discussion and Conclusions

For over more than a decade, the Swarm mission has greatly enhanced our understanding of the physics of the topside ionosphere and of magnetosphere–ionosphere coupling. This includes their corresponding space weather effects.

In this methodologically driven paper, we employ the Swarm time–frequency analysis toolbox to investigate the ultra-low-frequency magnetic activity, as well as the electron density variations, in the ionosphere. Our study focuses on the period around the time of the August 2016 Central Italy earthquake, during which the lower pair of Swarm satellites was crossing Italy. We find several interesting observations of perturbations, including the occurrence of an electron density disturbance at low latitudes in Swarm C’s time series, 6 h prior to the earthquake.

Regarding the lead time of a possible earthquake-related electromagnetic signal, we have a large variety of lead times reported in the literature both for ground-based and spaceborne observations. For instance, regarding ground measurements, the SESs, which are accompanied by magnetic field variations in the field vertical component, present a lead-time ranging from a few months to a few weeks before an earthquake’s occurrence [14]. Statistical anomalies have been detected from 1–2 months to a few weeks and 2–4 days before an earthquake for the direct ULF magnetic field power and ULF depression, respectively, whereas phase transition characteristics in the raw ULF magnetic field recordings have been identified even a few hours prior to an earthquake’s occurrence ([17,37] and references therein). The ULF/ELF pre-seismic anomalies have a lead time of 1 week to 2 days and the VLF/LF sub-ionospheric propagation anomalies have a lead-time of 1 week to 2 days, whereas ionosonde-detected anomalies have been reported as having lead times from 40 days to a few hours [38]. Finally, the MHz FEME/FEMR present a lead time ranging from 2 weeks to a few hours, whereas the kHz FEME/FEMR appear, after the MHz FEME/FEMR, a few days to a few hours before an earthquake’s occurrence ([19,20] and references therein).

Let us discuss some of the previously analyzed events and the different conclusions reached using satellite measurements. Marchetti and Akhoondzadeh [39] analyzed Swarm satellite data, including magnetic and electric fields and electron density, on the occasion of the 8 September 2017 M8.2 Mexico earthquake. Around 200 magnetic field anomalies were found in the 5 months before the earthquake. Constructing the cumulative number of these anomalies, an S-shaped pattern was revealed, with an inflection point about two months before the earthquake. Akhoondzadeh et al. [40] considered several kinds of atmospheric and ionospheric precursors, including Swarm magnetic field data, to the 2016 M7.6 Ecuador earthquake. The analysis of the magnetic field scalar and vector data show considerable anomalies from around 50 to 7 days before the main shock. De Santis et al. [41] analyzed Swarm satellite magnetic field and electron density data one month before and one month after 12 strong earthquakes that occurred in the first 2.5 years of the Swarm satellites’ mission lifetime in the Mediterranean region (magnitude M6.1+) or in the rest of the world (M6.7+). They then detected the anomalies using an objective method. Once the anomalies are normalized by the analyzed satellite tracks, they resemble a linear dependence with earthquake magnitude, so supporting a statistical correlation between anomalies and earthquakes and excluding a relationship by chance. In some cases, the anomalies were found also on the day of the earthquake, in particular for the 2015 M7.8 Nepal, the 2016 M7.9 Sumatra, and the 2016 M7 Japan earthquakes.

In Marchetti et al. [42], anomalous magnetic field signals were detected by Swarm satellites during geomagnetically quiet conditions and during major seismic events, from about 4 months before the start of the seismic sequence to the first 8 months after the seismic sequence (i.e., a total of one year of analyzed data). These authors found a very interesting increase in such anomalies, starting about 40 days before the beginning of the seismic sequence, coinciding with and following surface and atmospheric alterations, resulting in a temporal sequence of anomalies from the Earth’s surface up to the ionosphere. Akhoondzadeh et al. [43] analyzed three months of global positioning system (GPS) total electron content (TEC) data and four months of Swarm satellite magnetic field, electron density, and temperature data around the time of the 12 November 2017 M7.3 Iran–Iraq earthquake. Regarding the satellite magnetic field anomalies, they found tens of anomalies in the investigated period and reconstructed the cumulative number of anomalies, searching for an S-shaped pattern. An anomaly in total intensity was also found on the day of the earthquake, just before it occurred. Marchetti et al. [44] extended the analysis of Swarm satellite magnetic field data to 1000 days before the 2019 M7.1 Ridgecrest earthquake. They found that some geomagnetic field activity occurred 222 to 168 days before the earthquake, which was anomalous when compared to another confutation area. In addition, the Swarm B satellite passed above the epicentral area 15 min before the earthquake and detected a clear anomaly, mainly in the field east component.

Sasmal et al. [45] analyzed several parameters, including Swarm magnetic field data, during the 30 October 2020 M6.9 Samos (Greece) earthquake. They found a clear anomaly in the vertical component the day before the earthquake. Akhoondzadeh et al. [46] studied several kinds of potential ionospheric precursors before the 14 May 2019 M7.6 Papua New Guinea earthquake, including Swarm magnetic field satellite data. A few tens of magnetic field anomalies were found with a typical S-shape pattern in the four months before the earthquake. Ghamry et al. [47] also found some Swarm magnetic field data anomalies 12 and 13 days before the M7.8 25 April 2015 Nepal earthquake. Zhang et al. [48] studied the pre-occurrence phase of four strong Chinese earthquakes occurring in 2008–2020. They used several ionospheric data from ground and space, including Swarm satellite magnetic field data, for the 2014 and 2020 Yutian earthquakes, and found clear anomalies from two weeks to a week before each earthquake and a few just after. Zhang et al. [49] also analyzed the pre-occurrence phase of the 2022 M6.8 Luding (China) earthquake, by means of a multiparametric and multilayer approach, from ground to atmosphere and ionosphere. Regarding Swarm magnetic field data, they found some anomalies 83, 68, 67, 64, 12, and 11 days before the earthquake. Considering all anomalies with comprehensive accumulation, they recognized a two-way model of LAIC: a slow accelerating trend, which they called the diffusive–thermodynamic coupling, and an intermittent direct coupling, which was probably electromagnetic. Ozsoz et al. [50] analyzed Swarm satellite magnetic field data six months before and one month after five M6+ earthquakes in Turkey and Greece in 2017–2020, finding a few tens of anomalies (they found them when considering data from all three satellites, so the real number of anomalies would be actually one-third of that total) in the period of study, wherein some earthquakes showed a typical precursory S-shaped pattern (as found for the first time by De Santis et al. [27]).

Interestingly, Heki et al. ([51] and references therein), analyzing the total electron content, i.e., the total number of electrons in the column from the Earth to the satellites (mostly due to the ionospheric electrons), found total electron content anomalies some tens of minutes before a large earthquake; these were times comparable with those of the anomalies found in this work. Notably, our analysis reveals that, preceding and subsequent to the trajectory in question, the satellites traversed disparate longitudinal coordinates. However, during the temporal window corresponding to the capture of the electron density signal, the lower pair of Swarm satellites was coincidentally positioned directly above the geographical region of Italy in question. This fortuitous alignment underscores the spatial–temporal synchronicity between satellite positioning and the occurrence of electron density anomalies, reinforcing the relevance of our findings within the context of geophysical activity and earthquake occurrences.

Our methodology is data-driven, and an anomaly is objectively defined when it is clearly different from the background. We show what we find as something "peculiar” that happens before the M6 earthquake. Although we cannot fully trust it is earthquake-related, there are great chances it is.

It is true that, with some physical consideration relating to the earthquake’s magnitude, the size of its preparation area, and the speed of the satellite, a typical anomaly should last about 4–5 min, i.e., a little less time than what we have found. Indeed, according to the Dobrovolksy relationship, the size of the Central Italy earthquake’s preparation area was approximately 623 km (in radius) or 1250 km (in diameter), since the seismic sequence culminated with an M6.5 earthquake on 30 October 2016. As the satellite has a typical speed of 7.4 km/s, the time the anomalous area can interact with the ionosphere is of the order of 1250/7.4 = 169 s, which is about 3 min. Considering also the slight enlargement of the Dobrovolsky region when projecting it from the Earth’s surface to about 500 km above, in the ionosphere, the larger anomaly cannot last more than 4 min, which is still less than 5–10 min, as we have found. However, we cannot discard a possible distorting and spreading effect of the geomagnetic field and of the atmospheric dynamics, therefore justifying an anomaly lasting 5–10 min.

Another source of concern about the internal geophysical origin of the peculiar signal in the Swarm C time series of electron density data could be that a small concurrent geomagnetic storm was recorded (Figure 1), with the likely source being coronal hole 756, along with a disappearing solar filament. The arrival time of this disturbance is estimated to be around 19:00 on 23 August 2016, which corresponds to the time when the main features in the wavelet spectra of the magnetic field occurred (Figure 6). However, if the peculiar signal had an external origin, it would have been observed simultaneously in both the Swarm A and C time series of electron density data, which is clearly not the case.

The results obtained here pave the way for exploring other types of events using satellite data, as ionospheric processes and the space-based detection of natural hazards continue to be a multidisciplinary research area. The short- and long-term prospects are promising, even though our current understanding of the coupling between the lithosphere, atmosphere, and ionosphere remains limited. This applies not only to the generation of co-seismic and co-volcanic ionospheric disturbances, which are of particular interest, but also to other solid Earth phenomena, such as slow-slip earthquakes and landslides. To enhance our understanding of this complex coupling, it is essential to investigate the formation mechanisms of these ionospheric disturbances. Moreover, a deeper study of how this coupling varies with solar activity levels, atmospheric conditions, and other factors is necessary. In terms of observations, combining electromagnetic measurements with other data, such as high-resolution GNSS or gravity data, is crucial. This combination could provide new insights into the generation and evolution of ionospheric disturbances caused by natural hazard events and how they develop with altitude.