Characteristics of Recent Aftershocks Sequences (2014, 2015, 2018) Derived from New Seismological and Geodetic Data on the Ionian Islands, Greece

: In 2014–2018, four strong earthquakes occurred in the Ionian Sea, Greece. After these events, a rich aftershock sequence followed. More analytically, according to the manual solutions of the National Observatory of Athens, the first event occurred on 26 January 2014 in Cephalonia Island with magnitude M L = 5.8, followed by another in the same region on 3 February 2014 with magnitude M L = 5.7. The third event occurred on 17 November 2015, M L = 6.0 in Lefkas Island and the last on 25 October 2018, M L = 6.6 in Zakynthos Island. The first three of these earthquakes caused moderate structural damages, mainly in houses and produced particular unrest to the local popu-lation. This work determines a seismic moment tensor for both large and intermediate magnitude earthquakes (M > 4.0). Geodetic data from permanent GPS stations were analyzed to investigate the displacement due to the earthquakes. local faulting system. This evidence is getting stronger by Moment Tensors studies, which are the same at both events.


Introduction
A detailed analysis of both seismological and geodetic data of the Ionian Islands is performed in this study. It is the first time that such an extensive time period analysis with the latest state-of-the-art software and usage of updated data like the ITRF 2014 reference frame has been applied to Greece's most seismogenic area. This area is characterized by continuous seismic activity and frequent occurrence of large earthquakes (M > 6.0). The tectonic structure of the islands of Cephalonia and Ithaca results from the effect of compressive stresses in which periods of tensile stresses are inserted [1].
The first seismological indications for horizontal sliding movements in the study area were developed in the 1980s based on epicenters [10] and focal mechanisms of strong earthquakes and aftershocks of 17/01/1983 M = 7.0 and 23/03/1983 M = 6.2, [11,12]. Further study by many scientists confirmed the effect of the horizontal slip fault in the area using synthetic waveform modeling [13], with micro-seismic studies [14] and geodetic measurements [15,16]. The results from the micro-seismic [14,17] and geodetic measurements [15,16] led to the conclusion that this regime extends to the north in the area of Lefkas. Recent studies based on precise epicenters and focal mechanisms confirm the existence of this seismically active fault of Cephalonia that runs parallel to the west coast of Lefkas [3].
The complex seismotectonic structure that controls the wider area's seismicity is the Greek arc section's subsidence zone in the Ionian Sea area below Cephalonia [18]. Figure 1 represents a general structure map showing the prefectures of the broader region of Western Greece. For the study area (Ionian Islands), the portable seismological station sites' selection criterion was based on the existence and operation of permanent seismological and geodetic stations that had already been installed. Faults were collected from the international literature and digitized by ArcGIS Pro.
Aftershock sequences' statistical processing is an essential tool for understanding the earthquake process's focal mechanisms. Changes in the rate of seismicity during seismic sequences indicate precursors to the generation of a strong earthquake during the sequence [19][20][21][22]. Detailed seismological data of these large events and the strongest aftershocks were used in this work: The first and significant part focuses on the main event's extensional focal mechanisms and the comparison with those calculated from other institutions. In the second part, using GPS data and calculations of the GPS time series, the coseismic deformation is estimated. The above combination produces precise conclusions about unique seismic sequence characteristics.  [23]. Bathymetry is from Emodnet Bathymetr, [23] and DEM. Copernicus Land Monitoring Service, [24] red triangles represent the GNSS network stations, green triangles represent the stations of the portable seismograph network installed immediately after the main earthquake, yellow triangles indicate the permanent stations of the Unified Seismological Network (HUSN), and finally, the three yellow stars indicate the historical earthquake in the region of Cephalonia.

Data
In this section, seismological data were used to calculate the seismic parameters of both main earthquakes as well as the major aftershocks of them. The data were collected, analyzed, and processed using an appropriate methodology, analyzed in the next section, and in more detail in [25,26]. GNSS data analysis was performed along with the seismic data to calculate the displacement caused by seismic vibrations and compare it with seismic solutions.

Seismological Data
In this study, seismological broadband data from the Hellenic Unified Seismological Network were collected and analyzed to determine the source parameters of the events in the Ionian Sea. For this purpose, 13 broadband stations were equipped with three components seismometers from the Hellenic Unified Seismological Network (HUSN); detailed information is presented in Table 1.

Geodetic Data
The GPS data (30-s sampling interval) were collected and processed from stations VLSM (Valsamata Cephalonia), SPAN (Spanohori Lefkas), PONT (Ponti Lefkas) shown in Figure 1, which belong to the national geodetic network of the Institute of Geodynamics -National Observatory of Athens (NOANET; [31]). The exact location and instrument information of the permanent stations are shown in Table 2. After collecting GPS raw data from the online repository, a preprocessing procedure was applied to assure the best quality of post-processing results. TEQC software [32] was used to identify data incompatibilities, excess multipath, and data gaps.

Fault Plane Solutions
In this section, the Moment Tensor Solution for the strong events and all the moderate earthquakes from each aftershock sequence were calculated and presented. For this purpose, seismological broadband data from the Hellenic Unified Seismological Network (HUSN) were collected, analyzed, and used to determine the fault plane solution, the Moment Magnitude (Mw), and the Depth (d) of the strongest earthquakes of the K1,2, L1 and Z1 sequence.
Α methodology based on a moment tensor inversion was used, as analytically described in [26,33,34] using Ammon's software [35]. This method calculates synthetic seis-mograms directly compared with the observed ones for a given velocity structure. As implemented by Randall [36], Kennett's reflectivity method was applied to determine the Green Functions. Synthetics for the three fundamental faults are combined with an appropriate 1-D velocity model, which, in our case, is the one proposed by Haslinger [37]. The Haslinger velocity model was used. Initially, Green's functions for different depths were calculated. Regional data of five 3-component broadband stations belonging to the HUSN and situated at different azimuth coverage and epicentral distances less than 3° were selected and analyzed.
Initially, Green's functions for different depths were calculated by the analyst. Initial inversions were performed at a depth interval of 5 km followed by a finer one every 1-2 km around the depth that exhibited the lowest misfit.
Regional data of five broadband stations, at different azimuth coverage and epicentral distances less than 3°, equipped with three components seismometers, were selected and analyzed. The first step of the procedure is preparing the data, including the deconvolution of instrument response, the integration of the velocity to displacement, and the horizontal components' rotation to radial and transverse. Then the long period part of the signal was introduced to perform the inversion. After several attempts, our analysis was implemented using a seismic wave velocity model considered the most appropriate area under study. A bandpass filter is applied to both the observed waveforms and synthetics. We used a frequency band between 0.05 and 0.02 Hz, although the moderate magnitudes were in the range of 4.0 ≤ Mw ≤ 4.5. In all our inversions, we use a fixed waveform length of 80 s (the inversion results indicate that inverting waveforms longer than 80 s resulted in higher misfits). Moment tensor solutions' quality can be evaluated by considering the average misfit and the compensated linear vector dipole (CLVD; Column 9, Tables A1-A3).
There is a quality code that consists of the letters A-D for each solution for the minimum misfit and between the numbers 1-4 (Quality; Column 11, Tables A1-A3) for the percent of CLVD [38,39].

a. Cephalonia Seismic Sequence
The area of Cephalonia is characterized by too high seismicity, as shown by the past, for example, the seismic action of August 1953 with earthquakes of magnitude 6.5, 6.8, and 7.2 leveled Kefalonia, Zakynthos, and Ithaca and caused about 480 human casualties. The geodynamics and seismotectonic of the area are particularly complex [3,18]. After the destructive sequence of 1953, the strongest earthquakes that have taken place on the broader area are the one west of Cephalonia on 17 January 1983 with ML = 7.0, [10,13] and the one on 18 January 1997 with ML = 6.6 in the Strofades islands. The recent earthquake filled a seismic gap in the area.
On 26 January 2014 (13:55, UTC), two strong earthquakes of magnitude Mw = 6.1 and Mw = 5.2 (18:45, UTC) occurred in the island of Cephalonia, central Ionia Sea. These events induced extensive structural damages, mainly in the western and central parts. Eight days later, on 3 February 2014 (03:08, UTC), a second strong event with a magnitude similar to the first (Mw = 6.0) happened at the north section of Lixouri town. The geographical coordinates for the first events were manually located for this study and found φ = 38.252° Ν, λ = 20.443° Ε at a depth of 16 km. These two earthquakes (Mw = 6.1 and Mw = 6.0) occurred in the Cephalonia island as the destructive events of 1953. In the first days of August 1953 (9 August 1953 and 12 August 1953), three earthquakes of magnitude 6.4, 6.8, and 7.2 [2] took place in Cephalonia. For these three strong events, the source parameters were calculated and compared to the observed solutions from other institutes, and for the majority of them, a good agreement was found (Table 3). A large number of aftershocks followed these events. We note that 2462 events were recorded and analyzed for the first month, while from the beginning of the sequence until the end of 2019 more than 17,000 events took place and were recorded and viewed as a point cloud [40][41][42]. In Figure 2, the 3-D view clearly defines the earthquake epicenters depth zones that may overlay in a 2-D representation. From this catalog and for the largest events (Mw > 3.8), the activation fault, the Source Parameters, the Seismic Moment (M0) and the Moment Magnitude (Mw) were calculated using the moment tensor inversion [26]. Immediately after the earthquake on 26 January 2014, in Paliki of Cephalonia, the Geodynamic Institute sent a team of scientists and technicians. They installed a portable network of four stations (KEF1, KEF2, KEF3, and KEF4) since the number of existing permanent stations was insufficient for detailed recording of aftershock-seismic activity [43,44]. The significant result of the portable network installation was the possibility of recording thousands of earthquakes (n > 8000) over a year, with magnitudes M > 1.0, as seen in Figure 3. The seismological portable network's geographical distribution and the data used from the permanent seismological network stations for the inversion appear in Figure 1. Using the methodology described in the previous Section 3.1 and in the study [26] the Source Parameters (φ, δ, λ), the Moment Magnitude (Μw), the Seismic Moment (M0) and the depth were calculated and are presented in Figures 4-6 regarding the three strongest earthquakes in Cephalonia Island.   The Scheme 26 January 2014 (18:45 UTC), a few hours later, a second of magnitude Mw = 5.2 occurred in the same region as the first event. Using waveforms from Unified Seismological Network (HUSN), the epicenter was manually located at 38.1423° N, 20.2812° E. Broadband recordings from the HUSN network were collected, and those at epicentral distances less than 3° degrees were selected. For the inversion method, five stations were used with good azimuthal coverage. Reverse type faulting was revealed after applying the previous methodology. The obtained focal mechanism is φ = 19°, δ = 62° and λ = 170°. The seismic moment is equal to Mo = 5.99 × 10 23 dyn·cm, for a focal depth equal to 20 km. The inversion resulted in a double couple (DC) equal to 91%, while the compensated linear vector dipole was equal to 9% ( Figure 5).
Next, we present the results of inversion for the third-largest earthquake of this seismic sequence. This earthquake showed specificity as to the application of the method mainly because of its geographical position. The epicenter was calculated in the NW part of Cephalonia Island, according to the National Observatory of Athens (φ = 38.2462° N, λ = 20.3958° E). Due to insufficient azimuthal coverage of the Greek stations, trials were made for some fault plane solutions being recalculated by adding records either of Italian stations, either in mixed epicentral distances, more than 300 km, thus extending our azimuthal coverage to the west and south of the epicenter. To compute the focal mechanism of six stations for three components, each one was used in epicentral distances between 130 and 380 km to determine this event's source parameters. The source parameters were calculated using the method of moment tensor inversion outlined previously. The best fit solution is: strike = 20°, strike = 67°, rake = 174°, and the focal depth was calculated at 20 km. The seismic moment is determined as Mo = 2.46 × 10 24 dyn·cm, and the calculated double couple (DC) was found equal to 88%, while the compensated linear vector dipole (CLVD) to 12%. The results of the applied procedure are presented in Figure 5.

b. Lefkas Seismic Sequence
A strong earthquake with a magnitude Mw = 6.4 occurred in Lefkas Island, Greece. The geographical coordinates, as they calculated from the manual analysis of the Institute of Geodynamics-NOA [27] are φ = 38.6655° N, λ = 20.6002° E at a depth of 10 km. A few hours later, a second strong event with magnitude ML = 5.0 occurred in the same region. These earthquakes caused much structural damage to Agios Petros, Athani, Dragano, and Komilio [46,47] and environmental effects, including liquefaction, extensive rockfalls, and landslides, [48]. No surface ruptures were found in the field. The most recent strong earthquake occurred on 14 August 2003 with a magnitude of Mw = 6.2, offshore the western coast of Lefkas Island, causing severe damages around the whole island [49,50].
A rich seismic sequence followed these events for the following days. More specifically, from 17 November 2015 until the end of the month, 837 events were recorded, while for the first 24 h, the recorded events were 206. The 3D distribution of the epicenters compared to depths and the magnitudes and the distribution of the aftershocks relative to the years is shown in Figure 7a,b.
All the focal mechanisms from events with magnitude Mw > 4.0 were calculated using the proposed methodology (Appendix A, Table A2). The source parameter and the focal mechanism for the main event are shown in Figure 8. For this purpose, the data of six stations of three components, each one in epicentral distances less than 350 km, were used. The source parameters were calculated using the method of moment tensor inversion outlined previously. For the main event, the inversion indicates the activation of a strike-slip type faulting. The best fit solution is strike = 290°, strike = 88°, rake = −12°, and the focal depth is calculated at 10 km. The seismic moment was determined as M0 = 4.402 × 10 28 dyn·cm, and the calculated double couple (DC) was found equal to 85%, while the compensated linear vector dipole (CLVD) to 15%. The following table (Table 4) presents the moment tensor solutions from various institutions for the 17 November 2015 Lefkas earthquake.
A large number of aftershocks that followed this event appears in Figure 9a, b. The most recent seismic sequence occurred during April-May 2006. It consists of four moderate earthquakes (5.3 ≤ Mw ≤ 5.7) that were followed by significant seismic activity. Those focal mechanisms were calculated in the study [55].
All the focal mechanisms from events with magnitude Mw > 4.5 were calculated (Appendix A, Table A3). To determine the source parameters, the data of five stations of three components, each one in epicentral distances less than 350 km, were used. The source parameters were calculated using the method of moment tensor inversion outlined previously. Thrust type faulting was revealed after applying moment tensor inversion. The best fit solution was strike = 123°, strike = 42°, rake = 30°, and the focal depth was calculated at 17 km. The seismic moment was determined at M0 = 1.45 × 10 26 dyn·cm, and the calculated double couple (DC) was found equal to 80%, while the compensated linear vector dipole (CLVD) to 20%. The result of the applied modeling is presented in Figure 10. The following table (Table 5)

Processing of GPS Data
GIPSY/OASIS II software (ver. 6.4) developed by Jet Propulsion Laboratory (JPL, Pasadena, CA, USA) [56] was used to process our data from GPS. The post-processing software uses the precise Point Positioning strategy (PPP) [57]. This method's main advantage is that only a single specific station is needed to produce results, rather than analyzing tens of stations in a DGPS technique, saving valuable computational resources and time.
The GPS data format was RINEX 2.11 from all the three stations (VLSM, PONT, KLOK), a non-fiducial high precision clock, and orbit files that were used (flinnR_nf from JPL). According to international guidelines and experience [58] (a value of 5 × 10 8 cm/sec 2 ) random walk noise is recommended; zenith and tropospheric estimation were used. A troposphere mapping function GPT2 (Global Pressure and Temperature Mapping Function) [59] was used, which was shown to be better than the NIELL mapping function [60] and provides similar results as the VMF1 model (Vienna Mapping Function). Because we analyzed long-term recording GNSS stations [61] (we kept the advantage not to maintain a tropospheric database for this purpose), the use of GPT2 was found to be more appropriate. A lower elevation angle cutoff of 10° was used to eliminate the near-field multipath effect. Also, a receiver antenna calibration file was calculated and generated for each station separately from the IGS atx file, [62]. WahrK1 [63] tide model was used, and (Ocnld-Cpn) ocean load tide model was added, which uses 11 tidal frequencies to infer other frequencies; After the post-processing procedure, a reference frame transformation to ITRF2014 took place for each station and the time series further analyzed with TSANA-LYZER [64] to remove any outliers and extract the velocity of each station. The final time series is presented in Figure 11.  Figure 12 was changed (magnified) to better observe the displacement.

Discussions
Both seismological and geodetic data were used to analyze the strong events during the period 2014-2019 in the Ionian Islands. Source parameters were calculated for the seismic events using regional seismological data. The knowledge of the source parameters for strong as well as for moderate earthquakes is an essential tool for seismically active regions. In general, it allows analytical studies to reveal the tectonics and the seismogenic characteristics of a specific region. The modeling of the strongest earthquakes, using regional data of these sequences, was studied and revealed the region's tectonics, which is characterized by strike-slip and thrust faulting. Strike-slip faulting appears in Cephalonia and Lefkas strong earthquakes, while thrust type faulting appears south of Zakynthos island. Common events were compared with other studies [65,66], and they were found in very good agreement. The main reason to compare this study's solutions with those from other various institutes is to check the proposed methodology's stability as it is applied to earthquakes with different seismotectonic regimes and different magnitude ranges. For all the events used to calculate the source parameters, the epicenter was recalculated and appears in the first line of Tables 3-5 with bold using the velocity model [37]. To determine fault-plane solutions, a methodology and appropriate software implementing the moment tensor inversion was applied. The method calculates synthetic seismograms directly compared with the observed ones for a given velocity structure. The reflectivity method of Kennett, as implemented by Randall, was applied to determine the Green functions. Synthetics for the three fundamental faults are combined with an appropriate 1-D velocity model, which, in our case, is the one proposed by Hasslinger [37]. The focal mechanisms solutions determined in this study appear in Figure 12.
Moreover, the Geodetic Time Series of all the earthquake sequences was analyzed by PPP methodology, which finds the precise position only using each station and precise orbits and clocks products. The position is independent of long-baseline errors introduced by Differential GPS (DGPS) [67], which may produce over or underestimated displacements. The ITRF2014 reference frame [68] minimizes the error introduced from older reference frames like IGb08 and ITRF2008 seen in other studies [69].
We observe that the Lefkas event caused displacement in a broader area (in stations PONT, SPAN, and VLSM) but the Cephalonia event only in the limited area of Cephalonia (in station VLSM). These data contribute to a clearer understanding that the Lefkas event occurred as a result of the CTF regional system [70,71] in contradiction to the Cephalonia event that occurred as a result of the CTF local faulting system. This evidence is getting stronger by Moment Tensors studies, which are the same at both events. Future studies should consider that data from more seismological and geodetical stations in the broader area of Ionian will help better understand the local CTF Cephalonia faulting structure; also, a transformation of a new local reference system will give us more precise results in GPS time series.

Conclusions
In this study, the focal mechanisms of small earthquakes, west of Lefkas to North Cephalonia, are mainly characterized by horizontal slip faults. Specifically, as demonstrated by the method, the right-hand horizontal shift fault west of Cephalonia receives an address ~40° north, while for the department west of Lefkas, it turned out that the fault has a direction of 20° north. Also, the study of focal mechanisms indicates that west of Lefkas, the activation of a strike-slip fault occurred. East of Zakynthos up to the western part of the Peloponnese is predominated by horizontal sliding mechanisms, while in Zak-ynthos's bay, they respond mainly with reverse focal mechanisms. Most of the earthquakes in the area are associated with one right-hand horizontal slip. There are surface earthquakes with activation in the eastern part of the Ionian, mainly horizontal slip faults. Finally, in the southern part, inverted types of generating mechanisms are activated. The depth of the earthquakes studied in this area is between 12 and 30 km, which shows that they are increased compared to the usual depths are identified for surface earthquakes throughout Greece. Here the seismicity seems to be deeper than in other parts of Greece.
In the western part of Cephalonia Island, Ionian Sea, at the SSW-wards continuation of the Lefkas segment of the Cephalonia Transform Fault Zone (CTFZ), more than 8000 earthquakes occurred in the first year. The obtained results provided insights on the rupture mechanism and temporal distribution of the seismic sequence. Source parameters that are focal depth, fault plane solution, and seismic moment were determined by applying the moment tensor inversion methodology using regional data in distances less than 3°. The determined fault plane solutions represent a strike-slip fault's activation, and the depth distribution of the entire sequence ranges between 10 and 20 km. For the same region and events, the GNSS data were used and analyzed. The results indicate that an N-S displacement of ~7 mm and an E-W displacement of ~8 mm took place at station VLSM, and also at earthquake 3 February 2014 (03:08:44, UTC) ML = 5.7 an N-S displacement of ~11.2 mm and E-W displacement ~8 mm took place at the same station and none of the other two stations were significantly affected. The seismic sequence of Cephalonia in 2014 is evolving into a seismic area with a general direction from NW to NN. Τhe seismic sequence of Cephalonia in 2014 is evolving into a seismic area with a general direction from NW to NN and which is the continuation to the NN of the seismogenic area of Lefkas in 2003.
The second seismic sequence analyzed is the Lefkas Island earthquake (17 November 2015, 07:10:07, UTC) to the southwestern part of Lefkas island. In total, 2804 events that followed this event were recorded and analyzed to have a distribution of the epicenters. The moment tensor solution indicated the activation of a strike-slip fault with a rightlateral direction. The GNSS data analysis showed an N-S displacement of ~196 mm, an E-W displacement of ~374 mm, and a vertical displacement of ~58 mm at station PONT. From the analysis of the PONT station's geodetic data and the seismological waveforms' processing, it was observed that in the earthquake of 17 November 2015, the ruptured zone was broken, and not activated in the earthquake of 14 August 2003. Also, an N-S displacement of ~72 mm and E-W displacement of ~57.46 mm occurred at station SPAN; accordingly, an S-N displacement of ~6 mm and E-W displacement of ~18 mm took place at station VLSM. These results are confirmed by the focal mechanism that has already been calculated in the present study. We conclude that in Lefkada's earthquake, the seismic fault moved parallel to Lefkada's west coast in Cephalonia's direction. This is confirmed by the seismicity distribution that was presented in Figures 3 and 7b and the GNSS data processing. Finally, the earthquake sequence at the southern coast of Zakynthos Island that lasted from October 2018 April until the end of 2019 saw that this shallow event was followed by a rich aftershock sequence with a direction SW of the Zakynthos Island. In total, more than 12,000 events were recorded and analyzed in order to calculate the epicenter and the depth. The largest aftershocks of the sequence data were collected and processed to calculate the focal mechanism; the source parameters are the fault plane solution, the focal depth, and the moment magnitude. This analysis emerged for most of them the activation of a strike-slip fault for earthquakes with a focal depth of less than 10 km, while for those with a focal depth greater than 10 km, the activation of a thrust type faulting, with a component of strike-slip, indicates that the 25 October 2018 Mw = 6.7 event ruptured the Hellenic megathrust. This event highlights the high degree of seismic coupling in the Hellenic Arc western region. The GNSS data also point to a similar pattern between the coseismic strain released during 2014 until the end of the 2019 event and the long-term (interseismic) strain accumulation along the west Hellenic Arc. Also, the faultplane geometry is well constrained by GNSS; this is consistent with the distribution of the aftershocks. Figures containing maps were drawn using the Generic Mapping Tools (GMT) software (Wessel and Smith, 1998), http://gmt.soest.hawaii.edu/. We acknowledge concession for the use of ESRI products licensed to the Hellenic National Tsunami Warning Center, National Observatory of Athens, through the project "HELPOS-Hellenic Plate Observing System" (MIS 5002697). HELPOS is implemented under the Action "Reinforcement of the Research and Innovation Infrastructure", funded by the Operational Programme "Competitiveness, Entrepreneurship and Innovation" (NSRF 2014-2020) and co-financed by Greece and the European Union (European Regional Development Fund). We thank the NOANET network for GNSS data, http://geodesy.gein.noa.gr:8000/nginfo/.

Conflicts of Interest:
The authors declare no conflict of interest.   Table A3. Source parameters of the main events and the intermediate magnitudes for Zakynthos Seismic Sequence for the period 2018-2019. Nr is the event number. Lat and Lon are the geographical coordinates of each event, as calculated by the National Observatory of Athens; M0 is the seismic moment in dyn*cm, Mw is the moment magnitude; Strike, Dip, Rake of the two nodal planes are the seismic parameters as calculated from the inversion; CLVD is the percentage of Compensated Linear Vector Dipole, which describes seismic sources with no volume changes; Nr of stations is the number of stations used in inversion and finally the quality of the solution depending from the misfit and the percentage CLVD.