Upper crust seismic anisotropy study and temporal variations of shear-wave splitting parameters in the western Gulf of Corinth (Greece) during 2013

Accepted Manuscript Upper crust seismic anisotropy study and temporal variations of shear-wave splitting parameters in the Western Gulf of Corinth (Greece) during 2013. George Kaviris, Ioannis Spingos, Vasileios Kapetanidis, Panayotis Papadimitriou, Nicholas Voulgaris, Kostas Makropoulos PII: DOI: Reference:

S0031-9201(17)30025-0 http://dx.doi.org/10.1016/j.pepi.2017.06.006 PEPI 6052

To appear in:

Physics of the Earth and Planetary Interiors

Received Date: Revised Date: Accepted Date:

26 January 2017 12 June 2017 14 June 2017

Please cite this article as: Kaviris, G., Spingos, I., Kapetanidis, V., Papadimitriou, P., Voulgaris, N., Makropoulos, K., Upper crust seismic anisotropy study and temporal variations of shear-wave splitting parameters in the Western Gulf of Corinth (Greece) during 2013., Physics of the Earth and Planetary Interiors (2017), doi: http://dx.doi.org/ 10.1016/j.pepi.2017.06.006

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Upper crust seismic anisotropy study and temporal variations of shear-wave splitting parameters in the Western Gulf of Corinth (Greece) during 2013. George Kaviris*a, Ioannis Spingosa, Vasileios Kapetanidisa, Panayotis Papadimitrioua, Nicholas Voulgarisa, Kostas Makropoulosa a

Department of Geophysics–Geothermics, National and Kapodistrian

University of Athens, Panepistimiopolis, 157 84 Zografou, Greece * Corresponding author. Tel.: +30 210 7274841; fax: +30 210 7274787. E-mail addresses: [email protected] (G. Kaviris), [email protected] (I.

Spingos),

[email protected]

(V.

Kapetanidis),

[email protected] (P. Papadimitriou), [email protected] (N. Voulgaris), [email protected] (K. Makropoulos). Abstract During 2013, the Western Gulf of Corinth (WGoC, Central Greece) experienced a period of increased seismicity, with a total of over 4700 earthquakes. This fact in combination with the existence of dense seismological networks provided an excellent opportunity for the study of crustal seismic anisotropy. Of special note is the seismic crisis period of May – October, during which the main feature was the occurrence of the Helike seismic swarm. Polarigrams and hodograms were employed to analyze local waveforms. This method resulted in 659 measurements of shear-wave splitting parameters, namely the direction of the fast shear-wave (Sfast), the time-delay (Td) between the two split shear-waves and the source polarization direction. A pattern of a general WNW – ESE anisotropy direction, parallel to the GoC’s fault systems’ strike, is established, with the exception of two stations located in adjacent areas at the north. This is in agreement with the existence of fluid-filled microcracks, oriented according to the regional stress field. The obtained splitting parameters are compared to the results of other anisotropy studies performed in the WGoC. A detailed analysis of the temporal evolution of the normalized time-delay (Tn) was

1

performed to associate temporal stress changes to seismicity fluctuations. Increase in normalized time-delays and drop before the occurrence of the first significant event belonging to the “July Cluster”, which occurred between the 13th and the 16th of the same month, was observed for most of the analyzed stations. Keywords: Gulf of Corinth; Shear-wave splitting; Seismic anisotropy; EDA model; APE model; Temporal variation of normalized time-delays 1.

Introduction

The Gulf of Corinth (GoC) is one of the most tectonically active areas in Greece (Makropoulos and Burton, 1984; Papadimitriou et al., 1999, 2010; Tsapanos et al., 2010; Kapetanidis et al., 2015) and constitutes an area of major interest in studying intraplate seismicity. It is monitored by two dense seismological networks that provide high coverage of events, both offshore and around the Gulf. These are the Hellenic Unified Seismological Network (HUSN), operated by Greek research institutes (Papanastassiou, 2011)and the local Corinth Rift Laboratory Network (CRLN) that covers the western part of the GoC, around Aigion (Lyon-Caen et al., 2004). The GoC is a neotectonic structure which essentially separates Central Greece from the Peloponnese and is characterized as a complex asymmetric half-graben (Brooks and Ferentinos, 1984; Armijo et al., 1996), given that the southern shore is being uplifted by normal fault systems striking approximately E – W (Fig. 1), while dipping northward (Rigo et al., 1999; Hatzfeld et al., 2000). The length of each fault segment has been estimated up to 20 km, implying that earthquake magnitudes in the study area are moderate (Ambraseys and Jackson, 1990; Doutsos and Poulimenos, 1992). Intense seismicity in the southern coast of the Western GoC (WGoC) has occurred in several main fault zones. The most prominent faults (Fig. 1) are the Psathopyrgos (Houghton et al., 2003), Aigion (Koukouvelas and Doutsos, 1996; Pantosti et al., 2004) and Helike, which led to the destruction of the ancient city of the same name, after the generation of a

2

tsunami (Papadopoulos, 2003; Kortekaas et al., 2011; Soter and Katsonopoulou, 2011). The northern coast of the WGoC features three major systems (Eratini, Psaromita and Trizonia faults) and local subsidence (Stefatos et al., 2002). The results of Briole et al. (2000) indicate that the extension rate deduced from GPS data is 14 mm/yr oriented N9⁰E in the WGoC, whereas the western part of the gulf exhibits about 50% higher strain than the eastern one (Chousianitis et al., 2013). [Figure 1] Seismicity in the area exhibits a rich past with several strong historical (Galanopoulos, 1936; Papazachos and Papazachou, 2003; Kouskouna and Makropoulos, 2004; Stucchi et al., 2013) and instrumental (Makropoulos et al., 2012) events. During the past two decades, two major events occurred, namely the 1992 Galaxidi (Hatzfeld et al., 1996) and the 1995 Aigion earthquakes (Bernard et al., 1997a), whereas since 1995, the two strongest recorded events in the area occurred in 2010 near the city of Efpalio, both with a magnitude of M w = 5.1 (Kapetanidis and Papadimitriou, 2011; Sokos et al., 2012; Ganas et al., 2013). The seismicity in the WGoC, as presented in Fig. 1, is mostly expressed through the occurrence of seismic swarms (Lyon-Caen et al., 2004; Kaviris et al., 2010; Chouliaras et al., 2015; Duverger et al., 2015; Kapetanidis et al., 2015; Mesimeri et al., 2016). Shear-wave splitting has been attributed to a variety of causes in the upper crust, such as foliated rocks (Brocher and Christensen, 1990) and preferred orientation of minerals (Kaneshima, 1990; Valcke et al., 2006). In areas dominated by tectonic processes, fluid-saturated microcracks, oriented parallel to the maximum compressive stress component, exist in the rockmass and control the seismic anisotropy (Nur and Simmons, 1969; Crampin, 1978). The above observations were integrated into an anisotropy model, the Extensive Dilatancy Anisotropy (EDA) model, which eventually evolved into the Anisotropic Poro-Elasticity (APE) model that incorporates pore fluid pressure related to the mechanisms that govern seismic anisotropy

3

in the upper crust (Crampin and Zatsepin, 1997; Zatsepin and Crampin, 1997). Microcrack-controlled anisotropy has been observed in both volcanic and tectonic environments, such as Greece (e.g. Papadimitriou et al., 1999, 2010; Kaviris et al., 2008, 2010, 2014, 2015), Iceland (e.g. Crampin et al., 1999), Italy (e.g. Bianco et al., 2006; Bianco and Zaccarelli, 2008; Pastori et al., 2009) and the U.S.A. (e.g. Liu et al., 1997; Liu et al., 2008; Lewis and Gerstoft, 2012). Temporal variations of time-delays have been associated with changes in a region’s stress regime in several cases (e.g. Gao and Crampin, 2004). Increase of time-delay can be attributed to stress accumulation followed by a potential decline, related to microcrack coalescence. The stress-related alterations affect the microcrack geometry leading to shear-wave splitting variations. While the stress in the rockmass is increasing, the critical point where the fracture occurs can be reached. Nevertheless, the coalescence of continuously aspect-ratio-increasing microcracks into the fault plane coincides with the release of a specific amount of tectonic stress, independently of the earthquake’s magnitude, enabling critically-high pore fluid pressures to result in the eventual earthquake (Crampin et al., 2013). Such observations have taken place in different regions globally, such as Italy (Pastori et al., 2009) and Parkfield, USA (Liu et al., 1997). When a pattern of temporal variation is detected in a specific region, the stressforecast of moderate and large earthquakes can be achieved (Crampin et al., 1999; Gao and Crampin, 2006). It should be noted that such processes have not always been observed in anisotropy studies, as in the case of the 1999 Chi – Chi, Taiwan (Liu et al., 2004) as well as the Izmit and Duzce earthquakes in Turkey (Peng and Ben-Zion, 2005). In the current study, an anisotropy analysis was conducted in the WGoC using data recorded during 2013, including the period of the Helike seismic swarm (Chouliaras et al., 2015; Kapetanidis et al., 2015; Mesimeri et al., 2016). The data were recorded by stations from both the HUSN and the

4

CRLN. The manual processing of waveforms utilized the polarigram (Bernard and Zollo, 1989) and hodogram methods, leading to the measurement of the shear-wave splitting parameters (the polarization direction of the shear-wave component with the higher velocity, Sfast, and the time–delay between the arrivals of the Sfast and the slow component, Sslow) and of the source polarization direction. In addition, temporal variations of time-delays were investigated to examine their correlation to the occurrence of significant events.

2.

Data and Methodology

During 2013, over 4700 events were located using the HYPOINVERSE algorithm (Klein, 1989) and were later relocated via HYPODD (Waldhauser and Ellsworth, 2000). Statistical errors were constrained in sufficiently low levels, because of the available data from local stations, since the median absolute location uncertainties are equal to 0.47 km and 0.90 km, for the horizontal plane and the focal depth, respectively, with relative location errors after relocation being even smaller, especially for clustered events, while the mean RMS is equal to 0.26 sec. For the anisotropy study, a total of 537 local events were utilized. Strict selection criteria were applied, i.e. each selected set of waveforms was recorded by a station with an angle of incidence less than 45⁰, a feature which defines the shear-wave window, to avoid interaction with the free surface (Booth and Crampin, 1985). Ambiguous and disputable arrivals of shear-waves were excluded by using recordings with low noise level presenting clear and impulsive phases. Scattered and converted phases were rejected by ensuring greater recorded amplitudes in the horizontal components than in the vertical one. A band-pass Butterworth filter in the range 1 – 20 Hz was applied to all analyzed waveforms.

5

The chosen analysis method is the visual inspection, a process which involves the constant attention of the researcher, in antithesis with automatic or semi-automatic methods (i.e. Bowman and Ando, 1987; Aster et al., 1990; Silver and Chan, 1991). This manual approach was preferred over automatic ones, given that it offers more control over the measurement and ensures the quality assurance at every step. In addition, automatic methods which involve cross-correlation of the two horizontal components are significantly sensitive to the quality of the recording and, thus, a proper measurement demands stricter selection of waveforms. For the present study, it was decided to employ polarigrams (Bernard and Zollo, 1989) and hodograms, which display the particle motion. These methods have successfully been applied to previous anisotropy studies in Greece, including the WGoC (Papadimitriou et al., 1999; Kaviris et al., 2010, 2015). In the following paragraphs, an example of the analysis process is presented, to provide a clearer description of the method. The utilized waveforms were recorded by PSAR station. In Fig. 2a, the original waveforms are presented. The arrival of the fast shear-wave is marked by a red vertical line, while the angle of its polarization is noted on the polar diagram and is found equal to N105⁰E. The accuracy of the measurement is later confirmed by the hodogram at the bottom of Fig. 2a. [Figure 2a] The waveforms of the horizontal components are then rotated to the determined fast and slow polarization directions (Fig. 2b) to measure the time-delay. The vertical linearity of the Sfast polarization in the current axial system is of particular significance, as it provides a qualitative measure of the determined angle. The rotated waveform to the fast polarization direction is then temporally shifted with respect to the one in the slow direction, so that the arrival of the shear-wave is synchronous to both waveforms. This temporal offset between fast and slow shear-waves is the

6

time-delay, which in the presented example is equal to 9 sampling intervals or 90 ms for a sampling rate of 100 sps. [Figure 2b] After the estimation of the time-delay, the waveforms are rotated back to their original N – E axial system. The displayed waveforms (Fig. 2c) are what would be expected of the station to record if the propagation medium was isotropic. The source polarization direction, in this case N145⁰E, is measured by the corrected polar diagram and hodogram. Source polarization directions can be used for the determination of focal mechanisms, as in the case of the 2013 earthquake swarm in Helike (Kapetanidis et al., 2015). [Figure 2c] 3.

Shear-wave Splitting Results per Station

Anisotropy results were obtained for 9 stations and, in total, 659 measurements of splitting parameters were acquired from a dataset of 537 earthquakes. Average values for the direction of polarization and the timedelay are registered in Table 1, along with their standard error of the mean and the respective 90% confidence interval. In Fig. 3, rose diagrams for each station are presented, providing a concise view of the polarization’s general direction. Equal-area projections in Fig. 4, where the periphery of the circle corresponds to the maximum angle of incidence, i.e. 45⁰, bring out the independence of the polarization direction to the azimuth for the majority of the stations. Most stations exhibit a general WNW – ESE Sfast direction, with only two of them being differentiated in an approximate NE – SW direction. Circular statistics (Berens, 2009) have been employed for the calculation of the mean polarization directions and their respective errors, since they provide more reliable results for angular data by ignoring the wrapping that occurs at 360 for the orientations or 180 for simple orientation-independent directions. Time-delays are generally constrained in the range of 10 ms to 190 ms and, as a consequence, can set specific limits

7

to the magnitude of anisotropy. The Td standard errors of the mean were determined to be smaller than or equal to 5 ms, which is the reading error considering that the sampling rate is 100 sps, for all stations except for PYRG which had a typical error of mean value larger than the reading error. In addition, the 90% confidence interval, also presented in Table 1, is smaller than or equal to the reading error for 5 (KALE, PSAR, SERG, TRIZ and VVK) out of the 9 stations. [Table 1] [Figure 3] [Figure 4] EFP station is located in the NW part of the study area, near the city of Efpalio, yielding a total of 68 measurements of shear-wave splitting parameters. The location of EFP provides high coverage of events, both inland and offshore. The mean polarization direction is N94⁰E (Fig. 3a) with a range of N70⁰E to N120⁰E and a mean error of 5⁰. Time-delays present a mean value of 85 ms, with an error of 5 ms, equal to the reading error, and extreme values between 30 ms and 180 ms. This station is characterized by a deficit of events to the north (Fig. 4a), since there is no significant tectonic activity in that area. Moreover, periods that exhibit an absolute absence of events, like November and December, can be attributed to the lower activity of the seismogenic areas. These periods may be characterized by either no events or a very small amount of earthquakes within the spatial boundaries defined by the angle of incidence criterion. The station of KALE is installed at an elevation of 760 m, the highest of the presented stations, near the northern coast of the WGoC. A total of 60 anisotropy parameters led to a fairly constraint range of polarization direction values, limited between N70⁰E and N130⁰E, and a mean N100⁰E (Fig. 3b) with a standard error of mean equal to 5⁰. Measured time-delays exhibit typical values for the area, corresponding to a mean 79 ms in the

8

range between 10 ms and 150 ms. As in the case of EFP and other northern stations, analyzed events are concentrated to the south (Fig. 4b), where seismicity mostly occurs. LAKA is located in the SW part of the study area and is the closest utilized station to the Helike seismic swarm. This station features a mean anisotropy direction of N122⁰E (Fig. 3c) with a mean error of 5⁰, while the values are contained within N92⁰E and N150⁰E, as determined from 51 events. Concerning the time-delay, it presents a mean value of 71 ms and varies between 10 ms and 130 ms. The station’s azimuthal distribution is characterized by the absence of events to the NW (Fig. 4c). The major events that occurred in the WGoC during 2013 and their subsequent aftershock sequences have been recorded by LAKA, fact that implies a reliable normalized time-delay analysis (see paragraph 4.2). The easternmost station that was utilized in the present anisotropy study is PSAR, installed near the northern coast of the gulf. Exhibiting a highly uniform set of 135 anisotropy direction measurements, PSAR has a mean polarization direction equal to N98⁰E (Fig. 3d) and the smallest error value (along with TRIZ), i.e. 3⁰. The limits of the Sfast range are N70⁰E and N125⁰E. Azimuthal distribution of results is excellent, providing measurements for the whole range of possible azimuths, even though there is a slight gap to the SE (Fig 4d). Time-delays present a similar distribution with a mean value of 88 ms, while varying between 20 ms and 220 ms (the highest time-delay value for all stations). Analysis at PYRG station, located in the NW part of the study area, resulted in 30 shear-wave splitting parameters, presenting a significant alteration to the status of anisotropy. In contrast with previous stations, the mean anisotropy direction is N51⁰E (Fig. 3e), with a mean error equal to 11⁰, and a range between N25⁰E and N75⁰E. A significant azimuthal gap is observed to the N and NE (Fig. 4e) as a result of the relatively seismically quiet area

9

towards Central Greece. Time-delays exhibit a 67 ms mean value, accompanied by a 7 ms mean error value, greater than the reading error, in a range within 10 ms and 140 ms. As a part of the CRLN, ROD3 is installed near the western boundary of the GoC. This station, with a total set of 52 results, presents a mean value of anisotropy direction equal to N107⁰E (Fig. 3f) with an error of 5⁰ and a Sfast range between N80⁰E and N132⁰E. It is also noted that ROD3 is characterized by events with angles of incidence close to the upper limit of 45⁰, since the recorded seismicity is located in greater epicentral distances (in any case always smaller than 10 km) in comparison to other stations. Measurements at ROD3 have a very poor azimuthal distribution, since most of the events are located to the E and SE, with only a handful appearing in different directions. Time-delays exhibit a mean value of 83 ms, with a maximum value of 190 ms. SERG station exhibits a similar behavior to PYRG, rendering them the two exceptions in the general pattern of WNW – ESE anisotropy direction measurements throughout the GoC. Nevertheless, the splitting parameters for this station were obtained by a greater number of analyzed events (72) and, as such, provide a more reliable result about this differentiation. A N51⁰E mean Sfast value is determined (Fig. 3g), while the error corresponds to 6⁰ and the values range between N30⁰E and N80⁰E. Selected events for SERG station are mostly concentrated to the SW, while a complete absence of events to the N is evident, being a common trait among northern stations. The time-delay is represented by a mean value of (59±5) ms, the minimum measured Td is 10 ms, while the maximum is 130 ms. Such a small mean time-delay value is a characteristic shared among stations in the NW part of the study area, west of the Trizonia Island, with the exception of EFP. The TRIZ station is characterized by the highest number of analyzed events, equal to 152, ensuring reliable and well-documented results, and is located

10

in a central position. The area is characterized by a mean direction value of N115⁰E (Fig. 3h) and its corresponding error of 3⁰ reflects the very satisfactory distribution of values, while being constrained in an area of measurements within N85⁰E and N145⁰E. As a station installed on an islet near the northern coast of the GoC, recorded seismicity is mainly located offshore and, thus, an absence of results is observed to the N and NE (Fig 4h). A mean time-delay value equal to (62±5) ms is consistent with measurements in the stations to the NW (PYRG, SERG), while being significantly lower than results in stations to the E and NE (KALE, PSAR). Values are constrained between 10 ms and 160 ms. VVK is the westernmost station utilized in the study, providing a fairly low number of results (39). The mean value of the anisotropy direction, which is N103⁰E (Fig. 3i), with an error of 8⁰, is close to the one determined in EFP. The value range for the Sfast polarization directions is bounded by N75⁰E and N130⁰E. It is noted that these two stations provided coupled recordings of earthquakes since they are installed in adjacent areas. Nevertheless, timedelays do not follow such a pattern with VVK being characterized by a much lower mean value of 56 ms (in accordance with other northern stations) than EFP, limited between 20 ms and 120 ms. 4.

Temporal Variation of Splitting Parameters

Temporal variations of the splitting parameters and especially those of timedelays have been proved to be sensitive to changes of local stress in several sites around the world (Gao and Crampin, 2004). Normalized time-delays, where the measured value is divided by the hypocentral distance in order to exclude the effect of the ray-path’s length, usually increase during a period before the occurrence of a major event, the duration of which is related to the earthquake’s magnitude. Nevertheless, this phenomenon is followed by the unification of microcracks into the fault plane and is expressed by a magnitude-independent stress release (equal to 2- 4 MPa), as observed by a

11

decrease of time-delay values (Crampin et al., 2013). The expected period of time-delay increase for an M = 3.0 event is about 13 days and the decrease may last for roughly 0.1 days, while for an M = 4.2 earthquake these numbers are adjusted to 46 days and 1.3 days, respectively, according to the relations proposed by Crampin et al. (2015). In addition, these changes have been mostly observed in Band-1. The characterization of time-delay values belonging to either Band-1 or Band-2 refers to the ray-path between the hypocenter and the station, in accordance with the orientation of the vertical crack plane. When the ray-path is contained inside a solid angle ranging between 15⁰ and 45⁰ on either side of the crack plane, it is categorized in Band-1, whereas when the ray travels through the medium within the 15⁰ solid angle from the crack plane the respective measurements are defined as Band-2 (Crampin et al., 1999; Volti and Crampin, 2003; Gao and Crampin, 2006). Band-1 ray-paths are especially sensitive to the aspect ratio of microcracks, while the Band-2 ones are mainly affected by their density. In Fig. 5, the temporal variations of the direction of polarization of the Sfast, with their respective standard deviation margins, are shown for all of the utilized stations. As evident, the direction of polarization does not present significant temporal variations, being relatively stable around a mean level throughout different time periods for each station. This is indicative of the specific parameter, which represents the orientation of microcracks in the anisotropic medium, being affected only by long-term processes that take years or even decades to induce changes. [Figure 5] In the context of the present study, no major differences were observed between normalized and non-normalized time-delays, which can be attributed to the narrowly defined hypocentral distance range of our samples, varying between 3.6 km and 17.0 km with an average value of 10.6 km and median equal to 7.2 km.

12

Possible correlation of time-delay variations with specific earthquakes (Fig. 6 and 7) is observed, with the exception of stations PYRG, SERG and VVK, where such a relation could not be established. During July, a common characteristic presented among specific stations, is a steady increase in timedelays and a drop right before the occurrence of the first (or second in some cases) of nine significant events (the “July Cluster”) that occurred between the 13th and 16th

of the same month, according to the catalogue of

Kapetanidis et al. (2015). These earthquakes, with moment magnitudes in the range of 3.1 – 3.7, occurred south of the WGoC in NW Peloponnese, near the city of Helike (Kapetanidis et al., 2015). It is also worth noting that some stations present data gaps for certain time periods, whereas others don’t, due to various reasons (e.g. occurrence of events that do not satisfy our initial selection criteria for these stations, or periods of stations’ inactivity). Stress variations, inferred from the normalized time-delay changes, do not exhibit consistent behavior in all stations. The temporal variations of the normalized time-delays (Tn) in stations EFP, KALE, LAKA, PSAR, ROD3 and TRIZ in ‘‘Band-1’’ and “All-Bands” (both Band-1 and Band-2) are presented in Fig. 6 and Fig. 7, respectively, along with a histogram of the number of events that occurred per day in the broader study area. Errors for Td measurements, presented in Table 1, are the reading error of time-delays, δTd = 5 ms, while error-bars for the normalized time-delays, Tn, take into account the propagation of errors which is calculated using the formula (Del Pezzo et al., 2004):

(Equation 1)

where δTn the error of the normalized time-delay, D the hypocentral distance (path length) and δD its error, which is taken at a fixed value of δD = 0.2 km. The 3-point moving average is also displayed in Fig. 6 and Fig. 7 by a solid

13

line, bounded by the ±1. In addition, linear regression statistics for the temporal variations of Tn are presented for “Band-1” and “All-Bands” in Tables 2 and 3, respectively. [Table 2] [Table 3] [Figure 6a] [Figure 6b] [Figure 7a] [Figure 7b] EFP station presents a fairly steady increase and decrease of time-delays before the Mw = 3.7 event that occurred on 31/05 (Kapetanidis et al., 2015), with the increase starting right after the ML = 3.1 earthquake of 18/05 and persisting up to one day before the earthquake of 31/05. This observation is statistically significant when examined in the “All-Bands” set (Fig. 7a), with a correlation coefficient R2=0.99 and p=0.2% probability of zero slope, derived by performing a Student’s t-test. It is also followed by a significant decrease (Fig. 7a) in the period up to 13/06 (R2=0.96, p=0.8% in the “AllBands” set). The stress accumulation apparently starts again after the Mw = 3.5 event on 16/06. Nevertheless, the exact behavior cannot be extrapolated from the current data, since there is a lack of events after 20/06. After a long period of without available measurements (between July and mid-October) a drop can be observed, related to the three events at the end of October. Nevertheless, such a link requires more data to be established, as only two measurements of 13 and 14 ms/km are observed before the eventual drop to 6 ms/km, one day before the ML = 3.1 event on 24/10. At the end of the year, a continuous decrease of normalized time-delays is observed, starting a couple of days before the ML = 3.6 event of 17/12. While a slight increase is apparently taking place after the event, the occurrence of another earthquake

14

with ML = 3.4 four days after the last one (on 21/12) is accompanied by a decrease the day before. Nevertheless, the data in this case do not provide a strong relation between the last event and the time-delay drop, as widely different values (7 ms/km and 13 ms/km) have been determined on adjacent days. The station KALE has a plethora of results at the beginning of February which cannot be correlated with temporally adjacent events, since they exhibit significant fluctuations in a very short time. A trend of decrease is visible, right before the ML = 3.5 event on the 6th of February, but it consists of only N=4 measurements and lacks statistical significance. A steady and significant (R2=0.79, p=1.2% and N=9, all in Band-1) increase of the timedelay values begins on the 25th of May (Fig. 6a) and is continued until a couple of days before the occurrence of the 27/06 event with Mw = 3.6. The accumulation time period is longer than expected for an event of such magnitude, but it can be attributed to the occurrence of other significant events in that same period, enhancing the accumulation of stresses and creating the impression of a gradual increase associated with a forthcoming M = 3.5 event. The scattering of data from that temporal point to the end of the year does not permit the extraction of further links between significant earthquakes and time-delay values measured at station KALE. The Mw = 3.6 earthquake of 28/05 is accompanied by a statistically significant decrease (R2=0.93, p=0.3% and N=7, all in Band-1) of normalized time-delays in LAKA station one day before the event (Fig. 6a). Nevertheless, the accumulation period cannot be specified due to lack of recorded seismicity in the period before the event. A measurement equal to 5 ms/km on the 8th of May is unlikely to be linked with the specific event, since such long periods of accumulation are not expected for events with magnitudes of ~3.5. LAKA exhibits a similar image concerning the July Cluster, where a drop of time-delays is observed a couple of days before the first event on 13/07 and is continued well beyond the occurrence of all eight

15

significant events, until the Mw = 3.6 earthquake of 24/07, where the stresses begin to accumulate anew. The decrease is statistically significant (R2=0.78, p=0.8%) as determined from a set of N=10 measurements, all of which belong to Band-1 (Fig. 6a). PSAR station does not provide any conclusive results in the set of Band-1 measurements. While cycles of increasing and decreasing time-delays can be distinguished, the only relatively strong case is observed before the 27/06 event, but is later followed by notable irregularities. Meanwhile, a statistically significant increase (R2=0.99, p=1% for N=4 events in Band-1 or R2=0.98, p=0.3% for N=5 events in “All-Bands”) and, especially, a decrease (Fig. 6b and 7b) is recorded near the end of September, but is loosely linked with the ML = 3.7 earthquake of the 26th of September, since the hypothetical stress relaxation begins only after the event has occurred, with high values of normalized time-delays, equal to 15 ms/km, observed on the same day. Nevertheless, PSAR offers stronger earthquake – stress correspondence when utilizing measurements from “All-Bands” (Fig. 7b). There is a statistically significant increase (R2=0.86, p=0.3%, N=9) ending on 08/07, 5 days before the occurrence of the July Cluster. Despite the lack of seismicity that met the spatial selection criteria, ROD3 provides a coherent status of normalized time-delay values for two main periods. The first one refers to the months July – September, where a decrease of time-delays can be speculated on 08/07, a few days before the first event of the July Cluster, but the results are statistically inconclusive. An individual measurement of lower normalized time-delay on the 2nd of July cannot be utilized to specify an accumulation period of ~6 days. Nevertheless, the “All-Bands” image (Fig. 7b) during the same time-interval infers an accumulation period that ends on the 3rd of July and is supported by two additional results on the 25th and 26th of June, rendering the aforementioned low-value time-delay a significant irregularity. As a consequence, the July Cluster in ROD3 is registered mostly in the “All-

16

Bands” range than in the Band-1, a similarity shared with PSAR and TRIZ stations. In addition, another period, lasting through September and independent of any significant earthquakes, is well-documented in both Band-1 (Fig. 6b) and “All-Bands” (Fig. 7b) spaces. Such an oddity is not uncommon among the presented stations and requires further investigation. A second period with numerous results is observed since mid-October until the end of the year. The temporal variations do not appear to correlate with significant events. TRIZ station provides the highest number of measured parameters in the current study and, consequently, can offer a more complete picture of the variations of normalized time-delays. During May, a period of steady increase of time-delays is observed, from the 3rd to the 26th of the same month, before the occurrence of the Mw = 3.6 earthquake on 28/05. This variation is considered statistically significant in both “Band-1” (R2=1.00, p=0.003%, N=4) and in “All-Bands” (R2=0.87, p=0.2%, N=9). Concerning the July Cluster, a similar situation with PSAR and ROD3 is evident. While in Band-1 a temporally constrained increase is observed during early July (Fig. 6b), a different situation is exhibited in the “All-Bands” set (Fig. 7b), with the results in the period 5-10 July being statistically inconclusive. The stress accumulation is beginning as early as 18/06, affected by the Mw = 3.5 event on 16/06, and continues until the 5th of July, with a slight variation at the end of June caused by the earthquake that occurred on 27/06. The stress relaxation then takes place, with the eventual occurrence of the July Cluster, being statistically significant (Fig. 7b) in the period between 3 and 18 August (R2=0.98, p=2.3% in Band-1 and R2=0.77, p=4.3% in “All-Bands”). The decrease terminates after the occurrence of the 24/07 event. A later stress cycle related to the earthquakes that occurred on the 19th and 25th of September is observed in both Band-1 and “All-Bands”. 5.

Discussion - Conclusions

17

Extensive analysis of seismicity recorded by the HUSN and the CLRN during 2013 in the WGoC, including the occurrence of a seismic swarm between May and October (Chouliaras et al., 2015; Kapetanidis et al., 2015), led to the confirmation of an anisotropic seismic layer beneath the study area. This result is supported by the estimated shear-wave splitting parameters, which correspond to a total number of 659 measurements, including the polarization direction of the Sfast and the time-delay between the two split shear-wave components. A correlation is observed between the Sfast polarization direction and estimations of the stress orientation in the WGoC, as well as existing GPS measurements (Briole et al., 2000; Avallone et al., 2004; Chousianitis et al., 2013). Seven out of the nine stations, whose recordings were analyzed, provide general WNW - ESE anisotropy directions (Fig. 8), approximately perpendicular to the NNE - SSW extensional regime of the tectonic structure (Jackson et al., 1982; Armijo et al., 1996; Rigo et al., 1996; Bell et al., 2009; Taylor et al., 2011). This trend is independent of the azimuth of each event, since stations with adequate azimuthal distribution did not exhibit highly variable measurements (Fig. 4). The above observations are consistent with the existence of fluid-filled microcracks, oriented according to the regional stress field. [Figure 8] Previous studies have been conducted in the area and produced similar results, confirming the existence of an anisotropic upper crust. In the region of Efpalio, where EFP station is located, the mean fast shear-wave polarization direction has been previously measured equal to N105⁰E (Bernard et al., 1997b) and N75⁰E (Giannopoulos et al., 2015). These values are significantly different, while the first result is closer to the one obtained by the current study (N94⁰E). It is noted that Giannopoulos et al. (2015) used a cross-correlation method and selected events that occurred in the

18

framework of a specific seismic sequence during 2009 – 2010. In the area of Kallithea (KALE station), Bouin et al. (1996) and Giannopoulos et al. (2015) acquired polarization directions equal to N119⁰E and N125⁰E, respectively. Both results differ by about 20⁰ to the results of the current study (N100⁰E). Kaviris et al. (2010) measured a N123⁰E mean direction in station LAKA, using events of a local seismic swarm in 2008, which is regarded as the main one, as supported by previous results (Kaviris et al., 2008), while a secondary direction was also identified at N100⁰E. In addition, Giannopoulos et al. (2015) measured a mean polarization direction of N108⁰E in the same station, while we obtained a direction of N122⁰E. For station PSAR, the largest difference with previous studies was observed. Bouin et al. (1996) performed an anisotropy analysis on earthquakes that occurred in 1991 and in 1992 belonging to the aftershock sequence of the Ms = 5.9 event in Galaxidi on 18/11/1992 and measured polarization directions equal to N70E⁰ and N60⁰E, for each year respectively. These values significantly deviate from the proposed direction of the microcracks under the WGoC. This research team used an automatic method during their analysis, which might be the cause of the difference observed with the present result, i.e. N98⁰E. In addition, Kaviris et al. (2008), utilizing earthquakes recorded in 2000 by the same station, measured a mean anisotropy direction of to N105⁰E, using the same visual inspection method as the current study. The obtained mean polarization direction in the area of Rodini (ROD3 station), N107⁰E, varies by 21⁰ to the one measured by Giannopoulos et al (2015), which was found N86⁰E. In SERG, the significant deviation of the mean polarization direction (N51⁰E, as measured in the current study) from that of the major faults strikes in the rest of the WGoC was observed by Giannopoulos et al. (2015) as well, presenting an anisotropy direction of N68⁰E. Results from TRIZ station are pretty similar between the two studies, since Kaviris et al. (2008) observed a main direction equal to N130⁰E and a secondary of N110⁰E, while

19

Giannopoulos et al. (2015) measured a direction of N122⁰E. While these values do not deviate in a major degree, results of the present study are in agreement with both, given that a mean anisotropy direction of N115⁰E was obtained. Anisotropy results for stations PYRG and VVK are presented for the first time in the framework of the current study. As a consequence, the microcracks’ geometry is a seemingly stable feature of the anisotropic layer. The above observations, also evident by the negligible temporal variations of the Sfast polarization direction (Fig. 5), suggest that fractures are controlled by long-term tectonic processes, i.e. the rifting of the GoC. Fracture systems, as established by the general tectonic regime, lead to the migration of fluids in a specific direction, but not orientation. As a result, it is difficult to distinguish whether anisotropy is attributed to the EDA or the APE model. The mean anisotropy direction values in PYRG and SERG stations (Fig. 8) require further investigation. They differ from the results of the other stations, presenting a general NE – SW anisotropy direction. It is worth noting that these two stations are located in close proximity to each other. This exception can be explained by the existence of tectonic features which are different than the rest of the GoC. Structures striking NE – SW are generally observed close to and beyond the western edge of the WGoC, while an offshore fault with similar orientation and dextral component has been mapped by Beckers et al. (2015). Furthermore, four offshore and two minor onshore faults with NE-SW strike exist in the vicinity of PYRG and SERG

stations

(Loftus

and Tsoflias,

1971;

Paraschoudis,

1977;

Papanikolaou et al., 1997; Tranos, 2016). In addition, certain focal mechanisms in the area have presented nodal planes with strikes similar to the obtained mean polarization direction (Rigo et al., 1996; Karakostas et al., 2012; Sokos et al., 2012). Time-delay is a sensitive parameter and can be altered significantly in a timescale ranging from days to years, likely dependent of each sequence’s

20

main event (Gao and Crampin, 2004; Crampin et al., 2013, 2015). As a consequence, the comparison of measured mean time-delays to ones from previous studies does not provide any valuable information. On the contrary, single time-delay values compared in a temporal context may highlight important tectonic processes and stress variations. Stations can be classified in two groups in terms of mean time-delay values, the first between 67 ms and 90 ms (EFP, KALE, LAKA, PSAR, PYRG, ROD3) and the second about 60 ms (SERG, TRIZ, VVK). Nonetheless, no correlation between the time-delays and the stations’ location and altitude is observed. Temporal variations of normalized time-delays led to the determination of stress accumulation and relaxation processes, dependent of the occurrence and severity of earthquakes in the broader area of each station. While several studies have indicated the importance of utilizing the Band-1 criterion (Crampin et al., 1999; Gao and Crampin, 2004; Crampin et al., 2013), significant conclusions can be drawn by comparing data belonging to the Band-1 range (Fig. 6) and to the “All-Bands” range (Fig. 7). While stations PSAR, ROD3 and TRIZ exhibited a decrease of the normalized time-delay values (which corresponds to stress relaxation) in Band-1 before the July Cluster (a period of 6 days characterized by increased seismicity and the occurrence of 8 significant earthquakes), nevertheless, the “AllBands” set provided additional information, including a potential stress accumulation period. In general, analysis of temporal variations of normalized time-delays brought out some weak correlation of stress accumulation and relaxation cycles with significant local events. As a consequence, the phenomenon described by the APE model was observed for a limited number of earthquakes. Such discrepancies can be attributed to the magnitude of the defined “significant” events in comparison to the generally observed seismicity in the rift. Each of these earthquakes (Mw > 3.0), with the maximum recorded magnitude during 2013 being 4.2, cannot be characterized as much more intense than the common occurring

21

earthquakes with M  2.0. In other occasions through the literature, welldocumented stress-cycles refer to significantly stronger earthquakes that stand out among much weaker earthquakes of a region’s background by more than 3 orders of magnitude. Such cases have been observed by Piccinini et al (2006) where earthquakes with magnitudes less than 2.5 were utilized to determine stress variations before the occurrence of two events of magnitudes Mw = 5.2 and Mw = 5.4 in Central Italy. Nevertheless, additional research is required to better understand occasions of apparent stress cycles that occur during periods with an absence of any significant event. 6.

Acknowledgements

We would like to thank the scientists and personnel who participated in the installation or maintenance of the permanent and temporary stations belonging to the HUSN network, as well as the French research team of the CRL network. The present study was partially financed by the Special Account for Research Grants of the National and Kapodistrian University of Athens (UoA – S.A.R.G.). 7.

References Ambraseys, N. N., & Jackson, J. A. (1990). Seismicity and associated strain of central Greece between 1890 and 1988. Geophys. J. Int., 101, 663–708. Armijo, R., Meyer, B., King, G., Rigo, A., & Papanastassiou, D. (1996). Quaternary evolution of the Corinth Rift and its implications for the Late Cenozoic evolution of the Aegean. Geophys. J. Int., (126), 11–53. Aster, R. C., Shearer, P. M., & Berger, J. (1990). Quantitative measurements of shear wave polarizations at the Anza Seismic Network, southern California: Implications for shear wave splitting and earthquake

22

prediction.

J.

Geophys.

Res.,

95(B8),

12449.

https://doi.org/10.1029/JB095iB08p12449 Avallone, A., Briole, P., Agatza-Balodimou, A. M., Billiris, H., Charade, O., Mitsakaki, C., Nercessian, A., Papazissi, K., Paradissis, D., & Veis, G. (2004). Analysis of eleven years of deformation measured by GPS in the Corinth Rift Laboratory area. C. R. Geosci., 336(4–5), 301–311. https://doi.org/10.1016/j.crte.2003.12.007 Beckers, A., Beck, C., Sakellariou, D., & De Batist, M. (2015). Active faulting at the western tip of the Gulf of Corinth , Greece , from high-resolution

seismic

data.

Mar.

Geol.,

360,

55–69.

https://doi.org/10.1016/j.margeo.2014.12.003 Bell, R. E., McNeill, L. C., Bull, J. M., & Henstock, T. J. (2008). Evolution of the offshore western Gulf of Corinth. Bull. Geol. Soc. Am., 120(1–2), 156–178. https://doi.org/10.1130/B26212.1 Bell, R. E., McNeill, L. C., Bull, J. M., Henstock, T. J., Collier, R. E. L., & Leeder, M. R. (2009). Fault architecture, basin structure and evolution of the Gulf of Corinth rift, central Greece. Basin Res., 21(6), 824–855. https://doi.org/10.1111/j.1365-2117.2009.00401.x Berens, P. (2009). CircStat: a MATLAB toolbox for circular statistics. J. Stat. Softw., 31(10). Bernard, P., & Zollo, A. (1989). Inversion of near-source S polarization for parameters of double-couple point sources. Bull. Seismol. Soc. Am., 79(6), 1779–1809. Bernard, P., Briole, P., Meyer, B., Gomez, J., Tiberi, C., Berge, C., Cattin, R., Hatzfeld, D., Lachet, C., Lebrun, B., Deschamps, Courboulex, F., Larroque, C., Rigo,

A.,

A., Massonnet, D.,

Papadimitriou, P., Kassaras, J., Diagourtas, D., Makropoulos, K., Veis, G., Papazisi, E., Mitsakaki, C., Karakostas, V., &

23

Papadimitriou, E. (1997a). The Ms=6.2, June 15, 1995 Aigion earthquake (Greece): evidence for low-angle normal faulting in the Corinth

J.

rift.

Seismol.,

1(1),

131–150.

https://doi.org/10.1023/A:1009795618839 Bernard, P., Chouliaras, G., Tzanis, A., Briole, P., Bouin, M., Tellez, J., Stavrakakis, G., & Makropoulos, K. (1997b). Seismic and electrical anisotropy in the Mornos delta, Gulf of Corinth, Greece, and its relationship with GPS strain measurements. Geophys. Res. Lett., 24(17), 2227–2230. Bianco, F., & Zaccarelli, L. (2008). A reappraisal of shear wave splitting parameters from Italian active volcanic areas through a semiautomatic

algorithm.

J.

Seismol.,

13(2),

253–266.

https://doi.org/10.1007/s10950-008-9125-z Bianco, F., Scarfı, L., Del Pezzo, E., & Patanè, D. (2006). Shear wave splitting changes associated with the 2001 volcanic eruption on Mt Etna.

Geophys.

J.

Int.,

167(2),

959–967.

https://doi.org/10.1111/j.1365-246X.2006.03152.x Booth, D., & Crampin, S. (1985). Shear-wave polarizations on a curved wavefront at an isotropic free surface. Geophys. J. Int., 31–45. Bouin, M., Téllez, J., & Bernard, P. (1996). Seismic anisotropy around the Gulf of Corinth, Greece, deduced from three‐ component seismograms of local earthquakes and its relationship with crustal strain.

J.

Geophys.

Res.,

101(B3),

5797.

https://doi.org/10.1029/95JB03464 Bowman, J., & Ando, M. (1987). Shear-wave splitting in the upper-mantle wedge above the Tonga subduction zone. Geophys. J. Int., 25–41. Briole, P., Rigo, A., Lyon-Caen, H., Ruegg, J.C., Papazissi, K., Mitsakaki, C., Balodimou, A., Veis, G., Hatzfeld, D. & Deschamps, A. (2000).

24

Active deformation of the Corinth rift, Greece: Results from repeated Global Positioning System surveys between 1990 and 1995. J. Geophys. Res., 105(B11), 25605–25625. Brocher, T. M., & Christensen N. L. (1990). Seismic anisotropy due to preferred

mineral

orientation

observed

in shallow crustal

rocks in southern Alaska. Geology, 18, 737-740, 1990. Brooks, M., & Ferentinos, G. (1984). Tectonics and sedimentation in the Gulf of Corinth and the Zakynthos and Kefallinia channels, Western

Greece.

Tectonophysics,

101,

25–54.

https://doi.org/10.1016/0040-1951(84)90040-4 Chouliaras, G., Kassaras, I., Kapetanidis, V., Petrou, P., & Drakatos, G. (2015). Seismotectonic analysis of the 2013 seismic sequence at the

western

Corinth

Rift.

J.

Geodyn.,

90,

42–57.

https://doi.org/10.1016/j.jog.2015.07.001 Chousianitis, K., Ganas, A., & Gianniou, M. (2013). Kinematic interpretation of present-day crustal deformation in central Greece from continuous GPS measurements. J. Geodyn., 71, 1–13. Crampin, S. (1978). Seismic-wave propagation through a cracked solid: polarization as a possible dilatancy diagnostic. Geophys. J. Int., 53(3),

467–496.

https://doi.org/10.1111/j.1365-

246X.1978.tb03754.x Crampin, S., & Zatsepin, S. (1997). Modelling the compliance of crustal rock—II. Response to temporal changes before earthquakes. Geophys. J. Int., 129(495–506). Crampin, S., Volti, T., & Stefánsson, R. (1999). A successfully stressforecast earthquake. Geophys. J. Int., (138), 1–5.

25

Crampin, S., Gao, Y., & De Santis, A. (2013). A few earthquake conundrums resolved. J. Asian Earth Sci., 62, 501–509. https://doi.org/10.1016/j.jseaes.2012.10.036 Crampin, S., Gao, Y., & Bukits, J. (2015). A review of retrospective stressforecasts of earthquakes and eruptions. Phys. Earth Planet. Inter., 245, 76–87. https://doi.org/10.1016/j.pepi.2015.05.008 Del Pezzo, E., Bianco, F., Petrosino, S., & Saccorotti, G. (2004). Changes in the Coda Decay Rate and Shear-Wave Splitting Parameters Associated with Seismic Swarms at Mt . Vesuvius , Italy. Bull. Seismol. Soc. Am., 94(2), 439–452. Doutsos, T., & Poulimenos, G. (1992). Geometry and kinematics of active faults and their seismotectonic significance in the western CorinthPatras

rift

(Greece).

J.

Struct.

Geol.,

14(6),

689–699.

https://doi.org/10.1016/0191-8141(92)90126-H Duverger, C., Godano, M., Bernard, P., Lyon-Caen, H., & Lambotte, S. (2015). The 2003-2004 seismic swarm in the western Corinth rift: Evidence for a multiscale pore pressure diffusion process along a permeable fault system. Geophys. Res. Lett., 42(18), 7374–7382. https://doi.org/10.1002/2015GL065298 Galanopoulos, A. G. (1936). Seismicity of Achaia and its adjacent Mesolongi and Nafpaktos areas. PhD thesis, University of Athens, p. 23 (in Greek). Ganas, A., Chousianitis, K., Batsi, E., Kolligri, M., Agalos, A., Chouliaras, G., & Makropoulos, K. (2013). The January 2010 Efpalion earthquakes (Gulf of Corinth, Central Greece): Earthquake interactions and blind normal faulting. J. Seismol., 17(2), 465–484. https://doi.org/10.1007/s10950-012-9331-6

26

Gao, Y., & Crampin, S. (2004). Observations of stress relaxation before earthquakes.

Geophys.

J.

Int.,

157(2),

578–582.

https://doi.org/10.1111/j.1365-246X.2004.02207.x Gao, Y., & Crampin, S. (2006). A stress-forecast earthquake (with hindsight), where migration of source earthquakes causes anomalies in shear-wave polarisations. Tectonophysics, 426(3–4), 253–262. https://doi.org/10.1016/j.tecto.2006.07.013 Giannopoulos, D., Sokos, E., Konstantinou, K. I., & Tselentis, G. -A. (2015). Shear wave splitting and VP/VS variations before and after the Efpalio earthquake sequence, western Gulf of Corinth, Greece. Geophys.

J.

Int.,

200,

1436–1448.

https://doi.org/10.1093/gji/ggu467 Hatzfeld, D., Kementzetzidou, D., Karakostas, V., Ziazia, M., Nothard, S., Diagourtas, D., Deschamps, a., Karakaisis, G., Papadimitriou, P., Scordilis, M., Smith, R., Voulgaris, N., Kiratzi, S., Makropoulos, K., Bouin, M. P., & Bernard, P. (1996). The Galaxidi earthquake of 18 November 1992: A possible asperity within the normal fault system of the Gulf of Corinth (Greece). Bull. Seismol. Soc. Am., 86(6), 1987–1991. Hatzfeld, D., Karakostas, V., Ziazia, M., Kassaras, I., Papadimitriou, E., Makropoulos, K., Voulgaris, N., & Papaioannou, C. (2000). Microseismicity and faulting geometry in the Gulf of Corinth (Greece).

Geophys.

J.

Int.,

141(2),

438–456.

https://doi.org/10.1046/j.1365-246x.2000.00092.x Houghton, S., Roberts, G., Papanikolaou, I., McArthur, J., & Gilmour, M. (2003). New

234

U-230Th coral dates from the western Gulf of

Corinth: Implications for extensional tectonics. Geophys. Res. Lett., 30(19), 3–6. https://doi.org/10.1029/2003GL018112

27

Jackson, J. A., Gagnepain, J., Houseman, G., King, G. C. P., Papadimitriou, P., Soufleris, C., & Virieux, J. (1982). Seismicity, normal faulting and the geomorphological development of the Gulf of Corinth (Greece): the Corinth earthquakes of February and March 1981. Earth Planet. Sci. Lett., 57, 377-397. Kaneshima, S. (1990). Origin of crustal anisotropy: Shear Wave splitting studies

in

Japan.

J.

Geophys.

Res.,

95,

11121.

https://doi.org/10.1029/JB095iB07p11121 Kapetanidis, V., & Papadimitriou, P. (2011). Estimation of arrival-times in intense seismic sequences using a Master-Events methodology based on waveform similarity. Geophys. J. Int., 187(2), 889–917. https://doi.org/10.1111/j.1365-246X.2011.05178.x Kapetanidis, V., Deschamps,

A., Papadimitriou, P., Matrullo, E.,

Karakonstantis, A., Bozionelos, G., Kaviris, G., Serpetsidaki, A., Lyon-Caen, H., Voulgaris, N., Bernard, P., Sokos, E., & Makropoulos, K. (2015). The 2013 earthquake swarm in Helike, Greece: seismic activity at the root of old normal faults. Geophys. J. Int., 202(3), 2044–2073. https://doi.org/10.1093/gji/ggv249 Karakostas, V., Karagianni, E., & Paradisopoulou, P. (2012). Space-time analysis, faulting and triggering of the 2010 earthquake doublet in western

Corinth

Gulf.

Nat.

Hazards,

63(2),

1181–1202.

https://doi.org/10.1007/s11069-012-0219-0 Kaviris, G., Papadimitriou, P., & Makropoulos, K. (2008). An Overview of Anisotropy Studies in Central Greece using recordings around the Gulf of Corinth ( Greece ) and aftershocks of the 1999 Athens Earthquake. Full Paper, Proc. 31st General Assembly of the European Seismological Commission. 215–223, Hersonissos, Crete, Greece.

28

Kaviris, G., Papadimitriou, P., & Makropoulos, K. (2010). Anisotropy Study of the February 4th 2008 Swarm in NW Peloponnesus (Greece). In Bulletin of the Geological Society of Greece, Proceedings of the 12th International Congress, 2084–2092, Patras, Greece. Kaviris, G., Kapetanidis, V., Kravvariti, P., Karakonstantis, A., Bozionelos, G., Papadimitriou, P., Voulgaris, N., & Makropoulos, K. (2014). Anisotropy Study in Villia (E. Corinth Gulf, Greece). Full Paper, 2nd European Conference on Earthquake Engineering and Seismology, Istanbul, Turkey. Kaviris, G., Papadimitriou, P., Kravvariti, P. h., Kapetanidis, V., Karakonstantis, A., Voulgaris, N., & Makropoulos, K. (2015). A detailed seismic anisotropy study during the 2011-2012 unrest period in the Santorini Volcanic Complex. Phys. Earth Planet. Inter., 238, 51–88. https://doi.org/10.1016/j.pepi.2014.11.002 Klein, F.W (1989) HYPOINVERSE, a program for VAX computers to solve for earthquake locations and magnitudes. U.S. Geological Survey Open-File Report 89-314, 59 pp. Kortekaas, S., Papadopoulos, G. A., Ganas, A., Cundy, A. B., & Diakantoni, A. (2011). Geological identification of historical tsunamis in the Gulf of Corinth, Central Greece. Nat. Hazards Earth Syst. Sci., 11(7), 2029–2041. https://doi.org/10.5194/nhess-11-2029-2011 Koukouvelas, I. K. & Doutsos, T. (1996). Implication of structural segmentation during earthquakes: the 1995 Egion earthquake, Gulf of Corinth, Greece. J. Struct. Geol., 18(12), 1381-1388. Kouskouna, V., & Makropoulos, K. (2004). Historical earthquake investigations in Greece. Ann. Geophys., 47(2–3), 723–731. https://doi.org/10.4401/ag-3333

29

Lewis, M. A., & Gerstoft, P. (2012). Shear wave anisotropy from crosscorrelation of seismic noise in the Parkfield pilot hole. Geophys. J. Int.,

188(2),

626–630.

https://doi.org/10.1111/j.1365-

246X.2011.05285.x Liu, Y., Crampin, S., & Main, I. (1997). Shear-wave anisotropy: spatial and temporal variations in time delays at Parkfield, Central California. Geophys. J. Int., 130(3), 771–785. https://doi.org/10.1111/j.1365246X.1997.tb01872.x Liu, Y., Teng, T. L., & Ben-Zion, Y. (2004). Systematic analysis of shearwave splitting in the aftershock zone of the 1999 Chi-Chi, Taiwan, earthquake: Shallow crustal anisotropy and lack of precursory variations.

Bull.

Seismol.

Soc.

Am.,

94(6),

2330–2347.

https://doi.org/10.1785/0120030139 Liu, Y., Zhang, H., Thurber, C., & Roecker, S. (2008). Shear wave anisotropy in the crust around the San Andreas fault near Parkfield: Spatial and temporal analysis. Geophys. J. Int., 172(3), 957–970. https://doi.org/10.1111/j.1365-246X.2007.03618.x Loftus, D. L., & Tsoflias, P. (1971). Geological map of Greece, sheet Nafpaktos, 1:50.000. I.G.R.S., Athens. Lyon-Caen,

H.,

Papadimitriou,

P.,

Deschamps,

A.,

Bernard,

P.,

Makropoulos, K., Pacchiani, F., & Patau, G. (2004). First results of the CRLN seismic network in the western Corinth Rift: evidence for old-fault reactivation. C. R. Geosci., 336(4–5), 343–351. https://doi.org/10.1016/j.crte.2003.12.004 Makropoulos, K. C., & Burton, P. W. (1984). Greek Tectonics and Seismicity. Tectonophysics, 106, 275–304. Makropoulos, K., Kaviris, G., & Kouskouna, V. (2012). An updated and extended earthquake catalogue for Greece and adjacent areas since

30

1900. Nat. Hazards Earth Syst. Sci., 12(5), 1425–1430. https://doi.org/10.5194/nhess-12-1425-2012 Mesimeri, M., Karakostas, V., Papadimitriou, E., Schaff, D., & Tsaklidis, G. (2016). Spatio-temporal properties and evolution of the 2013 Aigion earthquake swarm (Corinth Gulf, Greece). J. Seismol., 20(2), 595–614. https://doi.org/10.1007/s10950-015-9546-4 Moretti, I., Sakellariou, D., Lykousis, V., & Micarelli, L. (2003). The Gulf of Corinth: An active half graben? J. Geodyn., 36(1–2), 323–340. https://doi.org/10.1016/S0264-3707(03)00053-X Nur, A., & Simmons, G. (1969). Stress-induced velocity anisotropy in rock: An experimental study. J. Geophys. Res., 74(27), 6667–6674. https://doi.org/10.1029/JB074i027p06667 Palyvos, N., Pantosti, D., De Martini, P. M., Lemeille, F., Sorel, D., & Pavlopoulos, K. (2005). The Aigion-Neos Erineos coastal normal fault

system

(western

Corinth

Gulf

Rift,

Greece):

Geomorphological signature, recent earthquake history, and evolution. J. Geophys. Res. B Solid Earth, 110(9), 1–15. https://doi.org/10.1029/2004JB003165 Pantosti, D., De Martini, P. M., Koukouvelas, I., Stamatopoulos, L., Palyvos, N., Pucci, S., Lemeille, F., & Pavlides, S. (2004). Palaeoseismological investigations of the Aigion Fault (Gulf of Corinth,

Greece).

C.

R.

Geosci.,

336(4–5),

335–342.

https://doi.org/10.1016/j.crte.2003.12.005 Papadimitriou, P., Kaviris, G., & Makropoulos, K. (1999). Evidence of shear-wave splitting in the eastern Corinthian Gulf (Greece). Phys. Earth Planet. Inter., 114(1–2), 3–13. Papadimitriou, P., Kaviris, G., Karakonstantis, A., & Makropoulos, K. (2010). The Cornet seismological network : 10 years of operation ,

31

recorded seismicity and significant applications. Ann. Geol Des Pays Hell., 45, 193–208. Papadopoulos, G. A. (2003). Tsunami hazard in the Eastern Mediterranean: Strong earthquakes and tsunamis in the Corinth Gulf, Central Greece. Nat. Hazards, 29, 437–464. Papanastassiou, D. (2011). Earthquake detection-location capability of the Hellenic Unified Seismological Network (HUSN) operating by the Institute of Geodynamics, National Observatory of Athens. Hellenic J. Geosc., 45, 209–216. Papanikolaou, D., Chronis, G., Lykousis, V., Sakellariou, D., & Papoulia, I. (1997) Submarine neotectonic structure of W. Korinthiakos gulf and

geodynamic

phenomena

of the

Egion

earthquake.

Proceedings of the 5th Panhellenic Symposium of Oceanography and Fisheries, I, 415 – 418. Papazachos, B.C., & Papazachou, C. (2003). The Earthquakes of Greece. Ziti Publ. Co., p. 286 (in Greek). Paraschoudis, B. (1977). Geological map of Greece, sheet Amygdalea, 1:50.000. I.G.M.R., Athens Pastori, M., Piccinini, D., Margheriti, L., Improta, L., Valoroso, L., Chiaraluce, L., & Chiarabba, C. (2009). Stress aligned cracks in the upper crust of the Val d’Agri region as revealed by shear wave splitting.

Geophys.

J.

Int.,

179(1),

601–614.

https://doi.org/10.1111/j.1365-246X.2009.04302.x Peng, Z., & Ben-Zion, Y. (2005). Spatiotemporal variations of crustal anisotropy from similar events in aftershocks of the 1999 M 7.4 İzmit and M 7.1 Düzce, Turkey, earthquake sequences. Geophys. J. Int.,

160(3),

1027–1043.

https://doi.org/10.1111/j.1365-

246X.2005.02569.x

32

Piccinini, D., Margheriti, L., Chiaraluce, L., & Cocco, M. (2006). Space and time variations of crustal anisotropy during the 1997 UmbriaMarche, central Italy, seismic sequence. Geophys. J. Int., 167(3), 1482–1490. https://doi.org/10.1111/j.1365-246X.2006.03112.x Rigo, A., Lyon-Caen, H., Armijo, R., Deschamps, A., Hatzfeld, D., Makropoulos, K., Papadimitriou, P., & Kassaras, I. (1996). A microseismic study in the western part of the Gulf of Corinth (Greece): implications for large-scale normal faulting mechanisms. Geophys. J. Int., 126, 663–688. Silver, P. G., & Chan, W. W. (1991). Shear Wave Splitting and Subcontinental Mantle Deformation. J. Geophys. Res., 96(B10), 16429–16454. https://doi.org/10.1029/91JB00899 Sokos, E., Zahradník, J., Kiratzi, A., Janský, J., Gallovič, F., Novotny, O., Kostelecký, J., Serpetsidaki, A., & Tselentis, G. - A. (2012). The January 2010 Efpalio earthquake sequence in the western Corinth Gulf

(Greece).

Tectonophysics,

530–531,

299–309.

https://doi.org/10.1016/j.tecto.2012.01.005 Soter, S., & Katsonopoulou, D. (2011). Submergence and uplift of settlements in the area of Helike, Greece, from the Early Bronze Age

to

late

antiquity.

Geoarchaeology,

26(4),

584–610.

https://doi.org/10.1002/gea.20366 Stefatos, A., Papatheodorou, G., Ferentinos, G., Leeder, M., & Collier, R. (2002). Seismic reflection imaging of active offshore faults in the Gulf of Corinth: their seismotectonic significance. Basin Res., 14(4), 487–502. https://doi.org/10.1046/j.1365-2117.2002.00176.x Stucchi, M., Rovida, A., Gomez Capera, a. a., Alexandre, P., Camelbeeck, T., Demircioglu, M. B., Gasperini, P., Kouskouna, V., Musson, R. M. W., Radulian, M., Sesetyan, K., Vilanova, S., Baumont, D.,

33

Bungum, H., Fäh, D., Lenhardt, W., Makropoulos, K., Martinez Solares, J. M., Scotti, O., Živčić, M., Albini, P., Batllo, J., Papaioannou, C., Tatevossian, R., Locati, M., Meletti, C., Viganò, D., & Giardini, D. (2013). The SHARE European Earthquake Catalogue (SHEEC) 1000–1899. J. Seismol., 17(2), 523–544. https://doi.org/10.1007/s10950-012-9335-2 Taylor, B., Weiss, J. R., Goodliffe, A. M., Sachpazi, M., Laigle, M., & Hirn, A. (2011). The structures, stratigraphy and evolution of the Gulf of Corinth rift, Greece. Geophys. J. Int., 185(3), 1189–1219. https://doi.org/10.1111/j.1365-246X.2011.05014.x Tranos, M.D. (2016). The seismogenic fault of the 2010 Efpalion moderatesize seismic sequence (western Corinth gulf, Central Greece). J. Seismol., doi: 10.1007/s10950-016-9601-9 Tsapanos, T., Koravos, G., Tsapanos, M., & Burton, P. W., (2010). Probabilities of earthquakes occurrence in the Corinth gulf and its vicinity inferred from combined information. Scientific Annals of the School of Geology, AUTH, Proceedings of the XIX CBGA Congress, Thessaloniki, Greece, 99, 403-409. Valcke, S. L. a., Casey, M., Lloyd, G. E., Kendall, J.-M., & Fisher, Q. J. (2006). Lattice preferred orientation and seismic anisotropy in sedimentary

rocks.

Geophys.

J.

Int.,

166(2),

652–666.

https://doi.org/10.1111/j.1365-246X.2006.02987.x Volti, T. & Crampin, S., (2003). A four-year study of shear-wave splitting in Iceland: 2-Temporal changes before earthquakes and volcanic eruptions. In: Nieuwland, R. (Ed.), New insights into structural interpretation and modelling, Geol. Soc., Spec. Publ., 212., p. 135. Waldhauser, F., & Ellsworth, W. (2000). A double-difference earthquake location algorithm: Method and application to the northern

34

Hayward fault, California. Bull. Seismol. Soc. Am., 90(6), 1353– 1368. Zatsepin, S., & Crampin, S. (1997). Modelling the compliance of crustal rock—I. Response of shear-wave splitting to differential stress. Geophys. J. Int., 129, 477–494.

35

Figure Captions Figure 1. Seismotectonic map of the broader area of the Western Gulf of Corinth. Blue circles indicate relocated epicenters of events that occurred during the period between September 2011 and October 2014, whereas red ones mark events that occurred in 2013 which were utilized in the current study. The dented lines indicate known faults in the area, mostly based on Armijo et al. (1996), Moretti et al. (2003), Palyvos et al. (2005), Bell et al. (2008) and Beckers et al. (2015). Figure 2a. Example of the polarigram and hodogram method, used to measure the shear-wave splitting parameters. The earthquake occurred on 06/05/2013 12:41 (GMT) at a depth of 9.6 km with M=2.2. The waveforms were recorded by PSAR station, while the ray-path presented an azimuth equal to 52⁰ and an angle of incidence equal to 15⁰ , within the shear-wave window. The respective panels show (top) the unfiltered waveforms, (upper middle) the filtered ones, (bottom middle) the polarigram and (bottom) the hodogram. The original waveforms and particle motion plots are presented, where the pick (red line) and the measurement of the polarization direction (105 N⁰ E, indicated by the black arrows in both the polarigram and the hodogram) of the Sfast were performed Figure 2b. Traces in the Sfast (F) and Sslow (S) directions, after rotation of the original traces according to the polarization direction (105 N⁰ E). In this stage, the time-delay (Td) between the two shear-wave components is measured (90 ms). Notation as in Fig. 2a. Figure 2c. Traces in the N-S and E-W components, after performing a temporal correction equal to the Td (90 ms) and re-rotation according to the polarization direction (-105 N⁰ E) The source polarization is then measured (145 N⁰ E). Notation as in Fig. 2a.

36

Figure 3. Rose diagrams of the Sfast polarization direction per station. Each bin is equal to 5⁰ , while F is the number of samples per gridline. Figure 4. Polar equal-area projections of the upper hemisphere for the Sfast polarizations at each station. The length of each bar is proportional to the observed time-delay (Td). Figure 5. (a) Temporal variation of the Sfast polarization direction for each station. Each circle represents a measurement, while the lines denote the mean value for each dataset. (b) Box-plots of the polarization direction per station. The circles indicate the mean values, while the horizontal lines inside the boxes indicate the median ones. The box edges denote the 25% and 75% limits of the values and the vertical whiskers mark the extreme values of the data. Figure 6a. Temporal variation of the normalized time-delays in “Band-1” for stations EFP, KALE and LAKA. At the top, the histogram of daily occurrence of earthquakes for 2013 and events with magnitude greater than or equal to 3.4 are presented. Each circle in the scatter plots represents a normalized time-delay value with its respective error bars. The red lines show the 3-point moving average for the datasets, while the dashed lines exhibit the respective standard deviation. Vertical downward arrows in the lower panels indicate the occurrence of events with ML ≥ 3.0. All three sketches, in each panel, share a common time axis. Figure 6b. Temporal variation of the normalized time-delays in “Band-1” for stations PSAR, ROD3 and TRIZ. Notation as in Fig. 6a. Figure 7a. Temporal variation of normalized time-delays in “All-bands” for stations EFP, KALE and LAKA. Notation as in Fig. 6a. Figure 7b. Figure 7. Temporal variation of normalized time-delays in “Allbands” for stations PSAR, ROD3 and TRIZ. Notation as in Fig. 6a.

37

Figure 8. Stations utilized in the study and their corresponding shear-wave splitting parameters. The red lines denote the Sfast polarization direction, while their length is proportional to the measured time-delay. Circles are the epicenters of events that led to successful shear-wave splitting measurements in any station. Arrows indicate the direction of regional extension in the WGoC. Table Captions Table 1. Shear-wave splitting parameters for 9 stations in the WGoC, where N is the number of measurements. The mean polarization direction (φ), time-delay (Td) and normalized time-delay (Tn) with their corresponding standard error of the mean (δφ, δTd and δTn) and the width of the 90% confidence interval (c.i.) are presented. For the stations where the Td errors are smaller than the reading error (5 ms), the latter is kept. Table 2. Linear regression statistics for the temporal variation of the normalized time-delays in “Band-1” for each station per time period that presented statistical significance. N is the number of samples, b the slope and its corresponding error δb, R2 the correlation coefficient and p the pvalue of the Student’s t-test for the statistical significance of the slope. Rows in Italics denote time periods related to the July Cluster while the ones in bold highlight time periods that presented statistical significance in both “Band-1” and “All-Bands” for the same station. Table 3. Linear regression statistics for the temporal variation of the normalized time-delays in “All-Bands” for each station per time period that presented statistical significance. The notation is the same as in Table 2.

38

39

40

41

42

43

44

45

46

47

48

49

50

Tables and Table Captions Table 1. Shear-wave splitting parameters for 9 stations in the WGoC, where N is the number of measurements. The mean polarization direction (φ), time-delay (Td) and normalized time-delay (Tn) with their corresponding standard error of the mean (δφ, δTd and δTn) and the width of the 90% confidence interval (c.i.) are presented. For the stations where the Td errors are smaller than the reading error (5 ms), the latter is kept.

Station

N

φ (N⁰E)

δφ (⁰)

Td (ms)

δTd (ms)

Td 90% c.i. (ms)

Tn (ms/km)

δTn (ms/km)

Tn 90% c.i. (ms/km)

EFP KALE LAKA PSAR PYRG ROD3 SERG TRIZ VVK

68 60 51 135 30 52 72 152 39

94 100 122 98 51 107 51 115 103

5 5 5 3 11 5 6 3 8

85 79 71 88 67 83 59 62 56

5 5 5 5 7 5 5 5 5

6 5 6 5 9 7 5 5 5

8.32 6.45 6.05 8.12 6.61 8.56 5.66 5.98 4.92

0.48 0.33 0.47 0.31 0.74 0.70 0.32 0.21 0.34

0.63 0.42 0.60 0.40 0.97 0.90 0.42 0.27 0.45

Table 2. Linear regression statistics for the temporal variation of the normalized time-delays in “Band-1” for each station per time period that presented statistical significance. N is the number of samples, b the slope and its corresponding error δb, R2 the correlation coefficient and p the pvalue of the Student’s t-test for the statistical significance of the slope. Rows in Italics denote time periods related to the July Cluster while the ones in bold highlight time periods that presented statistical significance in both “Band-1” and “All-Bands” for the same station.

“Band-1” Time Period

Station

N

23/05 – 22/06 27/05 – 30/05 12/07 – 22/07 16/09 – 27/09 06/07 – 11/07 18/08 – 28/09 03/05 – 26/05 03/08 – 18/08

KALE LAKA LAKA PSAR ROD3 ROD3 TRIZ TRIZ

9 7 10 4 5 5 4 4

b (mskm-1day-1) 0.228 -1.777 -0.638 1.311 -3.492 -0.359 0.220 -0.933

51

δb (mskm-1day-1) 0.068 0.323 0.148 0.134 1.655 0.213 0.001 0.143

R2

p (%)

0.78 0.93 0.78 0.99 0.77 0.70 1.00 0.98

0.12 0.30 0.80 1.00 1.25 2.70 0.00 2.30

Table 3. Linear regression statistics for the temporal variation of the normalized time-delays in “All-Bands” for each station per time period that presented statistical significance. The notation is the same as in Table 2.

“All-Bands” Time Period

Station

N

12/05 – 30/05 27/05 – 13/06 16/09 – 27/09 05/07 – 08/07 06/07 – 11/07 06/07 – 11/07 18/08 – 28/09 03/05 – 26/05 19/07 – 06/08 03/08 – 18/08

EFP EFP PSAR PSAR PSAR ROD3 ROD3 TRIZ TRIZ TRIZ

5 5 5 9 5 5 10 9 6 7

b (mskm-1day-1) 0.838 -0.756 1.282 5.689 6.880 -3.492 -0.215 0.208 0.609 -0.637

52

δb (mskm-1day-1) 0.084 0.120 0.141 1.284 1.525 1.655 0.092 0.043 0.246 0.237

R2

p (%)

0.99 0.96 0.98 0.86 0.93 0.77 0.77 0.88 0.78 0.77

0.20 0.80 0.30 0.30 2.00 1.25 1.00 0.20 6.80 4.30

Highlights     

Shear-wave splitting parameters were determined using polarigrams and hodograms. Results were obtained for 9 stations in the W. Gulf of Corinth (Greece) for 2013. Most anisotropy directions agree with the stress regime, the EDA and APE models. PYRG and SERG exhibited NE–SW Sfast directions, likely related to minor structures. Temporal variations of time-delays can be associated with significant events.

53