Relocation of the 2018 Zakynthos, Greece, aftershock sequence: spatiotemporal analysis deciphering mechanism diversity and aftershock statistics

Abstract

An M_w 6.8 earthquake occurred on October 25, 2018, 35 km offshore from the southwest coastlines of Zakynthos Island. The aftershock sequence appeared remarkably productive with six aftershocks of M ≥ 5.0 in the first month and tens of aftershocks with M ≥ 4.0 during the study period. The GCMT solution for the main shock suggests a very low angle plane (dip = 24°) for a dextral strike–slip faulting (rake = 165°). A similar solution is suggested for the largest aftershock (M_w 5.9) that occurred 5 days afterward. The proximity of the main shock location with the dextral active boundary of Kefalonia Transform Fault Zone (KTFZ) along with the Hellenic Subduction front supports this oblique faulting. The aftershock activity is comprised mostly in depths 5–12 km and forms eight distinctive clusters that accommodate regional strain and evidence strain partitioning. The role of stress transfer and statistical analysis are combined for detailing the highly productive aftershock sequence. Earthquake networks analysis reveals their random structure soon after the main shock, which became small-world structure after the first 200 days. Time series analysis constructed from the aftershock frequency and seismic moment release and manifested significant correlation among the eight seismicity clusters.

Appendices

Appendices

Appendix Α: Magnitudes conversion definition of the completeness magnitude, M_c

Local magnitudes were calculated during the routine earthquake analysis performed by the analysts of the Geophysics Department of Aristotle University of Thessaloniki (GD–AUTh) http://geophysics.geo.auth.gr/ss/). For our aftershock catalog the magnitude estimation was performed by using the recordings of the stations of the Hellenic Unified Seismic Network (HUSN) up to distances of 250 km from the central cluster of the sequence. Either magnitude overestimation or underestimation was consistently observed at many of these stations, when each estimate was compared to the final M_L assigned to the certain earthquake (Fig. 15). As a result, the final magnitudes would be over- or under-estimated depending upon the stations used in the calculation, thus producing a non-homogeneous catalog.

Fig. 15

Histogram of the medians of the residuals between the local magnitude estimated at a certain station, M_L-station, and the corresponding local magnitude of the earthquake as reported in the catalog, M_L-earthquake, for every station used in the M_L calculations, with the numbers on every bar denoting the number of times that the station contributed to this calculation. Consistent either over-estimation or under-estimation up to ± 0.2 can be seen on many of the stations with more than 1000 observations

In order to overcome the problem and the associated ones when defining the completeness magnitude, M_c, and when examining the variation of b-values we converted the calculated local magnitudes, M_L, to the moment magnitude scale, M_W. The moment magnitude is commonly accepted as a more reliable measure of earthquake size than local magnitude since the seismic moment is directly related to fault dimensions and slip (Aki and Richards 1980) and its uncertainties are much smaller (Kagan 2003). An M_W catalog was compiled comprising 62 earthquakes, using 14 centroid moment tensor solutions from the GCM.T catalog (https://www.globalcmt.org/CMTsearch.html), 42 calculated in the current study and 2 from the National Observatory of Athens (NOA) (denoted by a star, “*,” in Table 5). One of those two earthquakes occurred a few days before the main shock and one in February 2018, when the seismic network was at the same state as it was during the aftershock sequence. For those 62 earthquakes and each individual station, the M_L magnitudes were calculated using the relationships as in the routine analysis (Hutton and Boore 1987). The number of observations per station is plotted in Fig. 16, in order to ascertain for which stations an adequate set of M_L observations was available, for converting M_L to M_W, by establishing the proper scaling equations. For calculating the scaling relations, we used only twenty stations with more than ten observations each.

Fig. 16

Histogram of the M_L observations per station for the earthquakes for which centroid moment tensor solution is available. Dashed red line separates stations with more than ten (10) observations, used to calculate scaling relationships

Linear relationships between M_w and M_L were computed by orthogonal regression (Leptokaropoulos et al. 2013; and references therein) (Fig. 17). This approach was chosen over the standard linear regression, as the latter may introduce systematic effects leading to b-value bias (Castellaro et al. 2006). We used the obtained equations to convert the local magnitudes to moment magnitudes and then to recalculate the magnitudes for the entire aftershock catalog.

Fig. 17

Relationships between M_W and M_L for the stations used in the magnitude conversion, represented by the green lines. Blue scattered dots depict the M_W values of the centroid moment tensor catalog and their corresponding M_L, taken from the bulletin of the Seismological Station of GD–A.U.Th. The dashed red lines represent the y = x line

We proceeded to the definition of the completeness magnitude, M_C, for all data, applying the goodness-of-fit test (Wiemer and Wyss 2000) and using the converted catalog. The b- and a- values of the GR law are calculated from the observed distribution by applying a maximum likelihood estimation as a function of for all earthquakes above a minimum magnitude. Then, a perfect synthetic frequency magnitude distribution using the same GR law parameters and minimum magnitude is calculated and the absolute difference between the number of earthquakes in the observed and synthetic catalogs is calculated as in Wiemer and Wyss (2000). Mapping of the residuals of the absolute differences between observed and synthetic data shows that a completeness magnitude, M_c, of 3.2 is appropriate for this dataset at the 95% level (Fig. 18).

Fig. 18

Residuals of the goodness-of-fit method as a function of lower magnitude cutoff, for the converted catalog containing 4116 earthquakes. Horizontal dotted lines represent the 5% and 10% residual threshold. The red triangle marks the completeness magnitude, M_c, which equals to 3.2

Appendix B: Temporal variation of b-value in relation to cumulative seismic moment (ΣΜ_ο)

Figure 19 shows the temporal variation of b-value and the released cumulative seismic moment, both calculated in subsets of 50 aftershocks, in moving windows with a step of one event per time, and plotted at the occurrence time of the last event in each subset. In the first days after the main shock, the strongest aftershocks occurred, and the cumulative seismic moment (blue lines in Fig. 19) attains high values. At the same time, the b-values (red lines in Fig. 19) got the lower values, which is expressed by the anticorrelation in the temporal variation of the two estimates. After the first 50 days, the level of the cumulative seismic moment became lower and the corresponding b-values increased and remained relatively stable for each cluster.

Fig. 19

Temporal variation of the b-value calculated in subsets of 50 aftershocks, moving one event per calculation, along with the corresponding cumulative seismic moment (ΣΜ_ο) (in logarithmic scale), and the earthquakes with magnitude $M \ge 4.0$ for each cluster