- Email: [email protected]
A ten year Moment Tensor database for Western Greece
Accepted Manuscript A Ten Year Moment Tensor Database for Western Greece Dr Anna Serpetsidaki, Efthimios Sokos, G-Akis Tselentis PII:
S1474-7065(16)30054-7
DOI:
10.1016/j.pce.2016.04.007
Reference:
JPCE 2487
To appear in:
Physics and Chemistry of the Earth
Received Date: 29 June 2015 Revised Date:
15 April 2016
Accepted Date: 27 April 2016
Please cite this article as: Serpetsidaki, A., Sokos, E., Tselentis, G.-A., A Ten Year Moment Tensor Database for Western Greece, Physics and Chemistry of the Earth (2016), doi: 10.1016/ j.pce.2016.04.007. 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.
ACCEPTED MANUSCRIPT
A Ten Year Moment Tensor Database for Western Greece Anna Serpetsidaki1, Efthimios Sokos1 and G-Akis Tselentis1
Affiliations 1
RI PT
. Seismological Laboratory, Geology Department, University of Patras, Patras, Greece
e-mail addresses
SC
[email protected]
[email protected]
Corresponding author: Dr. Serpetsidaki Anna
TE D
Seismological Laboratory
M AN U
[email protected]
Patras University Campus, Rio Rio, Patras – 26504 Tel: +302610969369
EP
Fax: +302610990639
AC C
[email protected]
Keywords: Moment Tensor, Focal mechanisms, Seismotectonics, Western Greece
Abstract Moment Tensors (MTs) provide important information for seismotectonic, stress distribution and source studies. It is also important as a real time or near real time information in shakemaps, tsunami warning, and stress transfer. Therefore a reliable and rapid MT computation is a routine
1
ACCEPTED MANUSCRIPT task for modern seismic networks with broadband sensors and real-time digital telemetry. In this paper we present the database of Moment Tensor solutions computed during the last ten years in Western Greece by the University of Patras, Seismological Laboratory (UPSL). The data from UPSL broad band network were used together with the ISOLA Moment Tensor inversion package for routine MT calculation. The procedures followed and the comparison of UPSL derived solutions with
RI PT
the ones provided by other agencies for Western Greece region are presented as well. The moment tensor database includes solutions for events in the magnitude range 2.8 - 6.8 and provides a unique insight into the faulting characteristics of Western Greece. Moreover it paves the way for detailed studies of stress tensor and stress transfer. The weak events’ Moment Tensor included in UPSL’s
faults, which may be critical in seismic hazard estimation.
M AN U
1. Introduction
SC
database are important for the comprehension of local seismotectonics and reveal the role of minor
Since 2002 the University of Patras Seismological Laboratory (UPSL), has began an effort to upgrade its permanent seismic network in Western Greece. The main target was to upgrade the existing stations from analog to digital and install new stations in critical sites as regards the network geometry. Within two years the analog stations of the so called PATNET seismic network were substituted by digital ones while new stations were installed completing the new network called
TE D
PSLNET (code HP, Figure 1). Since 2008 PSLNET joined the three regional seismic networks of Greece to establish the new Hellenic Unified Seismic Network (HUSN); this fact increased enormously the availability of high quality broad band data, which could be used in standard seismological
EP
applications (i.e. location, magnitude determination, source inversion etc). Western Greece (Ionian Islands and Western Peloponnese) represents one of the most seismically
AC C
active regions in the Mediterranean (Hatzfled, et al, 1996). Within a small geographic area all faulting styles are found, moreover, the plate boundaries aren’t clearly defined in many places, leaving the space open for speculations on the existence of a microplate at the area (Vassilakis et al., 2011, Serpetsidaki et al., 2014, Perousse et al., 2011). The major tectonic features are the subduction of the African plate beneath the Aegean microplate along the western Hellenic trench, the Cephalonia transform fault at the northwestern end of the Hellenic arc and the Corinth Gulf continental rift. The Cephalonia Transform Fault (CTF) is a major strike-slip fault that links the subduction boundary to the continental collision between the Apulian microplate and the Hellenic foreland (Figure 1) (Louvari et al, 1999). The Corinth Gulf is an asymmetric graben and the fastest-spreading intra2
ACCEPTED MANUSCRIPT continental rift on Earth, with the geodetically measured extension varying from~5 mm/yr at the eastern part, to ~15 mm/yr at the western part (Briole et al., 2013), characterized by a high level of microseismicity. In this paper, the results of a ten year (2005-2015) routine determination of MT solutions in the area are presented. The solutions are determined for seismic events occurring for a wide range of
RI PT
magnitudes from weak Mw~3.0 to strong events Mw~6.8. In detail we present the uncertainty of the solutions in terms of quality factors such as variance reduction, condition number, inversion stability (Sokos and Zahradnik 2013) and compare the solutions to those published by other institutes. Similar attempts to routinely determine the Moment Tensor for Greece have been done by the
Aristotle University of Thessaloniki (Roumelioti et al., 2011).
SC
Geodynamical Institute of National Observatory of Athens (Konstantinou et al., 2010) and by the
M AN U
The routine determination of Moment Tensor solutions, not only complements the picture of the seismotectonic regime in Western Greece but it also reveals details in a local scale; the focal mechanism and the Mw determination lead to explicit description of the deformation in the source region and a better comprehension of stress distribution.
2. Data and Method
TE D
UPSL operates a permanent seismic network of 25 stations (Figure 1a). The stations are equipped with three-component seismometers recording in continuous mode. The sensors are mainly broad band i.e Trillium 40s (Nanometrics) and CMG-3T 120s (Guralp). The data are transmitted through
EP
satellite or telephone lines and the processing takes place in UPSL. Routine processing includes manual picking of P and S waves using the Atlas (www.nanometrics.com)
AC C
suite and location of the events using Hypoinverse (Klein, 2002); automatic location is done using SeisComp3 (www.seiscomp3.org). The events occurring in Western Greece with ML > 3.0 recorded by at least 4 stations, with a visually inspected, high signal-to-noise ratio, are used to calculate the Moment Tensor solution. The computed solutions are published in the European-Mediterranean Seismological Centre (EMSC) site (http://www.emsc-csem.org) and in the UPSL’s webpage (http://seismo.geology.upatras.gr/heliplots/mttable.html). The Moment Tensor (MT) inversion is performed for local and regional events using the ISOLA software (Sokos and Zahradnik, 2008). ISOLA is a well known software and has been widely applied during the last ten years to earthquakes of magnitudes ranging from Mw 0.2 to 9, at epicentral distances from ~1 to ~1000 km (Triantafyllis et al., 2015); for research applications see, for example, Agurto et al., (2012), Quintero et al., (2014). ISOLA is using the iterative deconvolution method of Kikuchi and Kanamori (1991), modified for 3
ACCEPTED MANUSCRIPT regional distances. Complete waveforms are used without separation of individual phases and Green functions are calculated by the discrete wave-number method. In ISOLA, MT is found by the leastsquares, time-domain minimization of the L2-norm misfit between the observed and synthetic waveforms, while the centroid position and time of the event are calculated by a spatio-temporal grid search. The computational options include inversion of the full MT, the deviatoric MT or the
RI PT
double-couple (DC) constrained MT. The mathematical background of ISOLA is described in detail in Křížová et al, 2013. For the database described in this paper moment tensor inversion was done for a single deviatoric source and in most cases the spatial grid was confined under the epicentre (centroid depth search).
SC
Several input parameters are important in Moment Tensor inversion, one of them is the frequency band used. In general the majority of inversions was performed in the frequency band 0.07-0.15Hz for weak events and in the range 0.05-0.09Hz for larger events (>4Mw). The above frequency bands
M AN U
provide the best results and are in agreement with similar studies (Konstantinou et al., 2010, Roumelioti et al., 2011). We have to denote that the two ends of the frequency band are defined by the noise and the instrument’s frequency response for the lower end and by the crustal model accuracy for the higher end. Thus, a key feature in MT inversion is the crustal model that will be used for Green function calculation. We used different velocity models during the inversion process
TE D
depending on the epicentral area. Crustal model provided by Haslinger et al., (1999) was used for ray paths crossing the Epirus and the Ionian Sea; Novonty et al., (2001) crustal model was used for the Corinth Gulf and neighboring areas while the one provided by Rigo et al., (1996) was employed for events located in the Western part of the Corinth Gulf and northern Peloponnese (Figure 1b).
EP
Besides the MT evaluation itself, of key importance is also the estimation of the solution’s quality, since it reflects the degree of MT validity. This is usually done by means of Variance Reduction (VR)
AC C
i.e. a measure of overall fit between observed and synthetic waveforms; nevertheless this is not always a reliable measure of the solution’s quality, see Sokos and Zahradnik, 2013 for a discussion on MT quality estimation. Thus in the standard MT inversion procedure, applied in UPSL, additional quality measures are also computed and reported in the final solution. These include besides VR, the number of stations, the solution’s stability based on linear inversion theory (condition number, CN) and two recently introduced indexes, the Focal-Mechanism VARiability Index (FMVAR), and the Space Time VARiability Index, described in detail in Sokos and Zahradnik, 2013. FMVAR and STVAR estimation is based on the correlation plot i.e. a plot of correlation as a function of the trial centroid position and time (Figure 2c). At each grid point a moment tensor is computed and the final solution is the one located at the grid point with the largest correlation. Naturally the 4
ACCEPTED MANUSCRIPT correlation function varies within the grid from low to high correlation areas, the type of this variation i.e. smooth or sharp, reflects the resolvability of the Moment Tensor position in space and time. Moreover the Moment Tensor solution at each grid point varies; again the variation is an indication of the solution’s stability. FMVAR quantifies the variation of the focal mechanism in the vicinity of the largest correlation value while STVAR is defined by the spatial extend of this area. The
RI PT
vicinity is defined by setting an arbitrary threshold in the correlation function, by default, in ISOLA, it is set to the 90% of the largest correlation value (Figure 2c). Large variation of FMVAR means that Moment Tensor solution is not stable close to maximum correlation indicating instability and large STVAR means that a large part of the spatiotemporal grid has correlation close to maximum, thus position and time of centroid is not defined precisely. In simple words if the correlation function has
SC
a single well defined peak at the largest correlation value, STVAR will be small. If the focal mechanism is stable around this peak then FMVAR will be small also. The quantification of the focal
M AN U
mechanism variability in FMVAR is done via the use of Kagan angle. Kagan angle between the final solution and all Moment Tensor solutions with correlation higher than the selected threshold is computed. Kagan’s method calculates an average rotation angle between two focal mechanisms and is valid for double couple mechanisms (Kagan, 1991). The angle can vary from 0o (perfect agreement between two solutions) to 120
o
(total disagreement), thus values below 60o indicate a good
agreement (Pondrelli et al., 2006). We have to note that although the spatio-temporal grid search of
TE D
centroid is applied in Moment Tensor inversion by other research groups, there is no estimation of errors in position and time of centroid, thus making difficult the comparison with the dataset presented here.
EP
Figure 2 presents an example of observed – synthetics waveform fit (Figure 2a), for a high quality (DC>90%, VR>90%) Moment Tensor solution (Figure 2b) and the corresponding correlation diagram (Figure 2c). The hatched area in Figure 2c depicts the vicinity of the largest correlation value marked
AC C
by a red beachball. The threshold that defines the vicinity, i.e. the hatched area extend, was at 90% of maximum correlation. The spatial extend of the hatched area normalized to the whole grid will result in STVAR, while the variation of focal mechanisms in it, will define FMVAR.
3. Results A decade extend database of 395 Moment Tensors is presented (Figure 3). The solutions represent mainly events of 3.0-4.5Mw (Figure 4a). The quality of the Moment Tensor solutions is evaluated through the condition number (CN), the number of stations, the VR, the FMVAR, and the STVAR (Figure 4b, c, d, e, f, respectively); 83% of the solutions were computed with coefficient CN=<10 indicating the stability of the computed Moment Tensors. The majority (80%) of the solutions were 5
ACCEPTED MANUSCRIPT determined using data from at least 5 stations, while 85% of the solutions had variance reduction (VR) higher than 50%. Furthermore the FMVAR index is smaller than 25o for the majority (84%) of the Moment Tensor solutions and the STVAR index is lower than 0.3 for the 75% of the focal mechanisms. These quality factors confirm the accuracy of the Moment Tensor solutions produced by high quality data in good azimuthal coverage.
RI PT
An important feature of the database is the inclusion of a large number of relatively weak events (<4.5Mw), computed in a routine basis, thus providing for the first time in western Greece a continuous record of such information. The inversion of weak events’ Moment Tensor can assist in the comprehension of local seismotectonics by revealing the slip character of minor faults. This is
SC
important, especially for Greece, where the published long span databases of Moment Tensor solutions (Kiratzi and Louvari, 2003) focus on events larger than 5Mw. Consequently, this database can assist the justification of detailed seismotectonic models by providing local scale information e.g.
M AN U
how major faults connect to each other through minor faults, how minor faults intersect with major faults, how minor faults are distributed on surface and how minor faults activate during an aftershock sequence.
More specific, the Moment Tensor analysis of weak events during the Movri 2008 aftershock sequence revealed the existence of strike slip focal mechanisms in the northern part of Western
TE D
Peloponnese (Figure 3). The Moment Tensors distribution suggested the existence, or at least the growth, of a large strike-slip structure in the western part of Peloponnese (Serpetsidaki et al., 2014). Furthermore, the Moment Tensor solution of weak events from the Efpalio 2010 sequence revealed that major normal faults are bounded by NE-SW trending strike-slip faults (Figure 3). This
EP
observation highlighted the role of transfer faults (Figure 3) in the western termination of Corinth Gulf, which provide the link with regional structures, such as the Trichonis and Rion-Patras fault
AC C
systems (Sokos et al., 2012). The Moment Tensor solutions of minor aftershocks of the Cephalonia 2014 sequence indicated the existence of thrust movements combined with the major strike slip motion (Figure 3) and implied the activation of a network of faults on-shore (Sokos et al., 2015). The routine Moment Tensor solution determination has implications for the seismic hazard assessment since it reveals the geometrical fault characteristics and the moment release. Although seismic hazard is controlled mainly by major faults, minor faults crossing populated areas are also significant. Such cases are abundant in western Greece, where urban areas are located at short distance or on top of active faults (e.g. Patras, Ag.Triada fault, Parcharidis et al., 2009).
6
ACCEPTED MANUSCRIPT
4. Comparison with other Moment Tensor catalogues Several international agencies routinely publish Moment Tensor solutions of events located in the region of Western Greece. The German Research Centre for Geosciences (GFZ), the Swiss Seismological Service (SED - ETH Zurich), the Global CMT group (GCMT, former Harvard) and the Regional CMT group (RCMT) by the Italian Istituto Nazionale di Geofysica e Vulcanologia (INGV) were
RI PT
selected. Depending on the availability of common Moment Tensor solutions, a comparison of the moment magnitude (Mw) and focal mechanism results was performed. The comparison of the Mw (Figure 5) reveals good agreement; the R2 coefficient varies from 0.80 to 0.97 indicating that UPSL’s results are compatible with the Mw estimates provided by international agencies. According to
SC
Figure 5, Mw determination by UPSL has a mean value of 0.1 unit difference with GFZ, ETH and GCMT. Larger differences appear in comparison with RCMT, which could be explained by a number of reasons a) different inversion method, RCMT is using intermediate period surface waves instead
M AN U
of full waveform inversion b) different treatment of centroid depth, RCMT keeps depth fixed for events shallower than 15km and c) the initial hypocentral data used by RCMT are based on teleseismic solutions (Podrelli et al, 2006).
The focal mechanism similarity between UPSL and other agencies was investigated using the Kagan angle as described previously. The results are presented in Figure 6 and show that the Kagan angle
TE D
for all cases is lower than 60o, while the mean value varies from 25o to 35o for all four agencies. More specific, for the 90% of common events with GFZ, ETH and RCMT the Kagan angle is smaller than 40o. The comparison indicates an excellent fit, as regards the focal mechanism, between UPSL solution
5. Discussion
EP
and GFZ, ETH, GCMT and RCMT solutions.
AC C
During the last ten years, ISOLA method is routinely used for Moment Tensor solutions of earthquakes, in Western Greece. Waveform data are provided by the permanent broad band network of UPSL. Analyzing strong and weak events (2.8≤Mw≤6.8), 395 Moment Tensor solutions were successfully computed. Along with the preferred solution, the method provides measures of several quality factors. A further comparison of common solutions with GFZ, ETH, GCMT and RCMT agencies, revealed a good agreement. The Moment Tensor solutions reveal the seismotectonic regime of Western Greece; around the Corinth Gulf they depict normal faulting of E-W orientation, the Kalamata fault (Southern Peloponnese) is well pictured through normal NNW-SSE focal mechanisms and the strike slip faults at the northwestern part of Peloponnese are also well described. Concerning the Ionian Sea, the 7
ACCEPTED MANUSCRIPT solutions are in agreement with the Ionian thrust in Zakynthos Island and the Cephalonia transform fault in the Western part of Cephalonia Island. The routine computation of Moment Tensor solutions is significant, for the following reasons; the vast number of solutions will lead to an accurate time – space stress tensor inversion, important for seismotectonic modeling. Furthermore, the calculation of weak to moderate events’ Moment Tensor
RI PT
reveals the slip character of small faults, such as Efpalio, Northern Peloponnese strike slip faults and the Cephalonia thrust faults. The systematic computation of Moment Tensor solutions can assist the comprehension of complex tectonic processes and reveal the role of minor faults, which may be critical during the activation of neighboring major faults i.e. stress transfer studies.
SC
Future efforts will focus on the automatic Moment Tensor inversion, which will provide real-time solutions making possible the immediate estimation of Moment Tensors during seismic crisis in
M AN U
Greece.
Acknowledgements
TE D
The authors would like to acknowledge the waveform data exchange within the Hellenic Unified Seismic Network. Figures 1 and 2 were produced using the GMT software (Wessel and Smith, 1998). Useful comments by two anonymous reviewers are highly appreciated.
References
Agurto, H., Rietbrock, A., Ryder, I. and Miller, M., 2012. Seismic-afterslip characterization of the 2010 MW 8.8
EP
Maule, Chile, earthquake based on moment tensor inversion, Geophys. Res. Lett., 39, L20303, doi:10.1029/2012GL053434.
AC C
Briole, P. & the CRL team, 2013. Ground deformation across the Corinth rift from 22 years of GPS observations, EGU General Assembly, 2013EGUGA.1513183B. Haslinger, F., Kissling, E., Ansorge, J., Hatzfeld, D., Papadimitriou, E., Karakostas, V., Makropoulos, K., Kahle, H.G., Peter, Y., 1999. 3D crustal structure from local earthquake tomography around the Gulf of Arta (Ionian region, NW Greece). Tectonophysics 304, 201–218. Hatzfeld D., L. Kassaras, D. Panagiotopoulos, D. Amorese, K. Makropoulos, G. Karakaisis, and O. Coutand (1995). Microseismicity and strain pattern in northwestern Greece, Tectonics, 14, 773-785. Kagan, Y. Y., 1991. 3-D rotation of double-couple earthquake sources, Geophys. J. Int. 106, no. 3, 709–716, doi:10.1111/j.1365-246X.1991.tb06343.x.
8
ACCEPTED MANUSCRIPT Kikuchi, M. and H., Kanamori, 1991. Inversion of complex body waves-III, Bull. Seism. Soc. Am., 81, 2335– 2350. Kiratzi, A., Louvari, E. (2003). Focal mechanisms of shallow earthquakes in the Aegean Sea and the surrounding lands determined by waveform modeling: a new database. J. Geodyn. 36, 251– 274.
and magnitudes, Open-File Rep.,U.S. Geol. Surv., 02 171, Version 1.0
RI PT
Klein, F.W., 2002. User’s guide to HYPOINVERSE-2000, a FORTRAN program to solve earthquake locations
Konstantinou, K. I., N. S. Melis, and K. Boukouras, 2010 Routine Regional Moment Tensor Inversion for Earthquakes in the Greek Region: The National Observatory of Athens (NOA) Database (2001–2006)
SC
Seismological Research Letters, v. 81, p. 750-760, doi:10.1785/gssrl.81.5.750
Křížová, D., Zahradník, J. and Kiratzi, A. (2013). Resolvability of Isotropic Component in Regional Seismic
M AN U
Moment Tensor Inversion, Bull.Seismol.Soc.Am., 103, 2460-2473.
Louvari, E., Kiratzi, A.A., Papazachos, B.C. (1999). The Cephalonia transform fault and its extension to western Lefkada island. Tectonophysics 308, 223– 236.
Novotny, O., Jansky, J., Plicka, V. and Lyon-Caen, H., 2008. A layered model of the upper crust in the Aigio region of Greece, inferred from arrival times of the 2001 earthquake sequence, Stud. Geophys. Geod., 52, 123–
TE D
131.
Parcharidis I., Kokkalas, S., Fountoulis, I. & Foumelis, M., 2009. Detection and monitoring of active faults in urban environments: Time series interferometry on the cities of Patras and Pyrgos (Peloponnese, Greece). Remote Sensing Journal 1, 676-696.
EP
Pérouse, E., Chamot-Rooke, N., Rabaute, A. and Briole P., 2011. A new model to link the present-day kinematics of the Central Mediterranean and the Eastern Mediterranean, Geophysical Research Abstract,. 13,
AC C
EGU2011-7711, EGU General Assembly, Vienna, Austria. Pondrelli, S., Salimbeni, S., Ekstrφm, G., Morelli, A., Gasperini, P., Vannucci, G., 2006. The Italian CMT dataset from 1977 to the present. Phys. Earth Planet. Int., 159/3-4, 286-303, doi:10.1016/j.pepi.2006.07.008. Quintero, R., Zahradník, J. and Sokos, E., 2014. Near-regional CMT and multiple-point source solution of the September 5, 2012, Nicoya, Costa Rica Mw 7.6 (GCMT) earthquake, Journal of South American Earth Sciences, 55, 155-165, doi:10.1016/j.jsames.2014.07.009. Rigo A., Lyon-Caen H., Armijo R., Deschamps A., Hatzfeld D., Makropoulos K., Papadimitriou P. and 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.
9
ACCEPTED MANUSCRIPT Roumelioti, Z., A. Kiratzi and Ch. Benetatos, 2011. Time Domain Moment Tensors of earthquakes in the broader Aegean Sea for the years 2006-2007: the database of the Aristotle University of Thessaloniki, Journal of Geodynamics, 51, 179-189. Serpetsidaki, A., P. Elias, M. Ilieva, P. Bernard, P. Briole, A. Deschamps, S. Lambotte, H. Lyon-Caen, E. Sokos and G.-A. Tselentis, 2014. New constraints from seismology and geodesy on the Mw = 6.4 2008 Movri (Greece) earthquake: evidence for a growing strike-slip fault system . Geophys. J. Int., 198 (3): 1373-1386
RI PT
doi:10.1093/gji/ggu212
Sokos, E., J. Zahradník, A. Kiratzi, J. Janský, F. Gallovič, O. Novotný, J. Kostelecký, A. Serpetsidaki, and G-A.
Tectonophysics, 530-531, 299-309.
SC
Tselentis, 2012. The January 2010 Efpalio earthquake sequence in the western Corinth Gulf (Greece).
Sokos, E. and Zahradnik, J., 2008. ISOLA-A Fortran code and a Matlab GUI to perform multiple-point source
M AN U
inversion of seismic data, Comput. Geosci., 34,967–977.
Sokos, E. and J. Zahradník, 2013. Evaluating centroid moment tensor uncertainty in new version of ISOLA software. Seismol. Res. Letters, 84, 656-665.
Triantafyllis N., Sokos E., Ilias A. and Zahradnik J. 2015, SCISOLA: Automatic moment tensor solution for Seiscomp3, Seismol. Res. Lett.,accepted.
TE D
Vassilakis, E., L. Royden, D. Papanikolaou, 2011. Kinematic links between subduction along the Hellenic trench and extension in the Gulf of Corinth, Greece: A multidisciplinary analysis, Earth and Planetary Science Letters, Volume 303, Issues 1–2, 15, 108-120, ISSN 0012-821X, doi:10.1016/j.epsl.2010.12.054. Wessel, P., and W. H. F. Smith, 1998. New, improved version of Generic Mapping Tools released, EOS Trans.
AC C
EP
Amer. Geophys. U. 79, no 47, 579.
Web References
geofon.gfz-potsdam.de
seismo.geology.upatras.gr/ www.nanometrics.com www.seiscomp3.org
10
ACCEPTED MANUSCRIPT www.seismo.ethz.ch www.globalcmt.org/
Figure Captions
RI PT
Figure 1. a) Map showing the main tectonic features in Western Greece; red squares represent the seismic stations operated by UPSL during the last ten years. The focal mechanism corresponds to the event of Figure 2. The cities discussed in the text are also shown b) crustal models used in Green function calculation determined by Rigo et al., 1996 (solid line), Haslinger et al., 1999 (dashed line) and Novotny et al., 2008 (dotted line).
SC
Figure 2. a) Waveform fit for a typical high quality Moment Tensor solution of event located in the Western Corinth Gulf. b) Summary of Moment Tensor solution of the event computed with ISOLA. c) correlation plot,
M AN U
hatched area designates the 90% of maximum correlation area, used in FMVAR, STVAR assessment. Figure 3. Map of the Moment Tensor solutions of the 395 events analyzed in the present study. Figure 4. Histograms showing the distribution of a) Moment Magnitude (Mw), b) Condition Number (CN), c) the number of stations used, d) Variance Reduction percentage (VR%), e) the Focal Mechanism Variability Index (FMVAR) and f) the Space-Time Variability Index (STVAR) of the computed Moment Tensor solutions.
TE D
Figure 5. Regression between moment magnitudes obtained by UPSL and a) GFZ, b) ETH, c) GCMT and d) RCMT. The dotted line represents the x=y line. The insets present the histograms of the differences between moment magnitudes of GFZ, ETH, GCMT, RCMT and UPSL. Figure 6. Diagram of the computed Kagan angle (for all common pair of events) between UPSL and GFZ, ETH,
AC C
EP
GCMT and RCMT versus cumulative percentage of Kagan angle.
11
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT We present a ten year moment tensor database for western Greece. We evaluate the database’s quality.
AC C
EP
TE D
M AN U
SC
RI PT
Our database complements the Moment Tensor catalogue for Greece including solutions of weak events.
S1474-7065(16)30054-7
DOI:
10.1016/j.pce.2016.04.007
Reference:
JPCE 2487
To appear in:
Physics and Chemistry of the Earth
Received Date: 29 June 2015 Revised Date:
15 April 2016
Accepted Date: 27 April 2016
Please cite this article as: Serpetsidaki, A., Sokos, E., Tselentis, G.-A., A Ten Year Moment Tensor Database for Western Greece, Physics and Chemistry of the Earth (2016), doi: 10.1016/ j.pce.2016.04.007. 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.
ACCEPTED MANUSCRIPT
A Ten Year Moment Tensor Database for Western Greece Anna Serpetsidaki1, Efthimios Sokos1 and G-Akis Tselentis1
Affiliations 1
RI PT
. Seismological Laboratory, Geology Department, University of Patras, Patras, Greece
e-mail addresses
SC
[email protected]
[email protected]
Corresponding author: Dr. Serpetsidaki Anna
TE D
Seismological Laboratory
M AN U
[email protected]
Patras University Campus, Rio Rio, Patras – 26504 Tel: +302610969369
EP
Fax: +302610990639
AC C
[email protected]
Keywords: Moment Tensor, Focal mechanisms, Seismotectonics, Western Greece
Abstract Moment Tensors (MTs) provide important information for seismotectonic, stress distribution and source studies. It is also important as a real time or near real time information in shakemaps, tsunami warning, and stress transfer. Therefore a reliable and rapid MT computation is a routine
1
ACCEPTED MANUSCRIPT task for modern seismic networks with broadband sensors and real-time digital telemetry. In this paper we present the database of Moment Tensor solutions computed during the last ten years in Western Greece by the University of Patras, Seismological Laboratory (UPSL). The data from UPSL broad band network were used together with the ISOLA Moment Tensor inversion package for routine MT calculation. The procedures followed and the comparison of UPSL derived solutions with
RI PT
the ones provided by other agencies for Western Greece region are presented as well. The moment tensor database includes solutions for events in the magnitude range 2.8 - 6.8 and provides a unique insight into the faulting characteristics of Western Greece. Moreover it paves the way for detailed studies of stress tensor and stress transfer. The weak events’ Moment Tensor included in UPSL’s
faults, which may be critical in seismic hazard estimation.
M AN U
1. Introduction
SC
database are important for the comprehension of local seismotectonics and reveal the role of minor
Since 2002 the University of Patras Seismological Laboratory (UPSL), has began an effort to upgrade its permanent seismic network in Western Greece. The main target was to upgrade the existing stations from analog to digital and install new stations in critical sites as regards the network geometry. Within two years the analog stations of the so called PATNET seismic network were substituted by digital ones while new stations were installed completing the new network called
TE D
PSLNET (code HP, Figure 1). Since 2008 PSLNET joined the three regional seismic networks of Greece to establish the new Hellenic Unified Seismic Network (HUSN); this fact increased enormously the availability of high quality broad band data, which could be used in standard seismological
EP
applications (i.e. location, magnitude determination, source inversion etc). Western Greece (Ionian Islands and Western Peloponnese) represents one of the most seismically
AC C
active regions in the Mediterranean (Hatzfled, et al, 1996). Within a small geographic area all faulting styles are found, moreover, the plate boundaries aren’t clearly defined in many places, leaving the space open for speculations on the existence of a microplate at the area (Vassilakis et al., 2011, Serpetsidaki et al., 2014, Perousse et al., 2011). The major tectonic features are the subduction of the African plate beneath the Aegean microplate along the western Hellenic trench, the Cephalonia transform fault at the northwestern end of the Hellenic arc and the Corinth Gulf continental rift. The Cephalonia Transform Fault (CTF) is a major strike-slip fault that links the subduction boundary to the continental collision between the Apulian microplate and the Hellenic foreland (Figure 1) (Louvari et al, 1999). The Corinth Gulf is an asymmetric graben and the fastest-spreading intra2
ACCEPTED MANUSCRIPT continental rift on Earth, with the geodetically measured extension varying from~5 mm/yr at the eastern part, to ~15 mm/yr at the western part (Briole et al., 2013), characterized by a high level of microseismicity. In this paper, the results of a ten year (2005-2015) routine determination of MT solutions in the area are presented. The solutions are determined for seismic events occurring for a wide range of
RI PT
magnitudes from weak Mw~3.0 to strong events Mw~6.8. In detail we present the uncertainty of the solutions in terms of quality factors such as variance reduction, condition number, inversion stability (Sokos and Zahradnik 2013) and compare the solutions to those published by other institutes. Similar attempts to routinely determine the Moment Tensor for Greece have been done by the
Aristotle University of Thessaloniki (Roumelioti et al., 2011).
SC
Geodynamical Institute of National Observatory of Athens (Konstantinou et al., 2010) and by the
M AN U
The routine determination of Moment Tensor solutions, not only complements the picture of the seismotectonic regime in Western Greece but it also reveals details in a local scale; the focal mechanism and the Mw determination lead to explicit description of the deformation in the source region and a better comprehension of stress distribution.
2. Data and Method
TE D
UPSL operates a permanent seismic network of 25 stations (Figure 1a). The stations are equipped with three-component seismometers recording in continuous mode. The sensors are mainly broad band i.e Trillium 40s (Nanometrics) and CMG-3T 120s (Guralp). The data are transmitted through
EP
satellite or telephone lines and the processing takes place in UPSL. Routine processing includes manual picking of P and S waves using the Atlas (www.nanometrics.com)
AC C
suite and location of the events using Hypoinverse (Klein, 2002); automatic location is done using SeisComp3 (www.seiscomp3.org). The events occurring in Western Greece with ML > 3.0 recorded by at least 4 stations, with a visually inspected, high signal-to-noise ratio, are used to calculate the Moment Tensor solution. The computed solutions are published in the European-Mediterranean Seismological Centre (EMSC) site (http://www.emsc-csem.org) and in the UPSL’s webpage (http://seismo.geology.upatras.gr/heliplots/mttable.html). The Moment Tensor (MT) inversion is performed for local and regional events using the ISOLA software (Sokos and Zahradnik, 2008). ISOLA is a well known software and has been widely applied during the last ten years to earthquakes of magnitudes ranging from Mw 0.2 to 9, at epicentral distances from ~1 to ~1000 km (Triantafyllis et al., 2015); for research applications see, for example, Agurto et al., (2012), Quintero et al., (2014). ISOLA is using the iterative deconvolution method of Kikuchi and Kanamori (1991), modified for 3
ACCEPTED MANUSCRIPT regional distances. Complete waveforms are used without separation of individual phases and Green functions are calculated by the discrete wave-number method. In ISOLA, MT is found by the leastsquares, time-domain minimization of the L2-norm misfit between the observed and synthetic waveforms, while the centroid position and time of the event are calculated by a spatio-temporal grid search. The computational options include inversion of the full MT, the deviatoric MT or the
RI PT
double-couple (DC) constrained MT. The mathematical background of ISOLA is described in detail in Křížová et al, 2013. For the database described in this paper moment tensor inversion was done for a single deviatoric source and in most cases the spatial grid was confined under the epicentre (centroid depth search).
SC
Several input parameters are important in Moment Tensor inversion, one of them is the frequency band used. In general the majority of inversions was performed in the frequency band 0.07-0.15Hz for weak events and in the range 0.05-0.09Hz for larger events (>4Mw). The above frequency bands
M AN U
provide the best results and are in agreement with similar studies (Konstantinou et al., 2010, Roumelioti et al., 2011). We have to denote that the two ends of the frequency band are defined by the noise and the instrument’s frequency response for the lower end and by the crustal model accuracy for the higher end. Thus, a key feature in MT inversion is the crustal model that will be used for Green function calculation. We used different velocity models during the inversion process
TE D
depending on the epicentral area. Crustal model provided by Haslinger et al., (1999) was used for ray paths crossing the Epirus and the Ionian Sea; Novonty et al., (2001) crustal model was used for the Corinth Gulf and neighboring areas while the one provided by Rigo et al., (1996) was employed for events located in the Western part of the Corinth Gulf and northern Peloponnese (Figure 1b).
EP
Besides the MT evaluation itself, of key importance is also the estimation of the solution’s quality, since it reflects the degree of MT validity. This is usually done by means of Variance Reduction (VR)
AC C
i.e. a measure of overall fit between observed and synthetic waveforms; nevertheless this is not always a reliable measure of the solution’s quality, see Sokos and Zahradnik, 2013 for a discussion on MT quality estimation. Thus in the standard MT inversion procedure, applied in UPSL, additional quality measures are also computed and reported in the final solution. These include besides VR, the number of stations, the solution’s stability based on linear inversion theory (condition number, CN) and two recently introduced indexes, the Focal-Mechanism VARiability Index (FMVAR), and the Space Time VARiability Index, described in detail in Sokos and Zahradnik, 2013. FMVAR and STVAR estimation is based on the correlation plot i.e. a plot of correlation as a function of the trial centroid position and time (Figure 2c). At each grid point a moment tensor is computed and the final solution is the one located at the grid point with the largest correlation. Naturally the 4
ACCEPTED MANUSCRIPT correlation function varies within the grid from low to high correlation areas, the type of this variation i.e. smooth or sharp, reflects the resolvability of the Moment Tensor position in space and time. Moreover the Moment Tensor solution at each grid point varies; again the variation is an indication of the solution’s stability. FMVAR quantifies the variation of the focal mechanism in the vicinity of the largest correlation value while STVAR is defined by the spatial extend of this area. The
RI PT
vicinity is defined by setting an arbitrary threshold in the correlation function, by default, in ISOLA, it is set to the 90% of the largest correlation value (Figure 2c). Large variation of FMVAR means that Moment Tensor solution is not stable close to maximum correlation indicating instability and large STVAR means that a large part of the spatiotemporal grid has correlation close to maximum, thus position and time of centroid is not defined precisely. In simple words if the correlation function has
SC
a single well defined peak at the largest correlation value, STVAR will be small. If the focal mechanism is stable around this peak then FMVAR will be small also. The quantification of the focal
M AN U
mechanism variability in FMVAR is done via the use of Kagan angle. Kagan angle between the final solution and all Moment Tensor solutions with correlation higher than the selected threshold is computed. Kagan’s method calculates an average rotation angle between two focal mechanisms and is valid for double couple mechanisms (Kagan, 1991). The angle can vary from 0o (perfect agreement between two solutions) to 120
o
(total disagreement), thus values below 60o indicate a good
agreement (Pondrelli et al., 2006). We have to note that although the spatio-temporal grid search of
TE D
centroid is applied in Moment Tensor inversion by other research groups, there is no estimation of errors in position and time of centroid, thus making difficult the comparison with the dataset presented here.
EP
Figure 2 presents an example of observed – synthetics waveform fit (Figure 2a), for a high quality (DC>90%, VR>90%) Moment Tensor solution (Figure 2b) and the corresponding correlation diagram (Figure 2c). The hatched area in Figure 2c depicts the vicinity of the largest correlation value marked
AC C
by a red beachball. The threshold that defines the vicinity, i.e. the hatched area extend, was at 90% of maximum correlation. The spatial extend of the hatched area normalized to the whole grid will result in STVAR, while the variation of focal mechanisms in it, will define FMVAR.
3. Results A decade extend database of 395 Moment Tensors is presented (Figure 3). The solutions represent mainly events of 3.0-4.5Mw (Figure 4a). The quality of the Moment Tensor solutions is evaluated through the condition number (CN), the number of stations, the VR, the FMVAR, and the STVAR (Figure 4b, c, d, e, f, respectively); 83% of the solutions were computed with coefficient CN=<10 indicating the stability of the computed Moment Tensors. The majority (80%) of the solutions were 5
ACCEPTED MANUSCRIPT determined using data from at least 5 stations, while 85% of the solutions had variance reduction (VR) higher than 50%. Furthermore the FMVAR index is smaller than 25o for the majority (84%) of the Moment Tensor solutions and the STVAR index is lower than 0.3 for the 75% of the focal mechanisms. These quality factors confirm the accuracy of the Moment Tensor solutions produced by high quality data in good azimuthal coverage.
RI PT
An important feature of the database is the inclusion of a large number of relatively weak events (<4.5Mw), computed in a routine basis, thus providing for the first time in western Greece a continuous record of such information. The inversion of weak events’ Moment Tensor can assist in the comprehension of local seismotectonics by revealing the slip character of minor faults. This is
SC
important, especially for Greece, where the published long span databases of Moment Tensor solutions (Kiratzi and Louvari, 2003) focus on events larger than 5Mw. Consequently, this database can assist the justification of detailed seismotectonic models by providing local scale information e.g.
M AN U
how major faults connect to each other through minor faults, how minor faults intersect with major faults, how minor faults are distributed on surface and how minor faults activate during an aftershock sequence.
More specific, the Moment Tensor analysis of weak events during the Movri 2008 aftershock sequence revealed the existence of strike slip focal mechanisms in the northern part of Western
TE D
Peloponnese (Figure 3). The Moment Tensors distribution suggested the existence, or at least the growth, of a large strike-slip structure in the western part of Peloponnese (Serpetsidaki et al., 2014). Furthermore, the Moment Tensor solution of weak events from the Efpalio 2010 sequence revealed that major normal faults are bounded by NE-SW trending strike-slip faults (Figure 3). This
EP
observation highlighted the role of transfer faults (Figure 3) in the western termination of Corinth Gulf, which provide the link with regional structures, such as the Trichonis and Rion-Patras fault
AC C
systems (Sokos et al., 2012). The Moment Tensor solutions of minor aftershocks of the Cephalonia 2014 sequence indicated the existence of thrust movements combined with the major strike slip motion (Figure 3) and implied the activation of a network of faults on-shore (Sokos et al., 2015). The routine Moment Tensor solution determination has implications for the seismic hazard assessment since it reveals the geometrical fault characteristics and the moment release. Although seismic hazard is controlled mainly by major faults, minor faults crossing populated areas are also significant. Such cases are abundant in western Greece, where urban areas are located at short distance or on top of active faults (e.g. Patras, Ag.Triada fault, Parcharidis et al., 2009).
6
ACCEPTED MANUSCRIPT
4. Comparison with other Moment Tensor catalogues Several international agencies routinely publish Moment Tensor solutions of events located in the region of Western Greece. The German Research Centre for Geosciences (GFZ), the Swiss Seismological Service (SED - ETH Zurich), the Global CMT group (GCMT, former Harvard) and the Regional CMT group (RCMT) by the Italian Istituto Nazionale di Geofysica e Vulcanologia (INGV) were
RI PT
selected. Depending on the availability of common Moment Tensor solutions, a comparison of the moment magnitude (Mw) and focal mechanism results was performed. The comparison of the Mw (Figure 5) reveals good agreement; the R2 coefficient varies from 0.80 to 0.97 indicating that UPSL’s results are compatible with the Mw estimates provided by international agencies. According to
SC
Figure 5, Mw determination by UPSL has a mean value of 0.1 unit difference with GFZ, ETH and GCMT. Larger differences appear in comparison with RCMT, which could be explained by a number of reasons a) different inversion method, RCMT is using intermediate period surface waves instead
M AN U
of full waveform inversion b) different treatment of centroid depth, RCMT keeps depth fixed for events shallower than 15km and c) the initial hypocentral data used by RCMT are based on teleseismic solutions (Podrelli et al, 2006).
The focal mechanism similarity between UPSL and other agencies was investigated using the Kagan angle as described previously. The results are presented in Figure 6 and show that the Kagan angle
TE D
for all cases is lower than 60o, while the mean value varies from 25o to 35o for all four agencies. More specific, for the 90% of common events with GFZ, ETH and RCMT the Kagan angle is smaller than 40o. The comparison indicates an excellent fit, as regards the focal mechanism, between UPSL solution
5. Discussion
EP
and GFZ, ETH, GCMT and RCMT solutions.
AC C
During the last ten years, ISOLA method is routinely used for Moment Tensor solutions of earthquakes, in Western Greece. Waveform data are provided by the permanent broad band network of UPSL. Analyzing strong and weak events (2.8≤Mw≤6.8), 395 Moment Tensor solutions were successfully computed. Along with the preferred solution, the method provides measures of several quality factors. A further comparison of common solutions with GFZ, ETH, GCMT and RCMT agencies, revealed a good agreement. The Moment Tensor solutions reveal the seismotectonic regime of Western Greece; around the Corinth Gulf they depict normal faulting of E-W orientation, the Kalamata fault (Southern Peloponnese) is well pictured through normal NNW-SSE focal mechanisms and the strike slip faults at the northwestern part of Peloponnese are also well described. Concerning the Ionian Sea, the 7
ACCEPTED MANUSCRIPT solutions are in agreement with the Ionian thrust in Zakynthos Island and the Cephalonia transform fault in the Western part of Cephalonia Island. The routine computation of Moment Tensor solutions is significant, for the following reasons; the vast number of solutions will lead to an accurate time – space stress tensor inversion, important for seismotectonic modeling. Furthermore, the calculation of weak to moderate events’ Moment Tensor
RI PT
reveals the slip character of small faults, such as Efpalio, Northern Peloponnese strike slip faults and the Cephalonia thrust faults. The systematic computation of Moment Tensor solutions can assist the comprehension of complex tectonic processes and reveal the role of minor faults, which may be critical during the activation of neighboring major faults i.e. stress transfer studies.
SC
Future efforts will focus on the automatic Moment Tensor inversion, which will provide real-time solutions making possible the immediate estimation of Moment Tensors during seismic crisis in
M AN U
Greece.
Acknowledgements
TE D
The authors would like to acknowledge the waveform data exchange within the Hellenic Unified Seismic Network. Figures 1 and 2 were produced using the GMT software (Wessel and Smith, 1998). Useful comments by two anonymous reviewers are highly appreciated.
References
Agurto, H., Rietbrock, A., Ryder, I. and Miller, M., 2012. Seismic-afterslip characterization of the 2010 MW 8.8
EP
Maule, Chile, earthquake based on moment tensor inversion, Geophys. Res. Lett., 39, L20303, doi:10.1029/2012GL053434.
AC C
Briole, P. & the CRL team, 2013. Ground deformation across the Corinth rift from 22 years of GPS observations, EGU General Assembly, 2013EGUGA.1513183B. Haslinger, F., Kissling, E., Ansorge, J., Hatzfeld, D., Papadimitriou, E., Karakostas, V., Makropoulos, K., Kahle, H.G., Peter, Y., 1999. 3D crustal structure from local earthquake tomography around the Gulf of Arta (Ionian region, NW Greece). Tectonophysics 304, 201–218. Hatzfeld D., L. Kassaras, D. Panagiotopoulos, D. Amorese, K. Makropoulos, G. Karakaisis, and O. Coutand (1995). Microseismicity and strain pattern in northwestern Greece, Tectonics, 14, 773-785. Kagan, Y. Y., 1991. 3-D rotation of double-couple earthquake sources, Geophys. J. Int. 106, no. 3, 709–716, doi:10.1111/j.1365-246X.1991.tb06343.x.
8
ACCEPTED MANUSCRIPT Kikuchi, M. and H., Kanamori, 1991. Inversion of complex body waves-III, Bull. Seism. Soc. Am., 81, 2335– 2350. Kiratzi, A., Louvari, E. (2003). Focal mechanisms of shallow earthquakes in the Aegean Sea and the surrounding lands determined by waveform modeling: a new database. J. Geodyn. 36, 251– 274.
and magnitudes, Open-File Rep.,U.S. Geol. Surv., 02 171, Version 1.0
RI PT
Klein, F.W., 2002. User’s guide to HYPOINVERSE-2000, a FORTRAN program to solve earthquake locations
Konstantinou, K. I., N. S. Melis, and K. Boukouras, 2010 Routine Regional Moment Tensor Inversion for Earthquakes in the Greek Region: The National Observatory of Athens (NOA) Database (2001–2006)
SC
Seismological Research Letters, v. 81, p. 750-760, doi:10.1785/gssrl.81.5.750
Křížová, D., Zahradník, J. and Kiratzi, A. (2013). Resolvability of Isotropic Component in Regional Seismic
M AN U
Moment Tensor Inversion, Bull.Seismol.Soc.Am., 103, 2460-2473.
Louvari, E., Kiratzi, A.A., Papazachos, B.C. (1999). The Cephalonia transform fault and its extension to western Lefkada island. Tectonophysics 308, 223– 236.
Novotny, O., Jansky, J., Plicka, V. and Lyon-Caen, H., 2008. A layered model of the upper crust in the Aigio region of Greece, inferred from arrival times of the 2001 earthquake sequence, Stud. Geophys. Geod., 52, 123–
TE D
131.
Parcharidis I., Kokkalas, S., Fountoulis, I. & Foumelis, M., 2009. Detection and monitoring of active faults in urban environments: Time series interferometry on the cities of Patras and Pyrgos (Peloponnese, Greece). Remote Sensing Journal 1, 676-696.
EP
Pérouse, E., Chamot-Rooke, N., Rabaute, A. and Briole P., 2011. A new model to link the present-day kinematics of the Central Mediterranean and the Eastern Mediterranean, Geophysical Research Abstract,. 13,
AC C
EGU2011-7711, EGU General Assembly, Vienna, Austria. Pondrelli, S., Salimbeni, S., Ekstrφm, G., Morelli, A., Gasperini, P., Vannucci, G., 2006. The Italian CMT dataset from 1977 to the present. Phys. Earth Planet. Int., 159/3-4, 286-303, doi:10.1016/j.pepi.2006.07.008. Quintero, R., Zahradník, J. and Sokos, E., 2014. Near-regional CMT and multiple-point source solution of the September 5, 2012, Nicoya, Costa Rica Mw 7.6 (GCMT) earthquake, Journal of South American Earth Sciences, 55, 155-165, doi:10.1016/j.jsames.2014.07.009. Rigo A., Lyon-Caen H., Armijo R., Deschamps A., Hatzfeld D., Makropoulos K., Papadimitriou P. and 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.
9
ACCEPTED MANUSCRIPT Roumelioti, Z., A. Kiratzi and Ch. Benetatos, 2011. Time Domain Moment Tensors of earthquakes in the broader Aegean Sea for the years 2006-2007: the database of the Aristotle University of Thessaloniki, Journal of Geodynamics, 51, 179-189. Serpetsidaki, A., P. Elias, M. Ilieva, P. Bernard, P. Briole, A. Deschamps, S. Lambotte, H. Lyon-Caen, E. Sokos and G.-A. Tselentis, 2014. New constraints from seismology and geodesy on the Mw = 6.4 2008 Movri (Greece) earthquake: evidence for a growing strike-slip fault system . Geophys. J. Int., 198 (3): 1373-1386
RI PT
doi:10.1093/gji/ggu212
Sokos, E., J. Zahradník, A. Kiratzi, J. Janský, F. Gallovič, O. Novotný, J. Kostelecký, A. Serpetsidaki, and G-A.
Tectonophysics, 530-531, 299-309.
SC
Tselentis, 2012. The January 2010 Efpalio earthquake sequence in the western Corinth Gulf (Greece).
Sokos, E. and Zahradnik, J., 2008. ISOLA-A Fortran code and a Matlab GUI to perform multiple-point source
M AN U
inversion of seismic data, Comput. Geosci., 34,967–977.
Sokos, E. and J. Zahradník, 2013. Evaluating centroid moment tensor uncertainty in new version of ISOLA software. Seismol. Res. Letters, 84, 656-665.
Triantafyllis N., Sokos E., Ilias A. and Zahradnik J. 2015, SCISOLA: Automatic moment tensor solution for Seiscomp3, Seismol. Res. Lett.,accepted.
TE D
Vassilakis, E., L. Royden, D. Papanikolaou, 2011. Kinematic links between subduction along the Hellenic trench and extension in the Gulf of Corinth, Greece: A multidisciplinary analysis, Earth and Planetary Science Letters, Volume 303, Issues 1–2, 15, 108-120, ISSN 0012-821X, doi:10.1016/j.epsl.2010.12.054. Wessel, P., and W. H. F. Smith, 1998. New, improved version of Generic Mapping Tools released, EOS Trans.
AC C
EP
Amer. Geophys. U. 79, no 47, 579.
Web References
geofon.gfz-potsdam.de
seismo.geology.upatras.gr/ www.nanometrics.com www.seiscomp3.org
10
ACCEPTED MANUSCRIPT www.seismo.ethz.ch www.globalcmt.org/
Figure Captions
RI PT
Figure 1. a) Map showing the main tectonic features in Western Greece; red squares represent the seismic stations operated by UPSL during the last ten years. The focal mechanism corresponds to the event of Figure 2. The cities discussed in the text are also shown b) crustal models used in Green function calculation determined by Rigo et al., 1996 (solid line), Haslinger et al., 1999 (dashed line) and Novotny et al., 2008 (dotted line).
SC
Figure 2. a) Waveform fit for a typical high quality Moment Tensor solution of event located in the Western Corinth Gulf. b) Summary of Moment Tensor solution of the event computed with ISOLA. c) correlation plot,
M AN U
hatched area designates the 90% of maximum correlation area, used in FMVAR, STVAR assessment. Figure 3. Map of the Moment Tensor solutions of the 395 events analyzed in the present study. Figure 4. Histograms showing the distribution of a) Moment Magnitude (Mw), b) Condition Number (CN), c) the number of stations used, d) Variance Reduction percentage (VR%), e) the Focal Mechanism Variability Index (FMVAR) and f) the Space-Time Variability Index (STVAR) of the computed Moment Tensor solutions.
TE D
Figure 5. Regression between moment magnitudes obtained by UPSL and a) GFZ, b) ETH, c) GCMT and d) RCMT. The dotted line represents the x=y line. The insets present the histograms of the differences between moment magnitudes of GFZ, ETH, GCMT, RCMT and UPSL. Figure 6. Diagram of the computed Kagan angle (for all common pair of events) between UPSL and GFZ, ETH,
AC C
EP
GCMT and RCMT versus cumulative percentage of Kagan angle.
11
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT We present a ten year moment tensor database for western Greece. We evaluate the database’s quality.
AC C
EP
TE D
M AN U
SC
RI PT
Our database complements the Moment Tensor catalogue for Greece including solutions of weak events.