A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, East Aegean Sea)

A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, East Aegean Sea)
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, EastAegean Sea) Anastasia A. Kiratzi ⁎ , Nikos Svigkas Department of Geophysics, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece a b s t r a c ta r t i c l e i n f o  Article history: Received 4 July 2013Received in revised form 1 September 2013Accepted 2 September 2013Available online xxxx Keywords: Lemnos BasinNorth Aegean BasinAegean SeaEarthquakeSlip The 8 January 2013 Mw5.8 earthquake sequence south of Lemnos Island (East Aegean Sea) is investigated. Thefocalmechanismsofthestrongestevents,obtainedbywaveformmodelling,togetherwiththeHypoDDrelocatedepicentres, clearly show the activation of a dextral strike-slip fault, no more than 14 km in length, trendingN60°E, and extending in depth from approximately 3 to 25 km, with the best determined focal depths in therange3to~13 km.Thedistributionofslipontothefaultplaneshowsa singlepatch of7 × 8 km 2 ,wheretheav-erageandthepeakslipareequalto30 cmand130 cm,respectively.Thelocusofthepeakslipiswellresolvedtobe ~2.5 kmaway from the hypocentre, towards SW. The slip modelwasused to forward modelthe distributionofPeak Ground Velocity (PGVincm/s)inthe near source.The PGV contours imply that ifany directivity isasso-ciatedwiththisevent,thisistowardsSW.Thesequenceinstudy,islocatedintheeastwardextensionoftheAgiosEfstratios Fault Zone, which was the locus of the February 19, 1968 Mw7.2 earthquake, which produced surfaceruptureontheislanditself.TheAgiosEfstratiosFaultZoneterminatesoffshore,southofLemnosIsland.The2013sequence is not connected, in a strict sense, with the Agios Efstratios Fault. It indicates rupture of another fault,approximatelyparalleltoit,andthecharacteristicsofthesequenceareinaccordancewiththeregionaltectonicsandpreviousknowledge.TheCoulombstresschangeassociatedwiththemainshock,isalsoexaminedtoevaluateany signi fi cant enhance of stresses along the Agios Efstratios Fault Zone.© 2013 Elsevier B.V. All rights reserved. 1. Introduction It is generally accepted that the North Aegean Basin (NAB) in theNorthern Aegean Sea (Fig. 1) has tectonically developed after the colli-sion of Arabia with Eurasia in late Miocene and the subsequent west-ward escape of Anatolia relative to Eurasia, during the early Pliocene(e.g. Angelier, 1979; Armijo et al., 1999; Jackson, 1994; LePichon et al.,1995; McKenzie, 1978; Papazachos et al., 2000; Taymaz et al., 1991;Tranos, 2009). This westward motion of Anatolia is facilitated by thestrike-slip faulting along the North Anatolian Fault Zone (NAFZ) andthe East Anatolian Fault Zone (EAFZ). It is further facilitated by the fast(~20 mm/yrrelativetoEurasia)south-westwardretreatoftheHellenicsubduction zone (e.g. Armijo et al., 2003; Kiratzi, 2002; Kreemer et al.,2004). As far as dynamics are concerned, Jolivet et al. (2013) suggest that theextrusionof Anatoliaand theAegeanextension arepartlydriv-en from below(asthenospheric fl ow) and partly from above (extrusionof a lid of rigid crust). The opening and deepening of the North AegeanBasinincreasetowardssouthwest,from20to40 km,andfrom950 mto1650 m, respectively (Papanikolaou et al., 2006). The most prominentstructures within the North Aegean Basin are the Sporades Basin in itswestern part, the Athos Basin in its central part and the Lemnos Basinand Saros Basin in its eastern part (Karabulut et al., 2006; Koukouvelasand Aydin, 2002). An approximately 160-km long NE – SW trendingfault, that bounds the southern margin of the NAB, and extends fromtheSporadesBasintoLemnosBasin,canbealsoseeninthebathymetry,which by Papanikolaou et al. (2006) is considered as the possible con-tinuation of the NAFZ into the Aegean Sea.South of the North Aegean Basin, another topographic depression isprevailing, the Skyros – Edremit Basin, which opens and deepens withthe same polarity as the North Aegean Basin (e.g. towards southwest).The focus of our study is the region eastwards of the Skyros – EdremitBasin, which is less well studied compared to the North Aegean Basin,at least from the seismological point of view, simply because no strongevent occurred there during the modern instrumental period. In thiscontext, we willstudy the8January 2013Mw5.8earthquakesequence,whichis located~25 kmSE of Lemnos Island aswell. Mostimportantlythe sequence is located in a broader region which also includes theAgiosEfstratios Fault Zone(Basili et al., 2013),a segmentof whichrup-tured during the 19 February 1968 earthquake (Kiratzi et al., 1991;Papazachos and Papazachou, 2002; Pavlides and Tranos, 1991). Thus,this sequence is important because it provides evidence for the easternsegments of this major deformation zone (Fig. 2).The 2013 sequence was well recorded by the stations of the Greekand TurkishSeismological Networks (see Fig. 1), whose broad band ve-locityrecordsareusedhere.Thesequenceincludedonlythreeeventsof magnitudeMw ≥ 4.0and greater. Althoughnoseriousdamage wasre-ported, the mainshock was strongly felt in wide parts of the Greek and Tectonophysics xxx (2013) xxx – xxx ⁎  Corresponding author. Tel.: +30 2310 998486. E-mail address: (A.A. Kiratzi). TECTO-126046; No of Pages 9 0040-1951/$  –  see front matter © 2013 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Tectonophysics  journal homepage: Please cite this article as: Kiratzi, A.A., Svigkas, N., A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, East Aegean Sea),Tectonophysics (2013),  Turkishterritories.Peakgroundaccelerationvalues,fromthemainshock,recorded at Bozcaada station, located at the city of Çanakkale in Turkey,was 33 gal in NS direction, 18 gal in EW direction and 10 gal in up-down direction (preliminary report, at lastaccessed Sept 16, 2013).Inthesectionsthatfollowwewillpresenta)thespatialevolutionof the sequence using HypoDD relocated epicentres; b) the focal mecha-nisms of the stronger events using waveform modelling; c) the slipmodel of the mainshock; and d) the Coulomb static stress changes.The results are discussed in the context of the general tectonic setting. Fig.1. Signi fi cantfeaturesofthegeneralgeodynamicsettingoftheNorthAegeanSeaarea.Thefocalmechanismsofthestrongest(Mw  N  6.0)events,arealsoplotted(KiratziandLouvari,2003; Kiratzi, 2013 unpublished data), together with the locations of the broadband stations (triangles) whose waveforms were used here. The location and the focal mechanism of the20130108 event studied here are also shown. Fig. 2.  Signi fi cant faults along the broader region, from the database of  Basili et al. (2013). The star denotes the location of the 2013 sequence and the close active structure of the AgiosEfstratios Fault Zone (discussed in the text) is shown in bold.2  A.A. Kiratzi, N. Svigkas / Tectonophysics xxx (2013) xxx –  xxx Please cite this article as: Kiratzi, A.A., Svigkas, N., A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, East Aegean Sea),Tectonophysics (2013),  2. Relocation of epicentres — focal mechanisms Digital broadband velocity records were available a) from theHellenic Uni fi ed Seismological Network (HUSNET), operated jointly bythe Geodynamic Institute of the National Observatory of Athens, theDept of Geophysics-Aristotle University of Thessaloniki, the PatrasUniversity and the Kapodistrian University of Athens, in Greece andb) from the Kandilli Observatory and Earthquake Research Institute(KOERI) in Turkey. The waveforms and the phase times from thesestations were our dataset for the subsequent sections.  2.1. Relocation approach and data used We selected 220 events which were well recorded by at least threestations for one month period following the mainshock, and we usedHYPOINVERSE(Klein,2002)to obtaininitial locations,withRMSvalues less than 0.7 s. From the initial set of events, 195 ful fi lled the criteria of stronglinkedeventsinordertoberelocatedusingthedoubledifferencerelocation algorithm, HypoDD (Waldhauser and Ellsworth, 2000). HypoDD is a relative earthquake location method (e.g. Got et al.,1994)whichminimizestheresidualdifferencesofneighbouringevents.For the procedure, the conjugate gradients method (LSQR) (Paige andSaunders, 1982) was adopted. We used Vp/Vs ratio of 1.77 which wecalculated from our database using the classic Wadati technique. Wewere able to improve earthquake locations for the one-month-periodthat the earthquake sequence mainly operated. The RMS values corre-sponding to the 195 relocated aftershocks are less than 0.4 s and theaverage value is of the order of 0.26 s. Our relocated  fi nal aftershockcatalogue is complete for Mw ≥ 2.1 and from the Gutenberg – Richtercurve the b-value was calculated to be b = 0.63 and the a-value =3.48 (plot not shown here).  2.2. Method of moment tensor inversion Toobtainthe moment tensorsolutionsfor thestrongest events of the 2013 sequence we used regionally recorded broad-band wave-forms. The quality (good signal-to-noise ratio) of the available dataallowed the computation of 16 earthquake focal mechanisms andsource depths (e.g. Table 1). Two inversion codes were used. TheTDMT_INV time-domain inversion code (e.g. Dreger, 2002; Kiratzi,2013 and references therein for details on the approach) was usedto calculate the moment tensor of the mainshock and of the stron-gest aftershock (no. 11 in Table 1) and the ISOLA inversion code(Sokos and Zahradnik, 2008) for the smaller magnitude aftershocks. In both cases we solve for a deviatoric moment tensor solution.Green'sfunctionswerecalculatedusingthefrequency – wavenumberintegration code (FKRPROG) developed by Saikia (1994) in the for-mer approach and the wavenumber method by Bouchon (1981) inthe latter. The velocity model of  Novotny et al. (2001) was used inboth cases, which has proven very suf  fi cient in explaining the low-frequency content along many paths in the Aegean Sea and the sur-roundings (Kiratzi, 2013 and references therein). The velocity wave- formswere correctedfortheinstrumentresponse, bandpass fi ltered,re-sampled at 1 Hz and integrated to displacement. Usually we  fi lterthe data and the synthetic Green's functions between 0.05 and0.08 Hz for the strongest events and between 0.05 and 0.10 Hz forthe smaller magnitude events, but other  fi lters are also tested. TheISOLA code is a Matlab GUI written in FORTRAN, and the inversionmethod is mainly based on Kikuchi and Kanamori (1991). Duringtheapplication,thepositionsoftheeventswere fi xedwhile calculat-ing the Green's functions and the depth was allowed to  fl uctuate inorder to de fi ne the best solution, which provided the best  fi t overall depths and times. The code is  fl exible and has the ability to ana-lyse both regional and local events; examples of ISOLA applicationscan be found in Zahradnik et al. (2005), Fojtíková et al. (2010), Sokos et al. (2012), Benetatos et al. (2012).  2.3. Application results From the spatial distribution of the epicentres (Fig. 3a) we observethat the epicentres have a distinct N60°E alignment, and that theyare con fi ned within the 100 and 200 km isodepths. The bathymetryshows that the seabed has a smooth to gently undulating morphologywith gradients ranging from 0.1° to 1.8° (Yalt ı rak et al., 2012). Thealong strike dimension of the activated zone is no more than ~14 km,anditswidthis~7 km,inaccordancetowhatisexpectedfromthemag-nitude of the largest event of the sequence (Mw5.8). Most of the after-shock focal mechanisms show right-lateral strike-slip faulting (Fig. 3ainset),as is thecaseof themainshock (no. 1).Based on the distributionof the aftershocks and the focal mechanisms, the ENE – WSW trendingplanesarethefaultplanes,whichinmostofthecasesareperpendicularto the local trend of the bathymetry lines (Fig. 3a inset).A number of aftershocks are located off the main aftershock cloud.This observation is also depicted in the cross-section perpendicular tothe strike of the fault rupture (Fig. 3b). The best constrained focaldepths (from waveform modelling) indicate that the sequence is mainly  Table 1 Source parameters of the mainshock and of the strongest aftershocks of the Lemnos 2013 sequence obtained by the inversion of broad band waveforms from regional stations (seeSection 2).No. Date Time hh:mm:ss Lat° Long° Depth Mw Nodal plane 1 Nodal plane 2 P axis T axisyr/mm/dd km Strike° Dip° Rake° Strike° Dip° Rake° az° pl° az° pl°1 20130108 14:16:09.14 39.641 25.611 11 5.8 241 86 175 331 85 4 286 1 196 62 20130108 15:38:38.20 39.652 25.657 9 3.2 234 82 175 325 85 8 99 2 190 93 20130108 20:46:22.00 39.637 25.630 17 3.2 293 48  − 8 28 84  − 138 259 33 153 244 20130108 21:12:31.20 39.637 25.589 7 3.2 62 71 180 152 89 19 285 13 19 145 20130109 06:37:10.90 39.640 25.643 7 3.5 285 54  − 75 80 39  − 110 243 75 4 86 20130109 15:41:33.15 39.660 25.663 9 4.4 62 79 165 155 75 11 109 3 18 197 20130109 17:28:12.26 39.651 25.654 7 3.5 232 69  − 133 121 47  − 29 97 47 352 138 20130110 05:49:59.03 39.625 25.586 4 3.8 56 50  − 178 325 88  − 40 274 26 19 289 20130111 00:30:21.15 39.621 25.572 5 4.2 71 54  − 173 337 84  − 36 288 29 29 2010 20130111 15:07:31.98 39.656 25.667 3 3.9 31 11 170 131 88 79 231 42 30 4611 20130113 08:55:15.49 39.624 25.584 8 4.5 246 75 174 338 84 15 111 6 203 1512 20130113 17:54:32.72 39.616 25.676 3 3.8 62 26  − 161 315 82  − 65 251 47 24 3213 20130119 15:19:05.70 39.683 25.496 3 3.5 78 83 168 169 78 7 124 4 33 1314 20130120 01:13:31.00 39.647 25.642 3 3.2 47 49  − 171 311 83  − 41 260 33 6 2215 20130130 04:01:03.00 39.665 25.627 5 3.5 68 76  − 162 334 73  − 15 291 22 201 216 20130130 22:30:01.00 39.637 25.636 13 3.7 88 79  − 174 357 84  − 11 312 12 43 43  A.A. Kiratzi, N. Svigkas / Tectonophysics xxx (2013) xxx –  xxx Please cite this article as: Kiratzi, A.A., Svigkas, N., A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, East Aegean Sea),Tectonophysics (2013),  con fi ned in the crust (h  b  ~15 km) and is operating in the approximatedepthrangefrom3to25 km.Themainaftershockcloud(Fig.3a)isclear-lydepictedinthe~7 kmwidthnarrowverticalzoneofthecross-section.Regarding the evolution of the sequence in time, we observed thatthe harsh seismic expression took place during the  fi rst 4 days(Fig. 4). The 63% of the sequence had already been expressed by the Fig.3. a)DistributionoftheHypoDDrelocatedepicentres,whichclearlydepicttheENE – WSWtrendingplaneasthefaultplane(seetheinsetforfocalmechanismsofthestrongestevents,Table1,obtained bywaveformmodelling).Noticealsothatthesequenceisboundingthe100and 200 kmisodepths; b) cross-sectionperpendicular tothestrike ofthefaultrupture.Thebestconstrainedfocaldepths(fromwaveformmodelling)arecon fi nedinthecrust(h  b  ~15 km),howeverthesequenceisoperatinginthedepthrange3to25 km.Themainaftershockcloud is clearly depicted in the narrow vertical zone of the cross-section. Fig.4. Evolutionofthesequenceintime.Noticethat~63%oftheonemonthaftershocks(195intotal)havealreadyoccurredduringthe fi rst4 days(e.g.by2013.03).BytheendofJanuary2013 the seismicity rates dropped to background levels.4  A.A. Kiratzi, N. Svigkas / Tectonophysics xxx (2013) xxx –  xxx Please cite this article as: Kiratzi, A.A., Svigkas, N., A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, East Aegean Sea),Tectonophysics (2013),  12th of January (e.g. 2013.03 in Fig. 4). By the end of January 2013 theseismicity rates have dropped to background levels. 3. Slip model  3.1. Method used The slip distribution onto the fault plane for the Mw5.8, January 8,2013 earthquake was investigated following the  fi nite-fault inversiondescribed in Dreger and Kaverina (2000) and Kaverina et al. (2002). A number of applications in the Aegean area have shown the robustnessof the method (e.g. Kiratzi, 2011 and references therein). In general,themathematicalformulationofslipinversionconsidersanearthquakeruptureasthespatialandtemporaldistributionofdisplacementdiscon-tinuity  Δ u (  x , t  ) across the fault planes,  Σ  . We consider only shear slip,whose direction is perpendicular to the local normal vector of faultplane,  v (  x ), so that everywhere Δ u (  x ,  t  )  v (  x ) = 0. Thus, in an elasticmedium, the displacement at (  x  ,  t  ) due to a slip distribution Δ u ( ξ ,  τ  )has as follows (Aki and Richards, 2002; eq. 10.1): u i  x ; t  ð Þ ¼ Z  ∞ − ∞ d τ  Z Z  Σ Δ u  j  ξ ; τ  ð Þ   C   jkpq G ip ; q  x ; t  ; ξ ; τ  ð Þ v κ   ξ ð Þ d ∑ ξ ð Þ where  C   jkpq  is the elastic constant,  G ip  is a Green tensor function thatis i thcomponentofdisplacementat(  x  , t  )duetoanimpulsiveforceat( ξ ,  τ  ), in  p th direction, and  q  represents the derivative in ( ξ q ) direc-tion. The summation convention is used for index  j ,  k ,  p ,  q . Therefore,theconstituentsof theinversionproblemare:data u i , slipmodel Δ u  j ,and Green's functions  G ip .Our data vector consistsof the broad band three component velocitywaveforms from the regional stations operating in Greece and thewestern coastal regions of Turkey. Prior to the inversion, the waveformsarecorrectedfortheinstrumentresponse, fi lteredusingaband-pass4thorder  fi lter applied twice, and integrated to displacement. Model errorsare mainly introduced by the inaccuracy of the Green's functions, as isthecaseinallsimilarinversions.Themagnitudeofsucherrorsisapprox-imately proportional to the amplitude of the calculated synthetic data.We model the fault as a rectangular plane, of length, L and width, wand we divide it in a number of sub-faults along strike and along dip.Each sub-fault is allowed to slip only once, when the rupture front ar-rives.Tostabilizetheinversion,slippositivityandsmoothingconstraintsare applied, following the suggestions in Dreger and Kaverina (2000).  3.2. Distribution of slip onto the fault plane For the January 8, 2013 mainshock we adopt the orientation of the causative planar fault, as it was derived in the previous section(e.g. strike = 241°; dip = 86º; rake = 175°), and we solve for theheterogeneous slip distribution onto the fault plane. The  fi nite faultmodel has dimensions of 20 × 20 km 2 discretized into 1 × 1 km 2 sub-faults. The dimensions are large enough to avoid an imposed con fi ne-ment of the slip. The rise time,  τ  , was given a constant value of 0.5 s(Somerville et al., 1999), using the value for scalar moment of   Μ 0  =5.4 × 10 24 dyn cm,ascomputedfromthetime-domainmomenttensorinversion. The rupture velocity,  V  r  , was considered constant (2.6 km/s)and equal to 80% of the shear-wave velocity at the source region.The distribution of slip onto the fault plane (Fig. 5) shows a single-patch, of approximate dimensions 7 × 8 km 2 along strike and alongdip, respectively. The smaller patch observed in the SW is less well re-solved, in the sense that it was not stable in the trial inversions. If wechoose to take this patch into account, then the dimensions of the slippatch increase to 11 × 8 km 2 . The coordinate system in Fig. 5 is thelocal coordinate system of the fault plane, and the absolute depth of the hypocentre, with local coordinates (0,0) is 11 km (Table 1). Tothis end, slip is con fi ned in the depth range 5 to 13 km. The peak slipis observed ~2.5 km off the hypocentre, towards the SW (e.g. towardsSkyrosIsland).Thisslipmodelproducedsatisfactoryvariancereduction(78%). The average slip is ~30 cm (in accordance with Somerville et al.,1999) and the peak slip, in the centre of the slip patch, is ~130 cm.The total seismic moment obtained from the slip model is  Μ 0  =6.4 × 10 24 dyn cm, approximately 20% larger than the scalar momentobtained from the moment tensor inversion.  3.3. Prediction of near source ground motions (ShakeMap) Herewe solvetheforward problem,todeterminenear-faultgroundmotionsusingthepreviouslyobtainedslipmodel.Todoso,weperformforward calculations of the two horizontal-components of velocity atphantom stations, over a grid extending +0.5° around the epicentre,both in latitude and longitude. Thus, eight hundred synthetic velocityrecords are determined, two for each grid point. Synthetics are lowpass  fi ltered with the high corner cut-off at 2.5 Hz and the peak valuesof each component are depicted. Instead of mapping median values of the two components, we chose to map peak values, because the peakvalueswere systematicallylowerintheN – S component. We did exam-ine the pattern of the median values, as well, both the arithmetic andgeometric mean, and the overall pattern did not change signi fi cantlyfrom the one presented here. The map of  Fig. 6 shows the contours of the simulated PGV's (in cm/s). The values comply with those expectedfrom empirical relations (as for example Skarlatoudis et al., 2003,2007) and the contours imply directional effects towards west, if any. Fig.5. Slipmodelforthe8January2013Mw5.8mainshockobtainedfromtheinversionof broadband waveforms,for the ENE – WSWtrending plane.Slip isconcentratedin a singlepatchofapproximatedimensions7 × 8 km 2 .Theaverageslipandthepeaksliparecalcu-lated equal to 30 and 130 cm, respectively. The asterisk denotes the hypocentre location,which lies at an absolute depth of 11 km.5  A.A. Kiratzi, N. Svigkas / Tectonophysics xxx (2013) xxx –  xxx Please cite this article as: Kiratzi, A.A., Svigkas, N., A study of the 8 January 2013 Mw5.8 earthquake sequence (Lemnos Island, East Aegean Sea),Tectonophysics (2013),
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks