Small-scale spatial variation of the stress field in the ...geophysics.geo.auth.gr/new_web_site_2007/download_files/costas_CV/...Aegean area: Results from the seismotectonic study
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
2006) 249–267www.elsevier.com/locate/tecto
Tectonophysics 417 (
Small-scale spatial variation of the stress field in the back-arcAegean area: Results from the seismotectonic study of the
broader area of Mygdonia basin (N. Greece)
D.A. Vamvakaris ⁎, C.B. Papazachos, E.E. Karagianni,E.M. Scordilis, P.M. Hatzidimitriou
Geophysical Laboratory, School of Geology, Aristotle University of Thessaloniki, GR-54124, Thessaloniki, Greece
Received 25 February 2005; received in revised form 25 January 2006; accepted 27 January 2006Available online 24 March 2006
Abstract
In the present work a detailed seismotectonic study of the broader area of the Mygdonia basin (N. Greece) is performed. Digitaldata for earthquakes which occurred in the broader Mygdonia basin and were recorded by the permanent telemetric network of theGeophysical Laboratory of the Aristotle University of Thessaloniki during the period 1989–1999 were collected and fault planesolutions for 50 earthquakes which occurred in the study area were calculated with a modified first motions approach whichincorporates amplitude and radiation pattern information. Fault plane solutions for the 3 main shocks of Volvi (23/05/78, MW=5.8and 20/06/78, MW=6.5) and Arnaia (04/05/95, MW=5.8) events and the 1978 aftershock sequence were additionally used.Moreover, data from two local networks established in the Mygdonia basin were also incorporated in the final dataset.
250 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
1. Introduction
Fault plane solutions constitute an importantelement in seismological research and a basic toolfor the stress field study. Several researchers (Etch-ecopar et al., 1981; Angelier et al., 1981, 1982; Armijoet al., 1982; Gephart and Forsyth, 1984; Gephart,1990a,b; Papazachos and Kiratzi, 1992; Loohuis andvan Eck, 1996) have developed methods that canemploy fault plane solutions to calculate the stresstensor describing the local stress field in a study area.In the present work, the stress tensor inversion methodof Gephart and Forsyth (1984) was used, using faultplane solutions for the Mygdonia basin area. A largenumber of publications which were based on thismethod, using earthquake sequences, have beenproposed for different seismotectonic regions, suchas Neri and Wyss (1993) (Tyrrhenian region — Italy);Horiuchi et al. (1995) (Nagano area — Japan);Caccamo et al. (1996) (western Sicily area — S.Italy); Cocina et al. (1997, 1998) (Mt. Etna — S.Italy); Eva and Solarino (1998) (western Alpine arc);Kiratzi (1999) (Kozani–Grevena area — C. Greece);
Fig. 1. Map of known earthquakes with MW≥3.0 which occurred in the broatill 2002. Black circles denote the epicenters of earthquakes with Mw≥Servomacedonian massif (shadowed area) and the North Aegean Trough. Th
Frepoli and Amato (2000) (Italy); Hinzen (2003)(Northern Rhine area, Central Europe).
The area examined in the present study is theMygdoniabasin, which is located in Northern Greece (CentralMacedonia). Its pre-alpine and alpine basement belongsto the Servomacedonian massif and the Circum RhodopeBelt Thrust System. Above this basement extensive NW–SE and E–W trending continental-type basins and grabenshave been filled with Neogene and Quaternary sediments.They were developed by a Miocene to present extensivebrittle extensional deformation that is mainly related tohigh-angle normal faults (Pavlidis and Kilias, 1987).Among these basins the E–W trending Mygdonia grabenand its precedingNW–SEPre-Mygdonia basin is located atthe center of the Servomacedonian massif (Fig. 1). Thebasin has a characteristic S-shape with its edges trending inaNW–SE direction, while the central part is approximatelyE–W oriented. Moreover, the active tectonics, thedeformation and seismicity of the broader area ofMygdonia basin have been studied in detail using bothseismological and neotectonic observations (Scordilis,1985; Pavlides, 1996; Martinod et al., 1997; Papazachoset al., 2001; Vamvakaris et al., 2003).
der area of central–northern Greece from the historical times (550 BC)6.5. A significant concentration of epicenters is identified in thee dashed line delineates the limits of the study area.
251D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
In the present work, seismological data from differentsources were used in order to create a complete databaseof focal mechanisms for earthquakes which occurred inthe broader Mygdonia basin area. Additional fault planesolutions for the last decade are computed using localand regional recordings of the permanent seismologicalnetwork of the area. This dataset is employed for thedetermination of the main seismotectonic characteristicsand the spatial variation of the stress field in the studyarea.
2. Seismotectonic setting
The seismicity of the Servomacedonian massif whichextends from the FYROM-Bulgaria border up to theNorth Aegean Trough is the highest in N. Greece withevents up to MW=7.6 (Scordilis, 1985). A large portionof this seismic activity is located in the Mygdonia basinand its continuation towards the Ierissos bay area, whichis one of the most active seismogenic regions in theback-arc Aegean area. This central section is the mostseismically active and has received a lot of scientificattention since the occurrence of the MW=6.5 main-shock of June 20, 1978, which was the first major eventin Greece with significant impact on a modern urbancenter such as the city of Thessaloniki (Papazachos etal., 1979a,b; Pavlidis and Kilias, 1987; Pavlides et al.,1988; Hatzidimitriou et al., 1991). This intense seismicactivity is directly related to the extensional deformationwhich is in effect from Miocene to present. In Fig. 1 allthe known earthquakes with M≥3.0 which occurredfrom the historical times (550 BC) up to the present arepresented for the Central–Northern part of Greece.Except from the SE part of the area where the highseismicity level is associated with the N. Aegean Troughwhich is the continuation of the North Anatolia Fault inthe Aegean, seismic activity of the continental central–northern Greece is mostly located in the Servomacedo-nian massive (shadowed area) with a significantconcentration in the broader Mygdonia basin. Eightstrong earthquakes with M≥6.5 have occurred in theServomacedonian massif during the 20th century, thestrongest in Cresna (SW Bulgaria,M=7.6, 1904) and inAgion Oros (M=7.4, 1906). In the Mygdonia basin,seven strong events (6.0≤M≤7.0) are known from thehistorical times to the present (Fig. 1, data fromPapazachos and Papazachou (1997)). During the 20thcentury two destructive earthquakes occurred in thebroader area of Thessaloniki along the E–W Mygdoniagraben (Assiros, 5/7/1902, MW=6.5 and Stivos, 20/6/1978, MW=6.5). The 1978 Thessaloniki (Stivos)earthquake sequence comprises of the larger foreshock
on May 23 (MW=5.3), the mainshock on June 20(MW=6.5) and the larger aftershock on July 4(MW=5.0). On May 4, 1995 a strong event with magni-tude MW=5.8 occurred in the SE part of the basin(Arnaia). Most recently, intense seismic activity close tothe city of Thessaloniki occurred in the summer of 1999with the largest event ofM=3.7 on June 29 (Papazachoset al., 2000).
The faulting pattern of the broader area is compli-cated and includes NE–SW, E–Wand SE–NW strikingfaults, where the E–W trending faults are the mostactive ones related with the current seismic activity. Themajor faults are presented in Fig. 2 and summarisedbelow.
2.1. E–W striking faults
These faults with a general E–W direction present alocal variability from ENE–WSW to ESE–WNW,suggesting integration with pre-existing faulting zones.The most important faulting zone in the CentralMacedonia area is the one of Volvi, which determinesthe south part of the Mygdonia graben. Severalhistorical earthquakes have occurred along this zone(Papazachos and Papazachou, 1997) and many of itsbranches are clearly related with the seismic sequence of1978 (Papazachos et al., 1979a,b). Many of the knownactive faults can be included in the 70 km long Volvifault zone, such as (see Fig. 2) the Asvestochori–Polichni fault (A–P), Thessaloniki–Gerakarou fault(T–G), Gerakarou–Nikomidino–Stivos–Peristeronasfault (G–N–S–P) and Loutra Volvis–Nea Appoloniafault (V–NA) (Tranos et al., 2003). Finally, theAnthemountas faulting zone (Anth), located in thesouthern part of the study area, has an E–W to ESE–WNW strike, is dipping to the north and has a length ofabout 40 km (Mountrakis et al., 1996, 1997).
2.2. SE–NW striking faults
Faults with SE–NW direction have a significantcontribution to the topography of the large neotectonicbasins which are found in the Central Macedonia area (i.e.Langadas basin). Specifically, the Assiros–Analipsi–Scholari faulting zone (As–An–Sc) is a zone of totallength of 30 km on the south-eastern boundaries ofLangadas graben. A brunch of this zone was activatedduring the occurrence of the 1978 Thessaloniki earthquake(Papazachos et al., 1979a; Mountrakis et al., 1983). TheLiti–Lagina–Ag. Vassilios fault (L–L–AgV) is a SE–NWstriking fault continuing for 20 km towards the east, as anextension of the WSW striking G–N–S–P fault.
252 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
2.3. NE–SW striking faults
NE–SW striking faults are not clearly recognized inthe Mygdonia basin. Topography and geomorphologi-cal features (Tranos et al., 2004) give a strong indicationfor the existence of faults having this direction. How-ever, the Neogene–Quarternary sedimentary depositscover such faults so that they have no clear surfaceexpression.
3. Applied method
3.1. Estimation of fault plane solutions
Focal mechanisms can provide useful informationabout the structure and kinematics of faults, describing atthe same time the dominant crustal stress field in whichan earthquake occurs. The FPFIT program (Reasenbergand Oppenheimer, 1985) is one of the most often usedapproaches to calculate the fault plane solutions, usingthe P-wave first motion polarities from short-period data.A modified version of this program (Vamvakaris et al.,
Fig. 2. Epicenters for the 182 earthquakes used in the present work and main fet al. (2003).
2004) was used in order to obtain improved results forthe calculation of the fault plane solutions. In compar-ison to the original version of FPFIT the modified ap-proach takes also into account the radiation pattern of SVand SH waves. For each earthquake the horizontal andvertical components of each station were used and thefirst arrivals of P and S waves were picked. Using themaximum peak-to-peak amplitude of P and S waves theratio Pmax / (SN
2max+SE
2max)
1 / 2 was estimated, whereSNmax and SEmax are the maximum amplitudes of thetwo horizontal components (N–S, E–W) for the Swavesand Pmax is the maximum amplitude of the vertical onefor the P-waves. This ratio for the observed data, as wellas the corresponding ratio Prad / (SH
2rad+SV
2rad)
1 / 2 ofthe synthetic data was used as a weight for thedetermination of the observed and theoretical P-wavepolarities, respectively (Vamvakaris et al., 2004). Testswith synthetic and real data showed that this modifiedapproach results in the minimization of the derivation ofmultiple solutions, as well as of the uncertainties in thelocation of P and T axes of the determined fault planesolutions, similar to the other methods which employ
aulting zones (explanation in text) in the study area according to Tranos
253D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
amplitude information (e.g. Hardebeck and Shearer,2003). We should point out that in those cases that nohorizontal components were available (only verticalseismometer available, less that 1 /3 of stations in ourcase) since theP /S ratio can not be estimated, the appliedsoftware was adapted in order to use only the availablepolarities of the first arrivals.
3.2. Stress tensor inversion
Several researchers have worked on the problem ofthe determination of a stress tensor related with a singlefault using field observations. Based on the work of Bott(1959), McKenzie (1969) and Angelier et al. (1982)studied the general case where a random discontinuitycould act as a fault, whose movement depends on asimple stress tensor. The principal constraint on thestress tensor in these analyses is that the shear stressdirection on any fault plane must coincide with theobserved slip direction in both orientation and sign. Inthe simplest case it is assumed that all slip events reflecta common stress field. Gephart (1990a,b) reviewed theprinciples of this analysis and described a straightfor-ward procedure for estimating stresses from groups offault geometries, defined as fault planes with associatedslip directions.
In the present work the method of Gephart andForsyth (1984) was applied for inverting earthquakefocal mechanism data to obtain the local stress field. Inthis method the orientation of fault planes and slipdirections provided by a large population of earthquakefocal mechanisms can be used to determine best fitregional principal stress directions and the parameterR=(σ2−σ1) / (σ3−σ1), which specifies the magnitudeof the intermediate σ2 compressive stress direction,relative to maximum σ1 and minimum σ3 compressivestress directions, under the assumption of uniform stressin the source region. The analysis allows for the pos-sibility that the failure occurs on pre-existing zones ofweakness of any orientation. The computer programFMSI (Gephart, 1990b) inverts observations of slipdirections on fault planes of known orientation in orderto determine the best-fitting four-parameter stresstensor, defined by the three principal stress directionsand the parameter R, as well the associated uncertainty.The definition of the best-fitting model and its iden-tification are two important aspects of the method. Themisfit between fault-plane solution observations and aspecific stress model is defined as a rotation of thecombined fault plane/slip vector that achieves an orient-ation for which the observed and predicted slip direc-tions on the fault plane are aligned. The best model is the
one that minimizes the sum of these values for all data,which predicts faults geometries that most closely matchthe observed ones. A grid search of stress models ratherthan a linearization scheme is used and a realistic erroranalysis can be performed in order to establish con-fidence limits for the preferred regional stresses. Themethod can be used to investigate possible stress in-homogeneities during earthquake sequences or a small-scale spatial variation of the stress field.
To confirm the reliability of the results produced bythe use of the stress tensor inversion we also applied ano-ther method proposed by Papazachos and Kiratzi (1992)based on the calculation of a representative “averagefocal mechanism tensor”. According to this method, forthe determination of the stress field the average “focalmechanism” tensor is calculated. This tensor is definedby the released moment Mo and a tensor F which is afunction of the strike, ζ, dip, δ, and rake, λ of the cor-responding fault plane (Aki and Richards, 1980), asshown by the equation:
Mn ¼ Mn0 d F
nðn; d; kÞ
From this equation, an average “focal mechanismtensor”,F̄, representing the deformation pattern of thearea, is estimated by the relation proposed by Papa-zachos and Kiratzi (1992),
F̄ ¼
XN
n¼1
Fn
N
According to the method, the eigenvalues of F̄correspond to the average P, T and N (null) axes of thelocal stress tensor. Therefore, the method allows thedefinition of an “average” kinematic (P, T and N) axiswhich is assumed to be identical with the principal stressaxes. It is clear that in areas where pre-existing struc-tures and complicated faulting dominates, deviationsfrom this assumption will lead to a systematic differenceof the stress axes obtained by this method in comparisonwith the method of Gephart and Forsyth (1984), whichcan be efficiently applied in such environments. How-ever, the additional application of this method allows: a)to obtain an independent stress field estimation andperform a comparison with the results obtained by themethod of Gephart and Forsyth (1984) in order toindirectly assess if such a complicated geotectonic set-tings exists for the study area and, b) to allow thedetermination of a starting stress field estimate, which isnecessary for the parametric search of the Gephart andForsyth (1984) method.
254 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
4. Data used
The data used for this study come from theSeismological Station of the Aristotle University ofThessaloniki and contain information about earth-quakes recorded by the permanent telemetric networkof the Geophysical Lab consists of 17 stations es-tablished mainly in N. Greece. Most of these stations(13 out of 17) are equipped with three-componentshort-period seismometers and only 4 with single ver-tical short-period seismometers, using a sampling fre-quency of 50 Hz. More than 473,000 waveforms inSAC format were obtained for 14,856 earthquakeswhich occurred in the broader Greek area during theyears 1989–1999.
Using this source database 50 earthquakes with epi-centers bounded by 22.85° E–23.70° E and 40.44°N–40.80° N (corresponding to the Mygdonia basin)were selected. This selection was made for events withmagnitude MW≥3.0, which mainly exhibit a good sig-nal to noise ratio and for which we were able to obtainreliable fault plane solutions, presented in Table 1. Theaverage number of first motion polarities used for these50 events was 9 (see column “FM” of Table 1) and theaverage horizontal uncertainty in the epicenter location(ERH) was of the order of 0.6 km. Note that the use ofthe radiation amplitude information for the FPS deter-mination allows to obtain good quality fault planesolutions (Vamvakaris et al., 2004), even if the numberof polarities is not as large as usually required in stan-dard first motion regional studies. Furthermore, allstudied events are at the center of the monitoring per-manent network, hence the azimuthal coverage used forthe fault plane solution determination was very good forall examined events. The analysis of the data was per-formed using the Seismic Analysis Code-SAC (Tull,1987; Tapley and Tull, 1992; Goldstein and Minner,1995).
Existing seismological data were also used for thecalculation of the stress field and the determination ofthe final seismotectonic model for the whole area.Due to their importance, 3 strong earthquakes of 1978in Volvi with magnitude MW=5.8, 6.5 (Papazachosand Papazachou, 1997) and MW=5.3, respectively(HARVARD CMT), and one which occurred in Arnaiaat 1995 with MW=5.8 (Papazachos and Papazachou,1997) were added to our dataset. Parameters for theseimportant earthquakes that occurred in the study areaare also presented in Table 1.
Furthermore, we have also used the aftershocks ofthe 1978 seismic sequence. Since most of them hadsimilar characteristics, they were grouped in 24 sets of
events with common parameters an4d an averageepicenter at the center of each cluster (Soufleris et al.,1983). Also, our dataset was enriched with 63 smallmagnitude earthquakes (MLb3.0) which occurred in theMygdonia basin in the spring of 1984 and 1985 andwere recorded by a local network (Christodoulou, 1986;Hatzfeld et al., 1987). Finally, 41 earthquakes whichoccurred during the period July 2001–April 2002 wereused, as they were recorded by a local network esta-blished to observe the microseismic activity in this area(Paradeisopoulou et al., 2004). These events correspondto the seismic sequence of October 8, 2001 with amainshock with ML=4.5 (Table 1). The epicenters ofthese events are mostly located in a narrow area in thesouthern border of the Langadas Lake with an averageepicentral error of approximately 1.0 km. The epicentersof the 182 earthquakes finally used in this work arepresented in Fig. 2.
It is interesting to notice that the average error for thedip/strike/rake of the 50 fault plane solutions determinedin the present work is of the order of 7°, 10° and 11°,respectively, while the corresponding errors of theprincipal kinematic axes (P, T and N) is of the sameorder (∼10°). Although the fault plane solutions for the1978 aftershock sequence and the other local experi-ments (1984–85 and 2001–2002) were determinedusing a standard first motion approach, the dense tem-porary networks employed resulted in similar magnitude(∼10–15°) of the corresponding axes errors (Soufleris etal., 1983; Christodoulou, 1986; Hatzfeld et al., 1987;Paradeisopoulou et al., 2004). Finally for the 4 strongershocks, the corresponding fault plane solutions havebeen determined by waveform modeling, as well as byadditional seismological and field information, hencethe corresponding errors of the geometrical character-istics of the stress axes should be of the same order oreven smaller. Therefore, despite the fact that the faultplane solutions form a quite heterogeneous dataset, thecorresponding errors of the individual should beconsidered of the order of 10° or slightly larger.
5. Estimation of stress field
In order to estimate the stress field in the Mygdoniabasin by the use of seismological data, we haveseparated the region in smaller areas with commonseismotectonic characteristics and a relative homogene-ity of stresses, so that the results can be considered aslocally representative. This zonation was performedtaking into account the information provided by theneotectonic rupture zones (Tranos et al., 2003)which are possibly related with the stress field, the
258 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
homogeneity of the fault plane solutions and thedirection of the T-axis from fault plane solutions.Thus, 11 zones were defined for the study area, whichwere considered to exhibit a uniform tectonic settingand a similar behavior.
We should point out that the adopted zonationapproach resulted in a few zones with a relativelysmall number of events (i.e. zone 7b, 8 and 10), howeverthe available fault plane solutions, as well as theneotectonic setting (observed active faults, etc.) of theseareas shows a clear differentiation which does not allowto merge them with vicinal zones and consider them aspart of a larger zone (i.e. zone 7a+7b). This is laterfurther discussed for the stress field estimation and itscorrelation with neotectonic information. Furthermore,
the selected subdivision of the study area resulted insome zones having a width of the order of 10 km.Nevertheless, the high precision of the epicentral
259D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
locations (less than 1.5 km— see Table 2) suggests thatthe qualitative characteristics and the accuracy of theadopted subdivision is not affected by epicentrallocations, since even when examining neighboringzones (for example zones 7b and 8), it is easily seen(e.g. Fig. 3) that a very small number of events couldhave been misplaced in the wrong zone, given the verysmall epicentral uncertainties.
In Fig. 3 the focal mechanism for the 182 earth-quakes used in the present work, as well as the mainfracture zones described before are presented for the 11zones. Examination of this figure shows that the faultplane solutions show a similarity throughout the wholestudy area, as normal ruptures with a W–E direction aredominant. Similar results can be obtained whenconsidering the P and T axes of the focal mechanisms.Each one of these zones was examined individually anda representative stress field was estimated for everyzone. Both methods previously described were appliedfor the stress field determination for the 11, as well as forthe whole study area using all the examined events as acommon dataset. For the Gephart and Forsyth (1984)inversion method, the initial principal stress solution
Fig. 3. Separation of the study area in eleven (11) zones, based on the 182 f
was adopted from the Papazachos and Kiratzi (1992)results. For the parametric search a 30° search areaaround this solution was performed using a 5°-spacingpredefined grid, which means that 85 different primarystress axes were tested within an angle of 30° “around”the initial primary stress axis.
The final stress models determined by the GF methodfor the various zones correspond to values of R mainlybetween 0.3 and 0.7 for each zone (Table 3). In generalthe observed R value are slightly greater than 0.5, whichsuggests that σ2 is closer in magnitude to σ3 than σ1
(RN0.5). This pattern corresponds to zones whereexcept from the dominant extension (and compression)giving typical double couple fault plane solutions, aslight secondary principal extensional stress is alsofound, which suggests a small-scale variation of thedominant extension even within each zone. Anexception is found for zones 5, 7a where R=0.3 andzone 9 with an extreme value of R=0.1, suggesting thatσ2 is closer or almost equal in magnitude to σ1 hence auniaxial extension dominates in these zones.
The obtained results shows an average azimuthaldifference of 11.5°±7° for the principal extension axis
ault plane solutions for the earthquakes used and main faulting zones.
Table 3Azimuth and dip angle values for σ1 and σ3 principle stress directionsfor each one of the 11 zones of the study area, as they have beendetermined by the methods of Papazachos and Kiratzi (1992) andGephart and Forsyth (1984)
Mean values from both applied methods are shown in the last line ofthe table. The average minimum rotation misfit and values forparameter R are also presented for each zone.
260 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
(σ3) between the two methods, as is shown in Fig. 4,where the white arrows present the results of thePapazachos and Kiratzi method and the black onespresent the results of the stress tensor inversion ofGephart and Forsyth. All stress vectors are plotted in arepresentative location in each zone, depending on theepicenters of the earthquakes which where used for thespecific zone. In the same figure, the average stress axesfor the major active neotectonic faults provided by thestudy of kinematics (Mountrakis et al., 2003) are alsopresented with thick gray arrows.
The results for the maximum σ1 and minimum σ3
principal stress directions, provided by the use of bothmethods are also shown in Table 3, both for the wholearea of interest as well as individually for the 11 zones.Moreover, R values and values for the average minimumrotation misfit provided by the inversion method for thezones selected are also presented in Table 3. It should bepointed out that when considering the angular differencebetween the two methods, a mean difference of 18° isfound for both σ1 and σ3 stress axes. These differencesare obviously associated with the inhomogeneous stressdistribution and pre-existing ruptures within each zoneand are within the range of the resolution of themethods. However, the observed bias is slightly largerthan the expected data error (∼10°) of the kinematicaxes from the fault planes of our dataset, which suggeststhat although both methods provide relatively similarresults, the Gephart and Forsyth (1984) results should be
adopted, as the method can handle such inhomogeneousstress effects.
This suggestion is further supported when consider-ing information for the mean extensional stress axesfrom field observations of the most important neotec-tonic fault zones (Mountrakis et al., 2003) which ispresented in Table 4 and compares them with thecorresponding seismological information for the zoneswhere the corresponding neotectonic faults are present.A good match of the values for the maximum extensionstress axes, T, is found for all 6 zones for whichcorresponding neotectonic information was available,when considering the results of the Gephart and Forsyth(1984) method, as σ3 seems to be almost parallel to the Taxes provided by the fault kinematics, with an azimuthmisfit between the stress axes from the seismologicaland neotectonic data less than 7°. This misfit increasesto 12° when the corresponding results of Papazachosand Kiratzi method are used, which further verifies thesuperiority of the inversion process in identifying thecorrect stress field. On the other hand, it is interesting tonotice that this seismological–neotectonic stress fieldcoincidence suggests that the stress field derived usingthe available fault plane solutions is practically identicalto the stress field shown by the available neotectonicfield data, despite its inhomogeneous original dataset orthe specific zonation scheme adopted. This observationverifies that the zonation procedure followed in thepresent study was not based on an over-interpretation ofthe available data (fault plane solutions, active faults,neotectonic setting), given the specific data uncertainties(e.g. epicentral locations).
5.1. Spatial variation of the stress field
Fig. 4 summarizes the stress field estimates derivedfrom seismological data for the broader area ofMygdonia basin. In general, the stress field in thestudy area is extensional, with a subhorizontal minimumstress axis (σ3) showing a∼N–S (6°) direction, as shownby the mean extensional stress vectors at the bottom ofthe Fig. 4. High seismicity and independent seismic(active) crustal deformation estimates show that theextension is a continuous process, mainly in the centralpart of the basin, following the lowest topographicdepression between the two lakes with an extension rateup to 3.4 mm/yr (Papazachos et al., 2001).
For most zones (zones 1, 3, 5, 6, 7a, 9 and 10) thedominant extension exhibits a general N–S direction.An extension with a NNW–SSE (∼320°− 335°)direction is found for zones 2, 4 and 7b while a NE–SW (∼25°) extension is only identified in zone 8.
Fig. 4. Maximum extensional principal stress direction (σ3) for each one of the 11 zones. White arrows correspond to the results of the Papazachosand Kiratzi (1992) method, while with black arrows the ones of Gephart and Forsyth (1984) method are shown. Thick arrows denote the averageminimum principal stress direction for the whole data (bottom part of the figure). Average stress axes for the major active faults in representativelocations (thick gray arrows) provided by the study of the kinematics (Mountrakis et al., 2003) are also presented in the same figure.
261D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
Similar variations are found from neotectonic observa-tions. Specifically, for the eastern part (zone 3) thestresses provided by the seismological data and theextensional neotectonic stresses related with the ob-served V–NA faulting zone are identical. The same isobserved for the central part of the study area containingthe G–N–S–P fault located in zone 4 and in thenorthern part (zone 5) related to the As–An–Sc faultingzone, where the misfit between the average T–axes forthe neotectonic and seismological data is less than 5°.
In the western part, for the zones 7b and 9 theextension stresses using the two methods seems to bealmost parallel for the seismological data (differenceless than 5°). Neotectonic observations for the centraland eastern part of T–G faulting zone, shows that theaverage stress is in good agreement with the seismo-logical data (7° misfit). In zone 8, kinematics for the A–P fault provides similar stress axes information with theseismological data. Finally, a misfit of about 11° bet-ween the stresses provided by the seismological andneotectonic data is found for zone 10 where the Anth
fault is located. However, it should be pointed out thatseismological information was available only for theeastern section of the fault (Fig. 3) where the fault“bends” towards the south, suggesting a possible localvariation of the obtained stress field.
For the examined zones a significant small-scalespatial variation of the orientation of the mean exten-sional axis from NNW–SSE to NNE–SSW (320–30°)is independently observed from both seismological(fault plane solutions) and geological field observations(faults, neotectonic ruptures). These results suggest thatthe improved fault plane solutions determined in thepresent study, the applied zonation scheme, as well asthe stress tensor inversion method have allowed thereliable estimation of the local stress field in the studyarea.
5.2. Stress field inversion and fault determination
Fault plane solutions define both nodal planesdescribing the rupture (main and auxiliary) but it is
Table 4Values for azimuth and dip angle of the mean extension stress axes inthe Mygdonia basin from neotectonic data (Mountrakis et al., 2003)and extensional principal stress directions (σ3) from the present work(seismological data), for the corresponding zones where neotectonicfaults are found
262 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
not clear which one of them is the fault plane associatedwith the earthquake. Application of the stress-tensorinversion, using the method of Gephart and Forsyth(1984) for the seismological data allows to chose theplane corresponding to the minimum misfit rotationabout any axis of general orientation which is needed tomatch an observed fault plane/slip direction with oneconsistent with the given stress model. These faults areidentified by the inversion method to be the “ideal” faultplanes on which the corresponding earthquakesoccurred.
As previously described, the nodal plane corres-ponding to the minimum misfit is identified by themethod as the fault plane. However, it is clear that thisselection is arbitrary if both nodal planes exhibit similarmisfit values which are either small (both planes areacceptable) or very large (both could be considered as“incompatible” with the determined stress field). Takinginto account the average uncertainty of 10° related withthe provided fault plane solutions and their kinematicaxes, we can consider the difference of the misfit valuesof the two planes provided by the method as a qualitymeasure for the selection procedure. In order to obtainmore realistic and reliable results we have assumed thatif the misfit calculated for both fault planes (main andauxiliary) is relatively small (less than 2–3 times largerthan the average uncertainty, i.e. 25°) and theirdifference is less than the average uncertainty (10°),then both planes should be considered and included inthe results of this study, as they are practicallyindistinguishable with respect to their misfit using thedetermined stress field. For larger rotation misfit dif-ference values (larger than 10°), we took the traditional
approach and kept the plane which was most compatiblewith the determined stress field (smaller misfit). For thecases for which the rotation misfit was too large (morethan 25°) for both fault planes, then neither plane wereadopted and both planes were rejected from furtherprocessing, as completely “incompatible” with the de-termined stress field.
This modified selection procedure resulted in reject-ing about one third of the fault planes provided by thefault plane solution database, leaving approximately240 fault planes which were selected as candidates forthe active faults which generated the examined earth-quakes. This rejection practically means that for aboutfor 65% of the examined fault planes a single nodalplane was selected as the “true” fault plane. Further-more, 225 out of 240 planes selected (more than 93%)exhibit a misfit less than 15° (upper limit of typicalkinematic axis error) and only in 5 cases (out of 182)both planes of the calculated focal mechanism exhibiteda misfit greater than 25° and were discarded. Thisremoval of “incompatible” data (∼2.5%) is too small toaffect the dataset used and change the obtained stressfield results. Furthermore, it should be noted that about45% of the fault planes selected show a misfit less than5°, verifying the good quality of the results.
In Figs. 5 and 6 fault planes identified by the stressinversion method of Gephart and Forsyth (1984) arepresented as linear elements, for two of the examinedzones worked (zones 3 and 7a, respectively). In eachfigure the identified faults are also presented using tworosediagrams. The upper one corresponds to the clas-sical presentation of the direction of the fault planes,while the lower one adopts the Aki and Richards(1980) convention, which gives additional informationabout the dip direction of the fault plane. For instance,in zone 7a (Fig. 6) all NW–SE trending faults dip tothe south, while NE–SW trending faults dip to thenorth. Both rosediagrams describe the dominant orien-tation of the fault planes, as this is determined by theuse of focal mechanisms and the stress inversionmethod.
A joint view of the major neotectonic fracture zonesand the directions of the faults provided by the seis-mological data is presented in Fig. 7 for each one of the11 studying zones, using rosediagrams for the presen-tation of the selected fault plane data. In the same figure,the average extension stresses derived by the use of thetwo methods employed are also shown.
In most cases, the faults “proposed” by the inversionmethod are in agreement with the faults observed in thefield or with the expected directions of the faults, asthese are depicted by the seismicity distribution.
Fig. 5. Fault planes that the stress tensor inversion method identified as candidates for the “real” seismic faults, using the data for zone 3. Therosediagrams present in two alternate ways the fault distribution within each zone (see text for explanation).
263D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
Application of the proposed method for the availableseismological data, for the eastern part of study areashows a group of normal E–W trending faults in verygood agreement with the field observations. Themethod seems to choose correctly in zone 1 the normalE–W faults and the NW–SE ones related with the
Fig. 6. Same as Fig.
Arnaia earthquake (4/5/95, MW=5.8), in zone 2 theE–W normal faults with southern dip, the numerousWNW–ESE normal faults in zone 3 and the large normalfaults with E–W azimuth related with the 1978 seismicsequence (zone 4). In the western part of the study area,the known dominant group of NW–SE faults is also
5 for zone 7a.
Fig. 8. Rosediagrams for the central and eastern part of theThessaloniki–Gerakarou faulting zone from neotectonic observations(gray color, Tranos et al., 2003) and from the fault planes availablefrom seismological data for the corresponding zones 7b and 9 (blackcolor).
Fig. 7. Joint presentation of the main neotectonic settings and rosediagrams of fault planes identified by the analysis of the seismological data for eachzone. In the same figure the average principal extensional stresses of Fig. 4 are also presented.
264 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
found by the seismological data, while another branchwith NE–SW faults is also found. More specifically, forzones 6, 7a and 7b faults with both NE–SWand NW–SEazimuth are identified. The WNW–ESE trending faultsmatch with the dominant rupture zones found in thisarea, while the second group of faults is almostperpendicular to the first one. These branches of NE–SW direction are very interesting because several small-scale fractures of the same azimuth are detected fromfield observations in this area (Tranos et al., 2004).Finally, there is a very good concurrence of the results ofthe inversion and the geological observations for zones 9and 10, where the WNW–ESE to E–WandWNW–ESEfaults identified by seismological data are also observedin the field, respectively.
The use of rosediagrams allows the derivation ofimportant information regarding the dominant faultingzones. In Fig. 8 the directions of faults provided by thestress inversion of seismological data and the rupturesdistribution determined by field measurements (Tranoset al., 2003, 2004) are presented for two zones. The graycolor rosediagram depicts the results of Tranos et al.
265D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
(2003) for the direction of 128 ruptures at the central andeastern part of the T–G fault zone and the 71 ruptures ofanother branch of the eastern part of the same fault zone.These branches correspond to zones 7b and 9, for whichthe corresponding fault-plane distributions are presentedin the black rosediagram for seismological data. It isimportant to notice that the presence of NE–SW rup-tures in zone 7b almost perpendicular to the dominantWNW–ESE to W–E structure, as the neotectonic datashow (Tranos et al., 2003, 2004), is independentlyproposed by the seismological data. In both cases, a verygood agreement of the fault rupture azimuth for bothneotectonic and seismological data shows that theapplied inversion method for the stress field estimationallowed not only the reliable determination of the stressfield but the identifications of the main active faultdistribution for the study area.
6. Conclusions
In the present work a detailed seismotectonic study ofthe Mygdonia basin is realized, showing a general N–Sextension that is also derived in this study. The ap-plication of inversion method for the stress field studyshows that information provided by the use of seis-mological data is in a good agreement with the stresspattern described by the available neotectonic observa-tions (Fig. 4). The average difference between theminimum principal stress directions and the extensionalstress axes for the seismological and neotectonic ap-proach is typically less than 10°, suggesting that thestress tensor inversion using fault plane solutions ispractically identical to the stress field shown byindividual neotectonic data. This agreement is importantand quite surprising, if we consider that the seismolog-ical information concerns a few mainshocks (MWN5.5),several intermediate magnitude events (5.5NMWN3.5)recorded by the permanent network and three localexperiments which monitored the 1978 aftershock se-quence and the 1984–1984 and 2001–2002 microearth-quake activity. Hence, we can conclude that, at least forthe study area, earthquake fault plane solutions can beefficiently used in order to determine the active seis-motectonic setting, even when coming from very diversesources.
The usage of fault plane solutions allowed the deter-mination of the stress pattern for areas not previouslystudied or not directly related with already known faultsidentified by field observations. The processing of thefocal mechanisms confirms the relation between theoccurred earthquakes and several rupture zones, evensome which are not clearly identified on the surface,
confirming the significant effect of the pre-existingseismotectonic environment on the development of activedeformations on existing faults through earthquakes.These secondary faulting branches (mostly in the westernpart of the Mygdonia basin) are probably related in mostcases with blind faults, without a significant effect on thegeomorphological structure of the area.
The detailed investigation of the study area by thedefinition of zones of similar fault plane solutions andneotectonic observations shows a significant small-scalevariation in the local extensional stress field, which canreach up to 70° (e.g. compare zones 2 and 8 — seeTable 3). Such variations are found even for neighboringzones at distances of the order of 10 km (e.g. zones 7band 8 show a difference of at least 25° of their T-axes—see Table 3). This significant small-scale variation of thestress field appears to contradict previous large-scaleresults (e.g. Papazachos and Kiratzi, 1996), which showa more or less gradual variation of the extensional axis inthe back-arc Aegean area from NNE–SSW in westernAnatolia to NW–SE in western Greece. However, suchlarge-scale studies can not depict the small-scale vari-ations of the stress field in smaller regions, such as theMygdonia basin area, as they fail to reflect the role of thepre-existing fault system on active tectonics. Hence, theinformation presented in the present work can provideaccurate information concerning active tectonics in thevicinity of a major metropolitan center such as the city ofThessaloniki, allowing the realistic modeling of strongearthquake generation scenarios, especially in zoneswhich are favorably oriented with the large-scale stressfield found from large-scale studies.
Acknowledgements
This work was partly supported by the EuropeanCommission Project No. EVG1-CT-2001-00040EUROSEIS-RISK.
References
Aki, K., Richards, P., 1980. Quantative Seismology: Theory andMethods. Freeman, San Francisco, California. 557 pp.
Angelier, J., Dumont, J.F., Karamandersei, H., Poisson, A., Sinsek, S.,Uysal, S., 1981. Analyses of fault mechanisms and expansion ofsouthwestern Anatolia since the late Miocene. Tectonophysics 75,T1–T9.
Angelier, J., Tarantola, A., Valette, B., Manoussis, S., 1982. Inversionof field data in fault tectonics to obtain the regional stress.-I. Singlephase fault populations: a new method of computing the stresstensor. Geophys. J. R. Astron. Soc. 69, 607–621.
Armijo, R., Carey, E., Cisternas, A., 1982. The inverse problem inmicrotectonics and the separation of tectonic phases. Tectonophy-sics 82, 145–160.
266 D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
Bott, M.H.P., 1959. The mechanics of oblique slip faulting. Geol. Mag.96, 109–117.
Caccamo, D., Neri, G., Sarao, A., Wyss, M., 1996. Estimates of stressdirections by inversion of earthquakes fault-plane solutions inSicily. Geophys. J. Int. 125, 85–868.
Christodoulou, A., Etude Sismotectonique et Inversion Tridimen-sionnelle en Grece du Nord. PhD thesis, University of Grenoble I,pp. 181, 1986.
Cocina, O., Neri, G., Privitera, E., Spampinato, S., 1997. Stress tensorcomputations in the Mt. Etna area (Southern Italy) and tectonicimplications. J. Geodyn. 23 (2), 109–127.
Cocina, O., Neri, G., Privitera, E., Spampinato, S., 1998. Seismogenicstress field beneath Mt. Etna (South Italy) and possible relation-ships with volcano–tectonic features. J. Volcanol. Geotherm. Res.83, 335–348.
Etchecopar, A., Vasseur, G., Daignieres, M., 1981. An inverse problemin microtectonics for the determination of stress tensors from faultstriation analysis. J. Struct. Geol. 3, 51–65.
Eva, E., Solarino, S., 1998. Variations of stress directions in thewestern Alpine arc. Geophys. J. Int. 135, 438–448.
Frepoli, A., Amato, A., 2000. Spatial variation in stresses in peninsularItaly and Sicily from background seismicity. Tectonophysics 317.
Gephart, J.W., 1990a. Stress and the direction of slip on fault planes.Tectonics 9 (4), 845–858.
Gephart, J.W., 1990b. FMSI: a FORTRAN program for inverting fault/slickenside and earthquake focal mechanism data to obtain theregional stress tensor. Comput. Geosci. 16 (7), 953–989.
Gephart, J.W., Forsyth, D.W., 1984. An improved method fordetermining the regional stress tensor using earthquake focalmechanism data: application to the San Fernando earthquakesequence. J. Geophys. Res. 89 (B11), 9305–9320.
Goldstein, P., Minner, L., 1995. A Status Report On The DevelopmentOf SAC2000: ANewSeismic Analysis Code. Lawrence LivermoreNational Laboratory, Livermore, CA. UCRL-ID-121523.
Hardebeck, J., Shearer, P., 2003. Using S /P amplitude ratios toconstrain the focal mechanisms of small earthquakes. Bull.Seismol. Soc. Am. 93, 2434–2444.
Hatzfeld, D., Christodoulou, A.A., Scordilis, E.M., Panagiotopoulos,D., Hatzidimitriou, P.M., 1987. A microearthquake study of theMygdonian graben (northern Greece). Earth Planet. Sci. Lett. 81,379–396.
Hatzidimitriou, P.M., Scordilis, E.M., Papadimitriou, E.E., Hatz-feld, D., Christodoulou, A.A., 1991. Microearthquake studyof the Thessaloniki area (northern Greece). Terra Nova 3,648–654.
Hinzen, K.-G., 2003. Stress field in the Northern Rhine area, CentralEurope, from earthquake fault plane solutions. Tectonophysics 377(3–4, 31), 325–356.
Horiuchi, S., Rocco, G., Hasegawa, A., 1995. Discrimination of faultplanes from auxiliary planes based on simultaneous determina-tion of stress tensor and a large number of fault plane solutions.J. Geophys. Res. 100 (B5), 8327–8338.
Kiratzi, A.A., 1999. Stress tensor inversion in Western Greece usingearthquake focal mechanisms from the Kozani–Grevena 1995seismic sequence. Ann. Geophys. 42 (4).
Loohuis, J., van Eck, T., 1996. Simultaneous focal mechanism andstress tensor inversion using a genetic algorithm. Phys. Chem.Earth 21 (4), 267–271.
Martinod, J., Hatzfeld, D., Savvaidis, P., Katsambalos, K., 1997. RapidN–S extension in Mygdonian graben (Northern Greece) deducedfrom repeated geodetic surveys. Geophys. Res. Lett. 24 (24),3293–3296.
McKenzie, D.P., 1969. The relationship between fault plane solutionsfor earthquakes and the directions of the principal stresses. Bull.Seismol. Soc. Am. 59 (2), 591–601.
Mountrakis, D., Psilovikos, A., Papazachos, B.C., 1983. Thegeotectonic regime of the 1978 Thessaloniki earthquake. In:Papazachos, B.C., Carydis, P.G. (Eds.), The Thessaloniki,Northern Greece, Earthquake of June 20, 1978 and Its SeismicSequence, pp. 11–27.
Mountrakis, D., Kilias, A., Pavlidis, S., Sotiriadis, L., Psilovikos, A.,Astaras, T., Vavliakis, E., Koufos, G., Dimopoulos, G., Soulios, G.,Christaras, B., Skordilis, E., Tranos, M., Spyropoulos, N., Patras,D., Siridis, G., Lamprinos, N., Laggalis, T., 1996. Neotectonic mapof Greece, Langadas Sheet, Scale 1 :100.000. Edition of theEarthquake Planning and Protection Organisation (OASP) and theEuropean Centre on Prevention and Forecasting of Earthquakes(ECPFE) (in Greek).
Mountrakis, D., Kilias, A., Pavlidis, S., Sotiriadis, L., Psilovikos, A.,Astaras, T., Vavliakis, E., Koufos, G., Dimopoulos, G., Soulios, G.,Christaras, B., Skordilis, E., Tranos, M., Spyropoulos, N., Patras,D., Siridis, G., Lamprinos, N. and Laggalis, T., 1997. Specialedition of the Neotectonic map of Greece, Langadas Sheet, Scale1 :100.000 (explanatory issue). Edition of the Earthquake Planningand Protection Organisation (OASP) and the European Centreon Prevention and Forecasting of Earthquakes (ECPFE), 1–79(in Greek).
Mountrakis, D., Kilias, A., Tranos, M., Thomaidou, E., Papazachos, C.,Karakaisis, G., Scordilis, E., Chatzidimitriou, P., Papadimitriou, E.,Vargemezis, G., Aidona, E., Karagianni, E., Vamvakaris, D.,Skarlatoudis, A., 2003. Determination of the settings and theseismotectonic behavior of themain seismic-active faults ofNorthernGreece area using neotectonic and seismological data. EarthquakePlanning and Protection Organisation (OASP) (in Greek).
Neri, G., Wyss, M., 1993. Preliminary results from stress tensorinversion of earthquake fault-plane solutions in the southernTyrrhenian region. Boll. Geofis. Teor. Appl. 35 (139), 349–362.
Papazachos, C.B., Kiratzi, A.A., 1992. A formulation for reliableestimation of active crustal deformation and its application tocentral Greece. Geophys. J. Int. 111, 424–432.
Papazachos, B.C., Papazachou, K., 1997. The Earthquakes of Greece.Ziti Publi, Thessaloniki, Greece. 304 pp.
Papazachos, B., Mountrakis, D., Psilovikos, A., Leventakis, G., 1979a.Surface fault traces and fault plane solutions of May–June 1978major shocks in the Thessaloniki area, Greece. Tectonophysics 53,171–183.
Papazachos, B.C., Mountrakis, D., Psilovikos, A., Leventakis, G.,1979b. Focal properties of the 1978 earthquakes in theThessaloniki area. Bulg. Geophys. J. 6, 72–80.
Papazachos, C., Soupios, P., Savvaidis, A., Roumelioti, Z., 2000.Identification of Small-Scale Active Faults Near MetropolitanAreas: An Example From The Asvestochori Fault NearThessaloniki.
Papazachos, C.B., Kiratzi, A.A., 1996. A detailed study of the activecrustal deformation in the Aegean and surrounding area.Tectonophysics 253, 129–153.
Papazachos, C.B, Vamvakaris, D.A., Vargemezis, G.N., Aidona, E.V.,2001. A study of the active tectonics and deformation in theMygdonia basin (N. Greece) using seismological and neotectonicdata. 9th International Congress of the Geological Society ofGreece. Bull. of the Geol. Soc. Of Greece., vol. XXXIV. No. 1,pp. 303–309.
267D.A. Vamvakaris et al. / Tectonophysics 417 (2006) 249–267
study of the broader Thessaloniki area. 5th International Symposiumon Eastern Mediterranean Geology, Thessaloniki.
Pavlidis, S.B., Kilias, A.A., 1987. Neotectonic and active faults alongthe Servomacedonian zone (SE Chalkidiki, northern Greece). Ann.Tecton. I (2), 9–104.
Pavlides, S.B., Kondopoulou, D.P., Kilias, A.A., Westphal, M., 1988.Complex rotational deformations in the Serbo-Macedonian massif(north Greece): structural and palaeomagnetic evidence. Tectono-physics 145, 329–335.
Pavlides, S., 1996. First palaeoseismological results from Greece.Ann. Geophys. XXXIX (3).
Reasenberg, P., Oppenheimer, D., 1985. FPFIT, FPPLOT andFPPAGE: fortan computer programs for calculating and displayingearthquake fault-plane solutions. USGS Open-File Report,vol. 85–739, p. 109.
Scordilis, E.M., 1985 Microseismic study of the Servomacedonianzone and the surrounding area. Ph.D. Thesis, University ofThessaloniki (in Greek).
Soufleris, C., Jackson, J.A., King, G.C.P., Spencer, C.P., 1983.Thessaloniki 1978 earthquakes: locally recorded aftershocks. In:Papazachos, B.C., Carydis, P.G. (Eds.), The Thessaloniki,Northern Greece, Earthquake of June 20, 1978 and Its SeismicSequence, pp. 159–185. Thessaloniki.
Tapley, W.C., Tull, J.E., 1992. SAC — Seismic Analysis Code: UsersManual. Lawrence Livermore National Laboratory. March 20,Revision 4. 388 pp.
Tranos, M.D., Papadimitriou, E.E., Kilias, A.A., 2003. Thessaloniki–Gerakarou fault zone (TGFZ): the western extension of the 1978Thessaloniki earthquake fault (Northern Greece) and seismichazard assessment. J. Struct. Geol.
Tranos, M.D., Meladiotis, I.D., Tsolakopoulos, E.M., 2004. Geomet-rical characteristics, scaling properties and seismic behaviour ofthe faulting of Chortiatis region and Anthemountas basin(Northern Greece). 5th International Symposium on EasternMediterranean Geology, Thessaloniki.
Tull, J.E., 1987. SAC — Seismic Analysis Code. Tutorial Guide forNew Users. Lawrence Livermore National Laboratory, Livermore,CA. UCRL-MA-112835.
Vamvakaris, D.A., Papazachos, C.B., Savvaidis, P.D., Tziavos, I.N.,Karagianni, E.E., Scordilis, E.M., Hatzidimitriou, P.M., 2003.Stress-field and time-variation of active crustal deformation in theMygdonia basin based on the joined interpretation of seismolog-ical, neotectonic and geodetic data. Geophys. Res. Abstr. 5, 08794.
Vamvakaris, D.A., Papazachos, C.B., Karagianni, E.E., Scordilis,E.M., Hatzidimitriou, P.M., 2004. Determination of fault planesolutions using waveform amplitudes and radiation pattern in theMydgonia basin (N. Greece). Bull. Geol. Soc. Greece XXXVI.