a probabilistic seismic hazard assessment for greece and the … · 2019. 11. 4. · an epicentre...

1675

ANNALS OF GEOPHYSICS, VOL. 47, N. 6, December 2004

Key words seismicity of Greece – probabilisticseismic hazard – peak ground acceleration – designearthquake procedure

1. Introduction

The basic lithospheric process in theAegean Sea is the subduction of the Africanplate under the Eurasian plate south of the Is-land of Crete, forming the Hellenic Arc andTrench (Papazachos and Comninakis, 1971;Comninakis and Papazachos, 1972; Makris,1973; Mercier, 1977; McKenzie, 1978; Deweyand Sengor, 1979; Makropoulos and Burton,1984). Because of this process, Greece is one of

A probabilistic seismic hazard assessmentfor Greece and the surrounding regionincluding site-specific considerations

Theodoros M. Tsapanos (1), Päivi Mäntyniemi (2) and Andrzej Kijko (3)(1) Aristotle University of Thessaloniki, School of Geology, Geophysical Laboratory, Thessaloniki, Greece

(2) Institute of Seismology, University of Helsinki, Finland(3) Council for Geoscience, Pretoria, South Africa

AbstractA probabilistic approach was applied to map the seismic hazard in Greece and the surrounding region. The pro-cedure does not require any specification of seismic sources or/and seismic zones and allows for the use of thewhole seismological record, comprising both historical and instrumental data, available for the region of inter-est. The new seismic hazard map prepared for Greece and its vicinity specifies a 10% probability of exceedanceof the given Peak Ground Acceleration (PGA) values for shallow seismicity and intermediate soil conditions foran exposure time of 50 years. When preparing the map, the new PGA attenuation relation given by Margaris etal. (2001) was employed. The new map shows a spatial distribution of the seismic hazard that corresponds wellwith the features of shallow seismicity within the examined region. It depicts the level of seismic hazard in whichthe exceedance of the PGA value of 0.25 g may be expected to occur within limited areas. The highest estimat-ed levels of seismic hazard inside the territory of Greece are found in the Northern Sporades Islands, where PGAvalues in excess of 0.50 g are reached at individual sites, and in the Zante Island in Western Greece, where PGAvalues in the range of 0.35 g to 0.40 g are obtained at more numerous localities. High values are also observedin the sea between the Karpathos and Rhodes islands, near the Island of Amorgos (Cyclades Archipelago) andin the Southwestern Peloponnesus. The levels of seismic hazard at the sites of seven Greek cities (Athens, Jan-nena, Kalamata, Kozani, Larisa, Rhodes and Thessaloniki) were also estimated in terms of probabilities that agiven PGA value will be exceeded at least once during a time interval of 1, 50 and 100 years at those sites. Theseprobabilities were based on the maximum horizontal PGA values obtained by applying the design earthquakeprocedure, and the respective median values obtained were 0.24 g for Athens, 0.28 g for Jannena, 0.30 g forKalamata, 0.21 g for Kozani, 0.24 g for Larisa, 0.43 g for Rhodes and 0.35 g for Thessaloniki. The probabilitiesof exceedance of the estimated maximum possible PGA value were also calculated for the cities to illustrate theuncertainty of maximum PGA assessment.

Mailing address: Dr. Päivi Mäntyniemi, Institute ofSeismology, University of Helsinki, P.O. Box 68, FI-00014,Helsinki, Finland; e-mail: [email protected]

1676

Theodoros M. Tsapanos, Päivi Mäntyniemi and Andrzej Kijko

the most seismically active regions in theworld. In one ranking of 50 seismogenic coun-tries worldwide, Greece took the sixth position(Tsapanos and Burton, 1991).

The objective of seismic hazard assessmentis to obtain long-term probabilities of the oc-currence of seismic events of a specified size ina given time interval. A large number of ap-proaches are currently available for this pur-pose. Abundant literature on the seismic hazardin Greece therefore exists for a number of dif-ferent seismic quantities such as the maximumexpected macroseismic intensity (Shebalin etal., 1976; Papaioannou, 1984; Papoulia andSlejko, 1997), Peak Ground Acceleration(PGA) or velocity (Algermissen et al., 1976;Makropoulos and Burton, 1985) and the dura-tion of strong ground motion (Margaris et al.,1990; Papazachos et al., 1992). One version ofthe geographic distribution of seismic hazard inGreece, based on seismic sources, has been pro-vided by Papazachos et al. (1993). Main andBurton (1989) estimated the seismicity inGreece by considering two different processes,namely subduction and stretching. Seismic haz-ard parameters have been calculated for Greeceusing the maximum likelihood method (Pa-padopoulos and Kijko, 1991), while Papaioan-nou and Papazachos (2000) estimated the seis-mic hazard using new seismotectonic data.

In the present work, the seismic hazard inGreece was assessed in terms of PGA using theapproach described in Kijko and Graham(1998, 1999) and using the recent Greek atten-uation relationship derived by Margaris et al.(2001). Seismic hazard analysis was done onthe basis of the whole seismological recordavailable for Greece, including historical obser-vations as well as the instrumental data record-ed during the past decades, covering a period ofabout two and a half millennia and comprisingsome 600 destructive earthquakes (Papazachosand Papazachou, 1997). Only the shallow seis-micity was taken into consideration, becausethe attenuation relationship given by Margariset al. (2001) was derived for shallow earth-quakes, and they account for the vast majorityof the observed events. A seismic hazard mapwas created for the Greek territory. In this map,the estimated seismic hazard is specified in

terms of the PGA with a 10% probability of ex-ceedance in 50 years, corresponding to a returnperiod of 475 years. In addition, a detailedanalysis of seismic hazard was carried out forthe Greek cities of Athens, Jannena, Kalamata,Kozani, Larisa, Rhodes and Thessaloniki. Thisincluded an assessment of the maximum possi-ble PGA at the sites of the cities and the calcu-lation of the probabilities that a given PGA val-ue will be exceeded at least once during timeintervals of 1, 50 and 100 years at each site. Themaximum PGA values for the seven sites wereobtained by applying the design earthquakeprocedure, assuming the occurrence of thestrongest possible earthquake at the distance of15 km from the site. The probabilities of ex-ceedance of the maximum possible PGA valueswere also calculated to illustrate the uncertain-ty of maximum PGA estimation.

2. The data

The examined region covers the territory ofGreece and its vicinity, including the Greekmainland, parts of its northern contiguouscountries, the Aegean Sea and Western Turkey,between latitudes 33.0°-43.0°N and longitudes19.0°-29.0°E. Figure 1 shows the epicentres ofshallow main shocks of magnitude M ≥ 5 forthis region.

Information on the seismicity of Greece ex-ists from the 6th century B.C. onwards, as de-scriptions of earthquake effects were made bythe ancient Greeks, Romans and Byzantines.The data used in the present study were re-trieved from the databank of the GeophysicalLaboratory of the University of Thessaloniki.This databank comprises information on a largenumber of Greek earthquakes since 550 B.C.(Papazachos et al., 2000). As knowledge of thecompleteness, homogeneity and accuracy ofearthquake data is necessary for a reliable esti-mation of various seismicity parameters, theuncertainties and thresholds for complete re-porting were assessed as follows: the catalogueof shallow seismicity is complete for the areaunder investigation for magnitudes M ≥ 8.0since 550 B.C., for M ≥ 7.3 since 1501 A.D.,M ≥ 6.0 since 1845, M ≥ 5.0 since 1911, M ≥ 4.5

1677

A probabilistic seismic hazard assessment for Greece and the surrounding region including site-specific considerations

since 1950, M ≥ 4.3 since 1964 and for magni-tudes M ≥ 4.0 since 1981 (Papazachos et al.,2000). All magnitudes are given on a scaleequivalent to the moment magnitude (for moredetails see: Papazachos et al., 1997). During theinstrumental measurement era after 1911, the er-ror for the epicentres is less than 20 km, whilethe uncertainty of magnitudes is 0.25 magnitudeunits (Papazachos and Papazachou, 1997). Theerrors affecting the historical data between 550B.C. and 1910 are of the same order, mainly be-cause these were strong shocks with substantial

macroseismic information available. Thus, forthe historical earthquakes the uncertainties forthe epicentre and magnitude are less than 30 kmand 0.4 magnitude units, respectively (Papaza-chos and Papazachou, 1997). If the number ofmacroseismic observations available for an eventis less than ten, the uncertainty of magnitudemay reach 0.5 magnitude units.

In the present study, foreshocks, aftershocksand earthquake swarms were omitted from theinitial earthquake catalogue. This removal wasbased on a relationship derived by Papazachos

Fig. 1. An epicentre map of shallow main shocks of magnitude Mw ≥ 5 for Greece and the surrounding regionbetween 550 B.C. and 1999. The letters mark the sites of the cities of Athens, Jannena, Kalamata, Kozani, Lar-isa, Rhodes and Thessaloniki for which site-specific analyses of seismic hazard were performed.

1678


and Papazachou (1997), in which the durationof the aftershock sequence depends on the mag-nitude of the main shock. The remaining shal-low main events served as the input data for theseismic hazard analyses.

3. Outline of the methodology

The method used to estimate the level ofseismic hazard in terms of PGA has been de-scribed in detail in Kijko and Graham (1998,1999). The first part of their work focuses onthe development and presentation of statisticaltechniques that can be used for the evaluation ofthe maximum regional magnitude, mmax. Thesecond part delineates a methodology for prob-abilistic seismic hazard assessment at a givensite. In the present study, emphasis is placed onthe second aspect.

Site-specific analyses of seismic hazard require a knowledge of the attenuation of theselected ground-motion parameter a, usuallyPGA, as a function of earthquake magnitudeand distance. According to the adopted method-ology, the attenuation relationship of PGA isassumed to be of the type

( ) ( )ln a c c m r1 2 $= + + +z f (3.1)

where c1 and c2 denote empirical coefficients, mis the earthquake magnitude, φ (r) is a functionof earthquake distance and ε is a normally dis-tributed random error.

To express seismic hazard in terms of PGA,the aim would be to calculate the conditionalprobability that an earthquake of random mag-nitude, occurring at a random distance from thesite, will cause a PGA value equal to, or greaterthan, the chosen threshold value, amin, at thesite. We accept the standard assumption (e.g.,Page, 1968) that the random earthquake magni-tude m, in the range of mmin≤ m ≤ mmax, is dis-tributed according to the doubly truncatedGutenberg-Richter relation with a CumulativeDistribution Function (CDF) of

( ) .exp exp

exp expF m

m m

m m

min max

min

M =- - -

- - -

b b

b b

_ _

_ _

i i

i i

(3.2)

In eq. (3.2), mmin is the minimum earthquakemagnitude corresponding to amin, which is theminimum value of PGA of engineering interestat the site, mmax is the maximum credible earth-quake magnitude and β = bln10, where b is theparameter of the Gutenberg-Richter magnitude-frequency relation. It can be shown (Kijko andGraham, 1999) that choosing eq. (3.1) as amodel for attenuation of PGA and eq. (3.2) as adistribution of earthquake magnitude, is equiv-alent to the assumption that the CDF of the log-arithm of PGA, x, is of the form

( ) .exp exp

exp expF x

x x

x x

min max

min

X =- - -

- - -

c c

c c

_ _

_ _

i i

i i(3.3)

Above, xmin = ln(amin), xmax = ln(amax), amax is themaximum possible PGA at the site, γ = β/c2where c2 is the coefficient related to the attenu-ation formula (3.1) and β is the parameter relat-ed to the Gutenberg-Richter distribution ofearthquake magnitude (3.2). One should notethat CDF (3.3) was derived under the conditionthat the spatial and magnitude distributions aremutually independent. In practice, the assump-tions of independence mean that within the areasurrounding the specified site, the parametersof earthquake magnitude distribution mmin, mmaxand β are the same.

It can be seen from formula (3.3) that thelogarithm of the PGA at a given site follows thesame type of distribution as the earthquake mag-nitude, i.e. doubly truncated negative exponen-tial – the form of the Gutenberg-Richter distribu-tion in eq. (3.2). The two distributions differ on-ly in the value of their parameters. If the param-eter of the magnitude distribution is equal to β,the parameter of the distribution of x = ln(PGA)is equal to β/c2.

From an engineering point of view, the largestPGA expected at least once at a given site duringa specified time interval, t, is of special interest.The CDF of the logarithm of the largest PGA val-ue, x, to be observed at least once at the site dur-ing a specified time interval t, can be written as

x( )( )

exp

exp expF

t

t F x t

1

1max

X

X

=- -

- - - -

m

m mt

]

]

g

g6 @# -

(3.4)

1679


where λ is the site-specific activity rate ofearthquakes that cause a PGA value, a, at thesite, exceeding the threshold value amin. Clearly,this CDF of the largest PGA values is doublytruncated: from below, by xmin = ln(amin), andfrom above, by xmax = ln(amax). The distributionin eq. (3.4) was derived under the assumptionthat the earthquakes that cause a PGA value a,a ≥ amin, at the site are distributed according tothe Gutenberg-Richter relation (3.2) and, intime, follow the Poisson process with mean ac-tivity rate λ(x) = λ [1−FX(x)], where x = ln(a).

The maximum likelihood method is used toestimate the site-characteristic seismic hazardparameters λ and γ. If a1,…, an are the largestPGA values recorded at the site during n suc-cessive time intervals t1,…, tn, the likelihoodfunction of the sample x1,…, xn, where xi = ln(ai)and i = 1,…, n, for a specified value amax can bewritten as

1

x, (L f maxX i in

=m c=

)ti

%_ i (3.5)

where fXmax(xiti) is the probability density func-

tion of the logarithm of the largest PGA valueobserved at a given site during a given time in-terval t. By definition, the probability densityfunction x x( (f xdmax maxX i i X i i=t t) )Fd . For a giv-en value of xmax (or equivalently, the maximumpossible PGA at the site), maximisation of thelikelihood function (3.5) leads to the determi-nation of the parameters λ and γ. However, thisprocedure for the estimation of the hazard pa-rameters is used only when the b-value (orequivalently β) of the Gutenberg-Richter fre-quency-magnitude relationship is not known.When the b-value is known, parameter γ is cal-culated as β/c2 and the maximum likelihoodsearch (3.5) reduces to the estimation of thesite-specific, mean seismic activity rate λ. In or-der to create seismic hazard maps, the proce-dure is repeatedly applied to grid points cover-ing the area of interest.

4. Attenuation relationship

A new attenuation relationship has recentlybeen proposed for Greece and the surrounding

region by Margaris et al. (2001). They used adata set consisting of records of 142 mainlynormal faulting earthquakes with magnitudes inthe range of 4.5< Mw < 7.0 and epicentral dis-tances between 1 and 150 km to derive the at-tenuation relationship. A total of 744 records ofhorizontal components and 338 records of ver-tical components of the peak ground accelera-tion, velocity and displacement were available.They took into consideration only the shallowearthquakes within the examined region. Theyconsidered soil classification according to theNEHRP (1994) recommendations and suggest-ed empirical predictive relations for the hori-zontal components of strong ground motion.

The new attenuation relation for PGA takesthe following form:

h

( )

.

ln

ln

c c M c

D c c

PGA

S/

w1 2 3

2

0

2 1 2

4 5

$ $

$ $ !

= + +

+ +^ h(4.1)

Above, D is epicentral distance (in km) and h0== 7 km. The parameters of the model are c1 == −3.36, c2 = 0.70, c3 = −1.14 and c4 = 0.12 andthe standard deviation of ln(PGA) is c5 = 0.70.With these values, the attenuation relationshipof eq. (4.1) provides acceleration values in unitsof g. The constant c3 in front of the term ln(D2++h0

2)1/2 includes both anelastic and geometricattenuation. The letter S describes soil classifi-cation by taking values 0, 1 or 2, correspondingto rock (hard), intermediate and alluvium (soft)conditions, respectively.

Earlier attenuation relationships for theGreek territory were proposed by Makropoulosand Burton (1985), following Makropoulos(1978), and also by Papaioannou (1984). Therelationships derived by Theodoulidis (1991)were regarded as the official attenuation rela-tionship for Greece during the 1990s. In orderto obtain more insight into the most recent at-tenuation relation given by Margaris et al.(2001), it was compared with the previous rela-tionships. Figure 2a-c shows the accelerationvalues as a function of distance computed ac-cording to the four different relationships forthree different magnitudes assuming intermedi-ate soil conditions. It can be seen that the newattenuation relationship gives lower values than

1680


the other relationships especially at near dis-tances. This is most pronounced in the case ofMw = 7.0, when the acceleration values based onthe most recent relation are the lowest up to adistance of 130 km. For magnitudes Mw = 5.0and Mw = 6.0, the corresponding distances are80 km and 100 km, respectively. For magni-tudes Mw = 6.0 and Mw = 7.0, only the relation-ship proposed by Papaioannou (1984) giveslower acceleration values for large distancesthan the most recent relationship, whereas formagnitude Mw = 5.0 also the acceleration valuesfrom the relationship of Theodoulidis (1991)are lower for distances in excess of 80 km. Acomparison between the attenuation relation-ships of Theodoulidis (1991) and of Margaris etal. (2001) shows that the difference betweencomputed acceleration values increases with in-creasing magnitude. For example, for a distanceof 20 km, assuming intermediate soil condi-tions and a magnitude Mw = 5.0, the differencebetween median PGA values is only 0.0042 g;for Mw = 6.0 it is 0.042 g and for Mw = 7.0 it is0.18 g, the higher values resulting from the ear-lier attenuation relationship.

The recent Greek attenuation relationship,eq. (4.1), was also compared with that derivedfor the European territory by Ambraseys et al.(1996). Since this relation only provides accel-eration for magnitudes expressed on the Msscale, the version given by Wahlström andGrünthal (2001), with Ms converted to Mw, wasused. The PGA values were computed as afunction of distance for the magnitude Mw = 6.0and a focal depth of 7 km assuming hard rockconditions (fig. 3). It can be seen that the medi-an values of acceleration (corresponding top = 0) given by the Greek relation are consider-ably lower than the values computed accordingto the Ambraseys et al. (1996) relationship for

a Fig. 2a-c. A comparison of Peak Ground Accelera-tion (PGA) values as a function of distance for mag-nitudes (a) Mw = 5.0, b) Mw = 6.0 and (c) Mw = 7.0computed for intermediate soil conditions accordingto the new Greek attenuation relationship given byMargaris et al. (2001) and the relationships proposedearlier for Greece by Makropoulos (1978), Papaioan-nou (1984) and Theodoulidis (1991).

b

c

1681


distances in excess of 5 km. The Greek valuescorresponding to the 84th percentile level(p = 1), computed by taking into account thestandard deviation of the normal distribution ofln(PGA), are lower than those computed withAmbraseys et al. (1996) relationship for dis-tances larger than about 20 km. This behaviourwas also observed for magnitudes Mw = 5.0 andMw = 7.0.

5. Results and discussion

5.1. Seismic hazard map

Figure 4 shows the new seismic hazard mapfor Greece, computed according to the above

procedure and the recently developed attenua-tion relation (4.1). The seismic hazard is ex-pressed in terms of PGA values with a 10%probability of exceedance at least once in 50years, corresponding to a return period of 475years. The map was prepared for shallow seis-micity, assuming intermediate soil conditionsand using a grid point mesh of 0.20°. The b pa-rameter was assumed to take the constant valueb = 1 over the whole region of interest (see, e.g.,Papazachos and Papazachou, 1997).

The overall spatial distribution of seismichazard as depicted in the map corresponds wellwith the main features of tectonics and shallowseismicity in the examined region. A belt of in-creased seismic hazard runs along the HellenicTrench, the Island of Crete and in WesternGreece, continuing north into Albania, whereasits eastern part joins the west coast of Turkey.This zone is related to the active convergentboundary between the Aegean and the Africanplates in the south and to the continental colli-sion between Apulia and the Aegean in the west.The highest estimated levels of seismic hazard inthis zone are located in the Zante Island in West-ern Greece. High values are also observed in thesea between the Karpathos and Rhodes islands,near the Island of Amorgos (Cyclades Archipel-ago) and in the Southwestern Peloponnesus. Thehighest estimated levels of seismic hazard insidethe territory of Greece are found in the NorthernSporades Islands, which corresponds to thewestern termination of the North AnatolianFault. Intermediate values dominate in the Gulfof Corinth, in the Athos Peninsula (Chalkidiki-Northern Greece), Thessaloniki and in the Ion-ian Islands. Northeastern Greece, the Sea ofCrete and Northwestern Greece appear as areasof the lowest seismic hazard. Most of the zone ofincreased seismic hazard corresponding to theHellenic Arc and Trench, as well as the NorthAegean and Western Turkey, shows PGA valuesin the range of 0.20 g to 0.25 g, while values inexcess of 0.25 g are reached within limited ar-eas. The maximum computed PGA values insidethe territory of Greece are in excess of 0.50 gand can be found at individual sites in the North-ern Sporades Islands, whereas PGA values in therange of 0.35 g to 0.40 g are obtained at morenumerous localities in the Ionian Islands.

Fig. 3. A comparison of Peak Ground Acceleration(PGA) values as a function of distance for the mag-nitude Mw = 6.0 and a focal depth of 7 km computedaccording to the new Greek attenuation relationshipgiven by Margaris et al. (2001) and that derived forthe European territory by Ambraseys et al. (1996).The displayed PGA values were computed for hardrock conditions. Notation p= 0 stands for medianPGA values, whereas in the case of p= 1 the PGAvalues were computed by taking into account thestandard deviation of the normal distribution ofln(PGA), corresponding to the 84th percentile levelof non-exceedance.

1682


Makropoulos and Burton (1985) presentedseismic hazard maps for Greece in terms ofPGA. They applied extreme value analyses andemployed an attenuation relationship based onthe averaging of eight formulae reported inworldwide studies. The spatial distribution ofseismic hazard depicted in their maps is rathersimilar to that shown in fig. 4. Makropoulos andBurton (1985) obtained a high PGA hazardaround Cephalonia and the Levkas islands aswell as around Lesvos and the Northeastern Spo-rades Islands. They reported a 30% probabilityof the PGA exceeding 0.2 g in 50 years withinthese areas.

5.2. Sites of the cities

Figure 5a-g shows the probabilities of thegiven PGA values being exceeded within timeintervals of 1, 50 and 100 years at the sitesAthens, Jannena, Kalamata, Kozani, Larisa,

Fig. 4. A new seismic hazard map for Greece based on the attenuation relationship of Margaris et al. (2001).The hazard is expressed in terms of Peak Ground Accelaration (PGA) values for intermediate soil conditionswith a 10% probability of exceedance at least once in 50 years.

Fig. 5a. Probabilities that the given Peak Ground Ac-celeration (PGA) values will be exceeded in 1, 50 and100 years at the Athens site. The maximum PGA val-ues are the median values obtained using the designearthquake procedure.

1683


b

c

d

e

Fig. 5b-g. Probabilities that the given Peak Ground Acceleration (PGA) values will be exceeded in 1, 50 and100 years at the sites: b) Jannena, c) Kalamata, d) Kozani, e) Larisa, f) Rhodes and g) Thessaloniki. The maxi-mum PGA values are the median values obtained for each site using the design earthquake procedure.

f

g

1684


Rhodes and Thessaloniki. These probabilitycurves are based on the estimated values for γ,λ and amax in formulae (3.3) and (3.4). The val-ues for the maximum possible PGA, amax, wereobtained using the attenuation relation (4.1) andassuming the occurrence of the strongest possi-ble earthquake, m̂max, at a distance of 15 kmfrom the site. The maximum credible magni-tudes, m̂max, were computed using data for a ra-dius of 180 km from each city centre and the K-S-B procedure (see Kijko and Graham, 1998:eq. 54). The maximum (median) PGA values,amax, were 0.24 g for Athens, 0.28 g for Jan-nena, 0.30 g for Kalamata, 0.21 g for Kozani,0.24 g for Larisa, 0.43 g for Rhodes and 0.35 gfor Thessaloniki. The probability plots forAthens and Thessaloniki can also be found inMäntyniemi et al. (2004).

Figure 6 illustrates the uncertainty of maxi-mum PGA assessment for the cities of Athens,Jannena, Kalamata, Kozani, Larisa, Rhodes andThessaloniki. It gives the probabilities that an

earthquake of maximum credible magnitude, asestimated for each area, will occur at a hypo-central distance of 15 (± 5) km from the respec-tive site and produce a PGA value exceedingthe estimated maximum value. In the calcula-tions it was assumed that the standard deviationof estimated maximum values of the PGA is af-fected by three factors: 1) the uncertainty in de-termination of the earthquake epicentral dis-tance; 2) the maximum earthquake magnitudemmax and 3) the uncertainty of attenuation rela-tionship. This kind of approach can be regardedas a probabilistic extension of the deterministic,design-earthquake scenario. The probabilitiesobtained for the exceedance of a high PGA val-ue, for example 0.70 g, range between 36% forRhodes and 27% for Kozani. The probabilitiescomputed for Athens and Larisa overlap, andalso the assessed maximum PGA values forthese sites are the same.

The computed exceedance probabilities ofthe given acceleration values at the site of eachcity for an exposure time of 100 years are com-pared in fig. 7. Rhodes stands out from the oth-er cities with the highest expected PGA ex-

Fig. 6. The probabilities that an earthquake of max-imum magnitude, as estimated for each area, will oc-cur at a hypocentral distance of 15 (±5) km from therespective site and produce a Peak Ground Accelera-tion (PGA) value exceeding the estimated maximumvalue at the sites Athens, Jannena, Kalamata, Kozani,Larisa, Rhodes and Thessaloniki. The probabilitiescomputed for Athens and Larisa overlap.

Fig. 7. The probabilities of exceedance of the givenPeak Ground Acceleration (PGA) values at the sitesAthens, Jannena, Kalamata, Kozani, Larisa, Rhodesand Thessaloniki during 100 years. The probabilitycurves for Athens and Kalamata overlap.

1685


ceedance probabilities. The city of Jannena dis-plays the second highest long-term probabili-ties, but they are much lower than those forRhodes. Despite its proximity to the HellenicArc and Trench, Kalamata does not rank veryhigh in the comparison. The computed proba-bilities for this site are practically identical tothose for Athens, whereas higher values wereobtained for the northern cities of Kozani andThessaloniki. This is in contrast with expecta-tions, as Kozani is situated in northwesternGreece, a region of negligible historical seis-micity (Papazachos and Papazachou, 1997).The instrumentally recorded seismicity thereseems to be mainly related to the filling of theartificial Lake Polyfytos (Papazachos et al.,1995). The area has been assumed to be of verylow seismic hazard, and therefore the destruc-tive Kozani earthquake (Ms= 6.6) of 13 May1995 was quite unexpected (Papazachos et al.,1995). Northwestern Greece appears as an areaof low seismic hazard also in fig. 4.

The seismicity features in areas adjacent toThessaloniki seem to be characterised by earth-quake sequences of long duration, extending tothe northern contiguous countries (Papazachoset al., 1979). The city of Larisa has the lowestprobabilities for the exceedance of the givenPGA values. The probabilities computed forthis site for exposure times of 1 and 100 yearsdo not deviate very much from each other (fig.5d), which can be attributed to the small activi-ty parameter λ S for this site. The computedprobabilities decrease rather quickly with in-creasing PGA values for all sites. For example,for the sites of Rhodes and Jannena, the PGAvalues with a 50% chance of exceedance with-in 100 years are 0.058 g and 0.040 g, respec-tively (fig. 7).

Previous studies have also provided someestimates of seismic hazard in terms of themaximum expected PGA at the sites of thecities investigated in the present work.Makropoulos and Burton (1985) estimatedground-motion for the cities of Athens, Thessa-loniki and Rhodes, among others. They appliedextreme value analysis to compute ground ac-celeration values and to produce seismic hazardparameters. The acceleration values they re-ported that had a 70% probability of non-ex-

ceedance for 50 and 100 years were, respec-tively, 0.09 g and 0.11 g for Athens, 0.15 g and0.17 g for Thessaloniki and 0.075 g and 0.08 gfor Rhodes. Thus, for the city of Rhodes theirestimated PGA values are the lowest, althoughin terms of earthquake size the area surroundingthis city has a high seismic potential. In the de-terministic approach used to compute the amaxvalues of the PGA shown in fig. 5a-g, the max-imum credible magnitude plays an importantrole, thus the highest amax value is obtained forthe site of Rhodes. Also the highest long-termprobabilities for the exceedance of the givenPGA values were obtained for this site (fig. 7).

Lyubushin et al. (2002) evaluated the levelof seismic hazard in terms of maximum valuesof PGA for six sites in Greece, including Kala-mata and Kozani, using a Bayesian procedure.They estimated a maximum PGA value of 0.36g for Kalamata and 0.38 g for Kozani. The at-tenuation relationship of Theodoulidis (1991)was employed to derive these estimates. Themaximum PGA values computed by Lyubushinet al. (2002) are higher than those estimated inthe present study, which probably largely fol-lows from the use of different attenuation rela-tionships. As discussed in Section 4, the mostrecent attenuation relationship elucidated byMargaris et al. (2001) gives faster attenuationthan do the previous Greek relations up to dis-tances between 80-130 km depending on themagnitude.

Directly measured ground motion valuesbecame available following the magnitude Mw== 6.0 event of 1986 near Kalamata (Anagnos-topoulos et al., 1987) and the magnitude Mw== 6.6 event of 1995 near Kozani (Theodoulidisand Lekidis, 1996). The observed value was0.27 g for an epicentral distance of 12 km fromKalamata. The largest aftershock of the Kala-mata earthquake was of magnitude Mw = 5.4 andthe respective recorded PGA value about 0.05 gat an epicentral distance of approximately 1 km(Anagnostopoulos et al., 1987). In the presentstudy, the design earthquake procedure yieldeda value of maximum PGA equal to 0.30 g,based on the maximum credible magnitude of7.50 (± 0.25). This maximum magnitude ex-ceeds the range of observed magnitudes avail-able in the derivation of the PGA attenuation re-

1686


lationship of Margaris et al. (2001), and there-fore the computed attenuation can be consid-ered more uncertain than that for lower magni-tudes. As displayed in fig. 6, the estimated max-imum PGA values can be exceeded with thegiven probabilities. Also, the observed PGAvalue may have been affected by site effects, asKalamata is built on the natural deposits of adried up river (Anagnostopoulos et al., 1987).The measured ground motion value followingthe Kozani earthquake was 0.21 g for a distanceof 19 km from the city (Theodoulidis andLekidis, 1996), which equals the present valuerelying on a maximum credible magnitude of7.0 (± 0.25). Both the observed values for theKalamata and Kozani main shocks are lowerthan the maximum PGA values estimated byLyubushin et al. (2002).

6. Conclusions

In the present study, seismic hazard mapsand site-specific assessments of seismic hazardwere prepared for the territory of Greece in-cluding its vicinity and the sites of seven Greekcities (Athens, Jannena, Kalamata, Kozani, Lar-isa, Rhodes and Thessaloniki) using themethodology developed by Kijko and Graham(1998, 1999) and the recent PGA attenuationrelationship derived for Greece by Margaris etal. (2001). The applied technique has been es-pecially developed for probabilistic seismichazard assessment at a specified site, and itdoes not rely on the definition of seismicsources and/or seismic zones. Only the shallowearthquakes were taken into consideration, be-cause the attenuation relationship given byMargaris et al. (2001) was derived using them,and they account for the vast majority of the ob-served events. A seismic hazard map was creat-ed by applying the procedure repeatedly to gridpoints covering the area of interest. In the newmap, the estimated seismic hazard is specifiedin terms of the horizontal PGA with intermedi-ate soil conditions with a 10% probability ofexceedance in 50 years. It depicts a spatial dis-tribution of the seismic hazard that correspondswell with the features of shallow seismicitywithin the examined region. It shows a level of

seismic hazard in which the exceedance of aPGA value of 0.25 g may be expected to occurwithin limited areas. The highest estimated lev-els of seismic hazard inside the territory ofGreece are found in the Northern Sporades Is-lands, where PGA values in excess of 0.50 g arereached at individual sites, and around theZante Island in Western Greece, where PGAvalues in the range of 0.35 g to 0.40 g are ob-tained at more numerous localities. High valuesare also observed in the sea between theKarpathos and Rhodes islands, near the Islandof Amorgos (Cyclades Archipelago) and in theSouthwestern Peloponnesus. Intermediate val-ues dominate in the Gulf of Corinth, in theAthos Peninsula (Chalkidiki-Northern Greece),Thessaloniki and in the Ionian Islands. North-eastern Greece, the Sea of Crete and North-western Greece appear as areas of the lowestseismic hazard. The levels of seismic hazard atthe sites of the seven Greek cities were assessedin terms of probabilities that a given PGA valuewill be exceeded at least once in 1, 50 and 100years at the sites of the cities. The maximum(median) PGA values obtained by applying thedesign earthquake approach were 0.24 g forAthens, 0.28 g for Jannena, 0.30 g for Kalama-ta, 0.21 g for Kozani, 0.24 g for Larisa, 0.43 gfor Rhodes and 0.35 g for Thessaloniki. Theprobabilities of exceedance of the estimatedmaximum possible PGA values were also cal-culated for the cities to illustrate the uncertain-ty involved in this kind of assessment. Some ofthe directly measured PGA values found in theliterature exceed those obtained through the de-sign earthquake procedure, although the respec-tive observed earthquakes were only of moder-ate magnitude. This discrepancy can probablyin part be attributed to site effects. Also, the re-cent Greek attenuation relationship employedin this study predicts rather rapid decrease inmedian PGA values as a function of distance incomparison with many previously published at-tenuation formulae. This applies especially tolarge earthquake magnitudes. The maximumcredible magnitudes used in the design earth-quake procedure also exceed the range of ob-served magnitudes available for its derivation,which adds to the uncertainty of the computedacceleration values.

1687


REFERENCES

ALGERMISSEN, S.T., D.M. PERKINS, W. ISHERWOOD, D. COR-DON, G. REAGOR and C. HOWARD (1976): Seismic riskevaluation of the Balkan region, in Proceedings of theSeminar on Seismic Zoning Maps, Skopje, Yugoslavia,1975, vol. 2, 172-240.

AMBRASEYS, N.N., K.A. SIMPSON and J.J. BOMMER (1996):Prediction of horizontal response spectra in Europe,Earthq. Eng. Struct. Dyn., 2, 371-400.

ANAGNOSTOPOLOUS, S.A., D. RINALDIS, V.A. LEKIDIS, V.N.MARGARIS and N.P. THEODULIDIS (1987): The Kalama-ta, Greece, earthquake of September 13, 1986, Earthq.Spectra, 3, 365-402.

COMNINAKIS, P.E. and B.C. PAPAZACHOS (1972): Seismicityof the Eastern Mediterranean and some tectonic fea-tures of the Mediterranean ridge, Bull. Geol. Soc. Am.,83, 1093-1102.

DEWEY, J.F. and A.M.C. SENGOR (1979): Aegean and sur-rounding regions: complex multiplate and continoustectonics in a convergent zone, Bull. Geol. Soc. Am.,90, 84-92.

KIJKO, A. and G. GRAHAM (1998): Parametric-historic pro-cedure for probabilistic seismic hazard analysis, Part I.Estimation of maximum regional magnitude mmax,Pure Appl. Geophys., 152, 413-442.

KIJKO, A. and G. GRAHAM (1999): «Parametric-historic»procedure for probabilistic seismic hazard analysis,Part II. Assessment of seismic hazard at specified site,Pure Appl. Geophys, 154, 1-22.

LYUBUSHIN, A.A., T.M. TSAPANOS, V.F. PISARENKO andG.CH. KORAVOS (2002): Seismic hazard for selectedsites in Greece: a Bayesian estimate of seismic peakground acceleration, Nat. Hazards, 25, 83-98.

MAIN, I.G. and P.W. BURTON (1989): Seismotectonics andthe earthquake frequency-magnitude distribution in theAegean area, Geophys. J. Int., 98, 575-586.

MAKRIS, J. (1973): Some geophysical aspects of the evolu-tion of the Hellenides, Bull. Geol. Soc. Greece, 10,206-213.

MAKROPOULOS, K.C. (1978): The statistics of large earth-quake magnitude and an evaluation of Greek seismici-ty, Ph.D. Thesis (Univ. Edinburgh, U.K.), pp. 193.

MAKROPOULOS, K.C. and P.W. BURTON (1984): Greek tec-tonics and seismicity, Tectonophysics, 106, 275-304.

MAKROPOULOS, K.C. and P.W. BURTON (1985): Seismichazard in Greece, II. Ground acceleration, Tectono-physics, 117, 259-294.

MÄNTYNIEMI, P., T.M. TSAPANOS and A. KIJKO (2004): Anestimate of probabilistic seismic hazard for five citiesin Greece by using the parametric-historic procedure,Eng. Geol., 72, 217-231.

MARGARIS, V.N., N.P. THEODULIDIS, CH.A. PAPAIOANNOUand B.C. PAPAZACHOS (1990): Strong motion durationof earthquakes in Greece, in Proceedings of the XXIIGen. Assembly ESC, vol. 2, 865-870.

MARGARIS, V., C.B. PAPAZACHOS, CH.A. PAPAIOANNOU, N.THEODOULIDIS, I. KALOGERAS and A. SKARLATOUDIS(2001): Empirical attenuation relations for the strongground motion parameters of shallow earthquakes inGreece, in Proceedings of the 2nd Congress of Engi-neering Seismology and Earthquake Engineering, 28-30 November 2001, Thessaloniki, Greece.

MCKENZIE, D. (1978): Active tectonics of the Alpine-Hi-malayan belt: the Aegean Sea and surrounding regions,Geophys. J. R. Astron. Soc., 55, 217-254.

MERCIER, J. (1977): Principal results of a neotectonic studyof the Aegean Arc and its location within the EasternMediterranean, in Proceedings of the VI Coll. Geol.Aegean Region, IGMR, Athens, Greece, vol. 3, 1281-1291.

NEHRP (1994): Recommended provisions for seismic reg-ulations for new buildings, Federal Emergency Man-agement Agency 222A/223A, May, 1 (Provisions) and2 (Compentary).

PAGE, R. (1968): Aftershocks and microaftershocks of theGreat Alaska earthquake of 1964, Bull. Seismol. Soc.Am., 58, 1131-1168.

PAPADOPOULOS, G.A. and A. KIJKO (1991): Maximum like-lihood estimation of earthquake hazard parameters inthe Aegean area from mixed data, Tectonophysics, 185,277-294.

PAPAIOANNOU, CH.A. (1984): Attenuation of seismic inten-sities and seismic hazard in the area of Greece, Ph.D.Thesis (Aristotle University of Thessaloniki, Greece),pp. 200 (in Greek).

PAPAIOANNOU, CH.A. and B.C. PAPAZACHOS (2000): Time-independent and time-dependent seismic hazard inGreece based on seismogenic sources, Bull. Seismol.Soc. Am., 90, 22-33.

PAPAZACHOS, B.C. and P.E. COMNINAKIS (1971): Geophysi-cal and tectonic features of the Aegean Arc, J. Geo-phys. Res., 76, 8517-8533.

PAPAZACHOS, B. and C. PAPAZACHOU (1997): The earthquakesof Greece (Ziti Publications, Thessaloniki), pp. 304.

PAPAZACHOS, B.C., D. MOUNTRAKIS, A. PSILOVIKOS and G.LEVENTAKIS (1979): Surface fault traces and fault planesolutions of the May-June 1978 major shocks in theThessaloniki area, Greece, Tectonophysics, 53, 171-183.

PAPAZACHOS, B.C., V.N. MARGARIS, N.P. THEODOULIDIS andCH.A. PAPAIOANNOU (1992): Seismic hazard assess-ment in Greece based on strong motion duration, inProceedings of the 10th WCEE, 1, 425-430.

PAPAZACHOS, B.C., CH.A. PAPAIOANNOU, B.N. MARGARISand N.P. THEODULIDIS (1993): Regionalization of seis-mic hazard in Greece based on seismic sources, Nat.Hazards, 8, 1-18.

PAPAZACHOS, B.C., D.G. PANAGIOTOPOULOS, E.M. SCORDILIS,G.F. KARAKAISIS, C.A. PAPAIOANNOU, B.G. KARA-KOSTAS, E.E. PAPADIMITRIOU, A.A. KIRATZI, P.M. HATZI-DIMITRIOU, G.N. LEVENTAKIS, P.G. VOIDOMATIS, K.I.PEFTITSELIS and T.M. TSAPANOS (1995): Focal propertiesof the 13 May 1995 large (Ms = 6.6) earthquake in theKozani area (North Greece), in Proceedings of the XVCongress of the Carpatho-Balkan Geological Associa-tion, 17-20 September 1995, Athens, Greece.

PAPAZACHOS, B.C., A.A. KIRATZI and B.G. KARACOSTAS(1997): Toward a homogeneous moment-magnitudedetermination of earthquakes in Greece and the sur-rounding area, Bull. Seismol. Soc. Am., 87, 474-483.

PAPAZACHOS, B.C., P.E. COMNINAKIS, G.F. KARAKAISIS,B.G. KARAKOSTAS, CH.A. PAPAIOANNOU, C.B. PAPAZA-CHOS and E.M. SCORDILIS (2000): A catalogue of earth-quakes in Greece and surrounding area for the period550 B.C.-1999, Publication Geophys. Lab., AristotleUniversity of Thessaloniki, Greece.

1688


PAPOULIA, J.E. and D. SLEJKO (1997): Seismic hazard assess-ment in the Ionian Islands based on observed macroseis-mic intensities, Nat. Hazards, 14, 179-187.

SHEBALIN, N.V., G.I. REISNER, A.V. DRUMEA, J.V. APTEK-MAN, V.N. SHOLPO, N.Y. STEPANEKS and A.J. ZACHARO-VA (1976): Earthquake origin zones and distribution ofmaximum expected seismic intensity for the Balkan re-gion, in Proceedings of the Seminar on Seismic ZoningMaps, 1975, Skopje, Yugoslavia, vol. 2, 68-171.

THEODOULIDIS, N.P. (1991): Contribution to the study ofstrong ground motion in Greece, Ph.D. Thesis (AristotleUniversity of Thessaloniki, Greece), pp. 499 (in Greek).

THEODOULIDIS, N. and V. LEKIDIS (1996): The Kozani-Grevena, Northern Greece, earthquake of May 13,

1995: strong motion data and structural response, Eur.Earthquake Eng., 1, 3-13.

TSAPANOS, T.M. and P.W. BURTON (1991): Seismic hazardevaluation for specific seismic regions of the world,Tectonophysics, 194, 153-169.

WAHLSTRÖM, R. and G. GRÜNTHAL (2001): Probabilisticseismic hazard assessment (horizontal PGA) forFennoscandia using the logic tree approach for region-alization and nonregionalization models, Seismol. Res.Lett., 72, 33-45.

(received July 16, 2002;accepted July 20, 2004)

a probabilistic seismic hazard assessment for greece and the … · 2019. 11. 4. · an epicentre...

Documents