Earthquake Research in China
Volume 21 , Number 4 , 2007
Recurrence Behaviors of
Earthquakes along the Kefallinia
Transform Fault , Ionian Sea ,
1
Greece
1)
2)
3)
3)
2)
Qi Cheng , Athanassios Ganas , Huang Fuqiong , Chen Yong , and George Drakatos
1) Institute of Geology and Geophysics , Chinese Academy of Sciences , Beijing 100029 , China
2) Institute of Geodynamics , National Observatory of Athens , 11810 Athens , Greece
3) Institute of Earthquake Science , CEA , Beijing 100036 , China
We examined the whole strong earthquake recurrence behaviors of t wo fault zones along the
Kefallinia Transform , Ionian Sea , Greece , using seismological data and statistical methods.
Our data include 29 events with M > 515 for the period 1636 ～ 2003. We found different
recurrence behaviors for the Kefallinia Fault Zone ( clustering and time2predictable recurrence
behaviors) and the Lefkada Fault Zone ( near random and non2slip2predictable or non2time2
predictable recurrence nature ) . The different modes may be attributed to : ( a ) segment
interaction along2strike ( Kefallinia ) by static triggering and ( b) the influence of fault systems
to the north and east on the recurrence on Lefkada. Within the active periods , earthquake
recurrence intervals are distributed in a more dispersed fashion , and can be fitted well by a
Weibull distribution. In contrast , the distribution of the quiet periods is relatively less
dispersed and difficult to describe by suitable probability functions.
Key words : Historical earthquake ; Active fault zone ; Earthquake recurrence ; Probability
distribution ; Greece
INTROD UCTION
Mid2 and long2term earthquake prediction and seismic hazard assessment usually depend on the
knowledge of historical strong earthquake recurrence characteristics. For the study on a single segment
of an active fault zone based on reliable historical and instrumental earthquake data , earthquake
Received in August 20071 This paper is part result of the program“ Investigation of recurrence modes of large
earthquakes along the main seismogenic faults in Greece and in China”(20051696) in the frame of bilateral cooperation
between China and Greece in the field of seismology and geophysics sponsored by the Ministry of Science and Technology
of China. This program is mainly funded by the NSF40674024 and the NSF40374019.
1
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
410
Earthquake Research in China
recurrence behavior usually takes on a relatively simple process , such as time2predictable and quasi2
periodic behaviors ( Shimazaki and Nakata , 1980 ; Bakun and McEvilly , 1984 ; Savage and
Cockerham , 1987 ; Nishenko and Buland , 1987 ; Wen Xueze , 1999) . However , based on ancient
earthquake data , study shows that the recurrence behaviors have an irregular and complicated process
(Jacoby , et . al. , 1988 ; Goes , 1996) . Therefore , reliable long historical earthquake data are very
important for the understanding of earthquake occurrence. Yi Guixi and Wen Xueze ( 2000 ) studied
the earthquake recurrence behavior of a whole intraplate active fault zone and its relation to those on
individual fault segments for the first time by using the earthquake data from 11 active fault zones on
the Chinese mainland. In this study , we studied the whole earthquake recurrence behaviors of
interplate active fault zones in western Greece. The active fault zones include many segments that can
independently generate strong earthquakes.
Due to its particular geological structure , western Greece may be considered a “ natural
laboratory”where many destructive earthquakes have occurred. The area is near a complex plate
boundary where strike2slip motion , thrusting and crustal extension co2exist within a few tens of
kilometers. We focus our attention in the region between Kefallinia and Lefkada of the Ionian Islands
in 38°
～39°North latitude and 20°
～21°East longitude ( Fig. 1) where the main structure is the 100
km2long NE2SW Kefallinia fault which defines the plate boundary. The Kefallinia strike2slip fault
(Scordilis , et al . , 1985 ; Kiratzi and Langston , 1991 ; Louvari , et al . , 1999) accommodates the
relative motion of the Apulia ( Africa ) and Aegean ( Eurasia ) lithospheric plates. The geology of the
)
Ionian Sea is characterized by a series of thrusts striking perpendicular to the E2S ( ± 30°
compression. This area is part of the Pre2Apulian geotectonic zone which was involved in Quaternary
shortening related to the westward propagation of the Hellenic fold and thrust system. The rock types
are sedimentary sequences comprising mainly Mesozoic carbonates. Extensional deformation ( i . e.
Quaternary2Holocene) is superimposed on the Miocene2Pliocene structures. Finally , diapiric movement
of Triassic evaporites has affected both the Alpine and the late Cenozoic to Holocene sedimentary
sequences ( Kokinou and Vafidis , 2003) .
The study area was selected for the following reasons : ( a) it is an area of high seismicity , ( b) it
is located at the plate boundary between Africa and Eurasia and c) there is good seismological and GPS
infrastructure. The region of the South Ionian Islands ( Lefkada , Ithaca , Kefallinia and Zante ) is
characterized by a high seismicity rate with earthquake magnitudes up to 714 ( Papazachos , 1990 ;
Louvari , et al . , 1999 ) . This is attributed to the complex crustal deformation resulting from the
subduction of the African plate towards the NE and the continental collision of the Apulian platform
further to the northwest . The Kefallinia fault zone , KFZ , plays a key role in this geodynamic
complexity ( Sorel , 1976 ; Mercier , et al . , 1987 ; Taymaz , et al . , 1991 ; Le Pichon , et al . , 1995 ;
Papazachos and Kiratzi , 1996 ; Louvari , et al . , 1999 ) and combines active subduction and
continental collision. Its slip2rate is about 20 mmΠa according to GPS measurements ( Serpelloni , et
al . , 2005 ) while , according to seismological data , it is about 3 cmΠa ( Papazachos and Kiratzi ,
1996) . On August 14 , 2003 an M W 612 strike2slip earthquake occurred offshore of Lefkada Island ,
(Fig. 1 ; Papadopoulos , et al . , 2003 ; Pavlides , et al . , 2004 ; Benetatos , et al . , 2005) . The
earthquake caused a lot of injuries and material damage , but no loss of human life. According to NOA
( National Observatory of Athens ) the epicenter was located 8 km offshore Lefkada , towards its
northwestern coast . According to static stress transfer studies ( e. g. Papadimitriou , 2002 ;
Papadimitiou , et al . , 2006) this earthquake may trigger another strong event in the area , so the short2
term seismic hazard may have increased. Our study will examine if it is possible to make a
probabilistic forecast for middle and long2term seismic hazard based only on the recurrence interval
distributions of strong earthquakes. The efficiency of the probabilistic forecast can be improved if
additional evidence can be obtained , from which , whether an active fault zone is in its seismically
active period or in a quiet period can be identified.
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
Volume 21 , Number 4
411
Fig. 1
Map of study area ( Ionian Sea) and of strike2slip fault zones
The earthquakes are divided into four (4) groups. From north to south the groups
are ( a) events associated with the Northern Lefkada Segment (NLS) , ( b) events
associated with the Southern Lefkada Segment ( SLS) , ( c) events associated with the
Northern Kefallinia Segment (NKS) and ( d) events associated with the Southern
Kefallinia Segment ( SKS) . All events are reported in Table 1 and Table 2 ,
respectively. Black arrows indicate plate motion
1 DATA
We selected two faults of the Kefallinia fault zone in the Ionian Sea ( Fig. 1) . The Fault Lefkada
in the north , which further includes the Northern Lefkada Segment ( NLS) and the Southern Lefkada
Segment ( SLS) and the Fault Kefallinia in the south , which includes the Northern Kefallinia Segment
( NKS) and the Southern Kefallinia Segment ( SKS) . The segmentation was mainly done on the basis of
macroseismic data by grouping the data into four regions : North & South Lefkada , North & South
Kefallinia , respectively. We found no evidence for a mega2earthquake rupturing all 120 km of the
Kefallinia transform. However , it is possible that this preliminary segment configuration can be split
into smaller segments but not enough data exists to support a finer segmentation structure. We analyzed
the whole recurrence behaviors of these two faults. On the basis of published focal mechanisms
(Louvari , et al . , 1999) and the location of earthquakes we suggest that all events of our catalogue
have occurred on fault planes with right2lateral strike2slip kinematics. We also assume that strong
earthquakes occur on weak pre2existing faults rather than initiating new faults. The magnitude range is
M S > 610 ( 1636～1983) and M W > 515 ( 1983～2003) . Our catalogue results from a compilation of
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
Earthquake Research in China
412
data sources including the historical catalogue of Comninakis and Papazachos ( 1986) , the instrumental
catalogue of Makropoulos , et al . ( 1989 ) , Papazachos and Papazachou ( 2003 ) , the NOA catalogue
(www. gein. noa. gr) , the work of Papadopoulos , et al . , ( 2003) , Fokaefs and Papadopoulos ( 2004)
and the work of Louvari , et al . , ( 1999 ) . The data are presented in the Tables 1 and Table 2 ,
respectively.
The dataset is subject to the following errors : time of event , location of event and magnitude of
event . First , the timing of strong events in the Ionian Sea is well constrained from many studies so our
data is suitable to determine the recurrence interval . Second , all events are located offshore because
the historical and instrumental descriptions do not mention surface ruptures. To determine the
magnitude we used two types of measurements that describe the size of an earthquake : ( a ) for
historical events the earthquake Intensity —a measure of the degree of earthquake shaking at a given
locality based on the amount of damage and ( b) for instrumental events the Magnitude —an estimate of
the amount of energy released at the source of the earthquake. The drawback of intensity scales is that
destruction may not be a true measure of the earthquakeπs actual severity. The amount of structural
damage attributable to earthquake vibrations depends on ( a ) intensity and duration of the vibrations ,
(b) nature of the material upon which the structure rests and ( c ) design of the structure. Another
drawback for historical events is the lack of consideration of the directivity effect , i . e. , fault raptures
in certain directions , along which seismic energy is more focused.
In order to group the northern (Lefkada) events into the two sets shown in Fig. 1 we also took into
account the evidence provided by the latest strong earthquake , a magnitude M W = 612 event offshore
the island of Lefkada that occurred on August 14 , 20031 The maximum intensity has been evaluated I0
= VIII ( EMS) at Lefkas municipality ( Papadopoulos , et al . , 2003) , while VI to VII + intensities
were evaluated at many other villages of the island. The offshore NNE2SSW oriented strike2slip right2
lateral fault was activated by the main shock ( Karakostas , et al . , 2004 ) . The most characteristic
macroseismic effects were extensive typical ground failures like rock falls , soil liquefactions ,
subsidence , compaction , ground cracks and landslides. These macroseismic effects are remarkably
similar to those reported from some historical Lefkada shocks ( Papadopoulos , et al . , 2003 ; Pavlides ,
et al . , 2004) e. g. during 1704 , 1914 ( M S = 613) and 1948 ( M S = 615) .
Table 1 Lefkas Offshore Seismicity since 1704
Date
1704211222
1722206205
1723202222
1741206223
1769210212
1783203223
1783206207
1815
1820202221
1825201219
1869212228
1914211227
1948204222
1948206230
1994202225
2003208214
Intensity
IX
VIII
IX
IX
X2LN
X2LS
VIII2LN
VIII
IX
X2LN
X2LN
IX
IX2LS
IX2LN
VIII2LN
Magnitude
613
614
616
616
617
617
613
612
614
615
614
613
615
614
515
612
Segment
NLS
SLS
NLS
SLS
NLS
SLS
NLS
NLS
NLS
NLS
NLS
NLS
SLS
NLS
NLS
NLS
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
Volume 21 , Number 4
413
Table 2 Kefallinia offshore seismicity since 1636 , related to the strike2slip plate boundary
Date
Intensity
Magnitude
Segment
1636209230
X
712
SKS
1658208201
X
619
NKS
1714208228
IX
616
SKS
1759206214
IX
615
NKS
1766207224
IX
617
SKS
1767207211
X
712
NKS
1862203214
IX
617
NKS
1867202204
XI
713
NKS
1972209217
613
NKS
1981206228
610
SKS
1983201217
617
SKS
1987202227
517
NKS
1996202201
515
SKS
3 METHOD
We empirically determine the lower limit M min of earthquake magnitudes for every fault according
to the relative intensity of seismicity. In the present study we take M min = 5151 Otherwise , to discern
the main behaviors of strong earthquake recurrences , the foreshocks and aftershocks are not treated as
independent events.
The whole earthquake recurrence behaviors of these two faults are analyzed in two different ways.
The first way does not consider the time2predictable or slip2predictable behaviors. The coefficient
of variation σ( the ratio of the standard deviation ∑ of recurrence time intervals to the average value
Tm of various cycle intervals) , however , is used to measure the variability of recurrence times. For
every fault , the coefficient of variation σof earthquake recurrence intervals is calculated.
Σ
σ=
Tm
The parameter σ reflects the complexity of the time process. The larger σ is , the more
complicated the recurrence time process is , and the more dispersive the distribution of recurrence time
intervals is. Investigations have shown that if σ< 1 , the recurrence presents quasi2periodic behavior ;
if σ > 1 , the recurrence shows clustering behavior ; and if σ = 1 , the recurrence behaves like
completely random Poisson process ( Kagan and Jackson , 1991) . However , generally speaking , the
case of σ< 1 doesnπt mean that the recurrence behavior of earthquakes is of a good quasi2periodicity.
But only when σν 1 , it behaves as quasi2periodic recurrence behavior in favor of long2term prediction
(Bakun and McEvilly , 1984) . If σ< 1 and σ is nearly equal to 1 , the recurrence is close to random
behavior.
The second way considers the time2predictable or slip2predictable behaviors of the earthquake
recurrence process. For every fault , two regression functions , corresponding to these two types of
behaviors respectively , are established as follows :
( 1)
ln T = a + bMπ
( 2)
Mλ = A + B ln T
T is the time interval between two successive events. Mπ and Mλ are the magnitude of the prior
and latter events respectively. a , b and A , B are all regression coefficients.
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
Earthquake Research in China
414
If function ( 1) is satisfied , the recurrence behavior is time2predictable ( Shimazaki and Nakata ,
1980 ; Papazhacos , 1989 ) . Assuming a constant rate of strain accumulation and constant fault
strength , Shimazaki and Nakata ( 1980) concluded that the time interval between two successive large
earthquakes is approximately proportional to the amount of co2seismic displacement of the preceding
earthquake and not of the following earthquake. After an earthquake with larger co2seismic
displacement , it will take a longer time for the next earthquake to occur. If the amount of slip during
the past earthquake is known , the time of the next earthquake can be predicted.
If function ( 2 ) is satisfied , the recurrence behavior is slip2 ( or magnitude2) predictable
( Shimazaki and Nakata , 1980 ; Wen Xueze , 1999) . In this case , both the recurrence time and the
amount of time elapsed since the previous shock is taken into account . These behaviors are compatible
with the seismic gaps hypothesis which suggests that the probability for a future earthquake to fill up
the gap is small immediately following the occurrence of a characteristic earthquake and increases with
the time elapsed since the previous event . These are distinct from the time2independent behaviors
( i . e. , Poisson distributions ) in that the latter ones use only information concerning recurrence
intervals and do not take into account the amount of time elapsed since a prior event . These behaviors
cannot be used to predict when an earthquake will occur , but they can be used to predict the
magnitude of the earthquake that would occur at any given time. The potential fault slip at any time is
proportional to the shear stress on the fault . Thus , if the time of the last earthquake is known , the
magnitude of the next earthquake can be determined at any particular time.
According to whether the correlation coefficient r is larger than or equal to the critical value rα
with a given confidence degreeα, the recurrence behaviors of time2predictable or slip2predictable can
be determined. The confidence degreeα is usually taken as 01051
4 ANALYSIS OF THE WHOL E RECURRENCE BEHAVIORS
By use of the methods described above , we analyzed these two faults in the Ionian Sea. From
Table 3 , we can see that the earthquake recurrence behavior of Fault Lefkada is near2random with a
variation coefficient σ= 01824 and that of Fault Kefallinia presents clustering behavior with a variation
coefficient σ= 11191 Yi Guixi , et al . ( 2003 ) found that the greater the number of segments of one
fault is , the more complex the whole earthquake recurrence process of this fault is. As the variance
coefficient σof Fault Kefallinia is larger than that of Fault Lefkada , the number of segments of Fault
Kefallinia may be greater than that of Fault Lefkada.
According to the calculated correlation coefficient r and the judgments of the recurrence behaviors
( Table 3) , for Fault Lefkada , both time2predictable and slip2predictable recurrence behaviors are not
existent . For Fault Kefallinia , however , the time2predictable recurrence behavior is existent , but the
Table 3 Earthquake recurrence parameters and behavior judgments
Fault
name
( number
of events)
Function ln T = a + bMπ ,
Variance coefficients σ
The correlation coefficient r
and average recurrence
and judgment of time2
intervals Tm
predictable behavior
Function Mλ = A + B ln T ,
The correlation coefficient r
and judgment of slip2
predictable behavior
Lefkada
(16)
σ= 01824
Tm = 19192
r = 0106
| r| < rα= 0. 05 = 0. 514
Not existent
r = - 0140
| r| < rα= 0. 05 = 0. 514
Not existent
Kefallinia
(13)
σ= 1119
Tm = 29194
r = 0166
| r| > < rα= 0. 05 = 0. 576
Existent
r = - 01005
| r| < < rα= 0. 05 = 0. 576
Not existent
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
Recurrence
behavior
Near2random
Clustering
http://www.cnki.net
Volume 21 , Number 4
415
slip2predictable recurrence behavior is not existent . It means that for Fault Kefallinia , one can make
earthquake prediction to a certain extent according to the magnitude of the prior strong earthquakes.
Fig. 2 shows the statistical distribution map of earthquake recurrence time intervals of these two
faults. In order to erase the effect of different average earthquake recurrence time intervals of different
faults , we took a normalized earthquake recurrence time interval TΠTm as a statistical sample
(Nishenko and Buland , 1987) , which is the ratio of an exact recurrence time interval T to average
recurrence time interval Tm . In this figure , the distribution trend is that short recurrence time intervals
have a high probability. In Fig. 2 ( b) , the statistical map of Fault Kefallinia indicates a second peak at
3 times the average recurrence time interval , which suggests that the earthquake recurrence time
intervals of Fault Kefallinia obviously alternate between relatively active periods and relatively quiet
periods. If σ > 1 , such as it is for Fault Kefallinia , the second peak in the statistical distribution
map moves obviously to the right with the increase of the σ value. It suggests that the quiet periods are
longer than the active periods and the recurrence behavior shows a true clustering. Studies show that
whole earthquake clustering of an active fault zone is the result of a process of different fault segments
breaking successively in a relatively short period ( Perkins , 1987) . The process is also the rupture
triggering process between segments.
Fig. 2
Statistical distribution of earthquake recurrence time intervals
(σ is variance coefficient and n is the number of recurrence time intervals)
( a) Fault Lefkada ; ( b) Fault Kefallinia
For these two faults , we also calculated the normalized earthquake recurrence time intervals of
relatively quiet periods ( TqΠTqm ) and relatively active periods ( TaΠTam ) . Ta and Tq are relatively
quiet and relatively active recurrence time intervals respectively. Tqm and Tam are average intervals
respectively. Fig. 3 shows the probability density distribution of relatively active recurrence time
intervals ( Fig. 3a) and relative quiet recurrence time intervals ( Fig. 3b) of these two faults. Several
different probability density functions ( PDFs) are used to fit the statistical distribution.
It is easy to see in Fig. 3 that the time interval distribution of active periods is more dispersive
than that of quiet periods. Within active periods , it is proper to use the Weibull PDF with two
parameters to accurately fit this distribution , while not the negative exponential PDF. The probability
density function of Weibull is :
0,
x ≤0
p ( x) =
(λ > 0 , α > 0)
α- 1
α
α
λx exp ( - λx ) , x > 0
We estimated the parameters α= 111231 , λ= 019535 with the method of moment . The Weibull
( stretched exponential ) PDF has often been applied to recurrence statistics ( Hagiwara , 1974 ;
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
Earthquake Research in China
416
Fig. 3
Statistical distribution of earthquake recurrence time intervals of relatively active
periods ( a) and relatively quiet periods ( b)
Rikitake , 1976 , 1982 ; Utsu , 1984 ) , and was firstly introduced into the study on earthquake
recurrence as early as 1974 ( Hagiwara , 1974 ) . It has proven useful in a broad variety of similar
applications , particularly lifetime modeling. For using mathematical treatments of failure time can
simulate the rupture time of the earthπs crust , it is successfully applied to reliability considerations of
the earthπs crust with respect to earthquake occurrences. The Weibull is characterized by its steadily
increasing hazard function for a range of parameter values. Rikitake ( 1975 ) analyzed crustal
movements associated with an earthquake and showed that ultimate strain to rupture of the earthπs crust
can be well approximated by a Weibull distribution. In recent years a numerical simulation2based study
showed the Weibull can describe the simulation data substantially better than other distributions , not
only the distribution of interval times , but also the failure of a group of fault segments interacting by
means of elastic stress transfer ( Rundle , et al . , 2005) .
2
Within quiet periods , we tried to use normal distribution ( a = 1100 , σ = 010747) , logarithmic
2
2
normal distribution ( a = - 01036 , σ = ( 01268 ) ) and Weibull (α = 411171 , λ = 016718 )
distribution to depict the probability density distribution. It is not easy to say which function is more
like the true distribution due to a lack of samples.
From the analysis above , due to the complexity of earthquake occurrence processes , it is
impossible to give an exact middle2 and long2term earthquake prediction and seismic hazard assessment
for a whole active fault zone. However it is possible for us to estimate the probability according to the
characteristic of the distribution. If additional evidences can be acquired , the efficiency of the
probabilistic forecast can be improved , and whether an active fault zone is in its seismically active
period or in a quiet period can be identified.
5 DISCUSSION AND CONCL USIONS
We examined the whole strong earthquake recurrence behaviors of two fault zones along the
Kefallinia Transform , Ionian Sea , Greece. The fault zones are Fault Lefkada and Fault Kefallinia. Our
results can be summarized as follows.
( 1) For Fault Lefkada , both time2predictable and slip2predictable recurrence behaviors are not
existent . For Fault Kefallinia , however , the time2predictable recurrence behavior is existent , but the
slip2predictable recurrence behavior is not existent . It means that for Fault Kefallinia , one can make
earthquake prediction to a certain extent according to the magnitude of prior strong earthquakes.
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
Volume 21 , Number 4
417
(2 ) The recurrence behavior of Fault Lefkada is near random while Fault Kefallinia shows
clustering behavior possibly due to static stress transfer ( Stein , et al . , 1997 ; Papadimitriou , 2002) .
A physical basis for this difference may be that Fault Lefkada is at the end of the plate boundary where
segmentation may not be so well developed as along Fault Kefallinia. It is possible that several smaller
structures exist that help release part of the energy stored along the plate boundary ( this has to be
tested by future seismicity studies) . A second reason may be the mechanical interaction of Fault
Lefkada with thrust faults of the collisional plate boundary further north. Fault Lefkada is located at the
end of this major , dextral strike2slip system where deformation is taking place northwards by crustal
shortening along the Greek and Albanian coasts.
(3) The seismicity of the two faults alternates between relatively active periods and relatively
quiet periods. The time intervals distribution of active periods is more dispersive than that of quiet
periods , and it is proper to use the Weibull PDF with parameters α = 111231 , λ = 01953 to accurately
fit this distribution. However , the distribution of quiet periods can be depicted either by the normal
distribution , the logarithmic normal distribution or the Weibull distribution.
ACKNOWL ED GEMENTS
This study is part of the results from the study on the earthquake recurrence behaviors of the main
seismogenic fault zones in China and Greece , which belongs to a framework of Sino2Greece scientific
cooperation project , the Ministry of Science of Technology of the Peopleπs Republic of China and the
General Secretariat for Research and Technology of Greece. The authors thank George Stavrakakis and
Spyros Pavlides for discussions. George Papathanassiou provided the historical earthquake catalogue for
Table 11
REFERENCES
Bakun W. H. , McEvilly T. V. Recurrence models and Parkfield , California , earthquakes [J ] . J . Geophys . Res . ,
1984 , B89 :3051～30581
Benetatos Ch. , Kiratzi A. , Roumelioti Z. , Starrakakis G. , Drakatos G. and Latousakis J . The 14 August 2003
Lefkada Island ( Greece ) earthquake : focal mechanisms of the main shock and of the aftershock sequence [J ] .
Journal of Seismology , 2005 , 9 :171～190.
Comninakis P. E. , Papazachos B. C. A Catalogue of Earthquakes in Greece and the Surrounding Area for the Period 1901
～1985[M] . Univ. Thessaloniki Geohys. Lab. , 1986.
Fokaefs A. , Papadopoulos G. A. Historical earthquakes in the region of Lefkada Island , Ionian Sea estimation of
magnitudes from epicentral intensities[J ] . Bulletin of the Geological Society of Greece , 2004 , 36 :1389～13951
Goes S. D. B. Irregular recurrence of large earthquakes : an analysis of historic and paleoseismic catalogs [J ] . J .
Geophys . Res . , 1996 , 101 (B3) :5739～57491
Hagiwara Y. Probability of earthquake occurrence as obtained from a Weibull distribution analysis of crustal strain[J ] .
Tectonophysics , 1974 , 23 : 313～3181
Jacoby , G. C. , Sheppard P. R. , Sieh K. E. Irregular recurrence of large earthquakes along the San Andreas fault :
Evidence from trees[J ] . Science , 1988 , 241 :196～1991
Kagan Y. Y. , Jackson D. D. Seismic gap hypothesis : Ten years after[J ] . J . Geophy . Res . , 1991 , 96 (B13) :21 ,419
～21 ,4311
Karakostas V. G. , Papadimitriou E. E. , Papazachos C. B. Properties of the 2003 Lefkada , Ionian Islands , Greece ,
earthquake seismic sequence and seismicity triggering[J ] . Bull . Seism . Soc . Am . , 2004 , 94 (5) :1976～19811
Kiratzi A. A. , Langston C. A. Moment Tensor Inversion of the 1983 January 17 Kefallinia event of Ionian Islands
( Greece) [J ] . Geophysical Journal International , 1991 , 105 :529～5351
Kokinou E. , Vafidis A. Seismic modeling of marine reflection data from Ionian Sea [ J ] . Journal of the Balkan
Geophysical Society , 2003 , 6 (1) :21～361
Le Pichon X. , Chamot2Rooke N. , Lallemant S. , et al. Geodetic determination of the kinematics of central Greece with
respect to Europe. Implications for Eastern Mediterranean tectonics[J ] . J . Geophy . Res . , 1995 , 100 :12675～
126901
Louvari , E. , Kiratzi , A. A. , Papazachos , B. C. The cephalonia transform fault and its extension to western Lefkada
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
418
Earthquake Research in China
Island ( Greece) [J ] . Tectonophysics , 1999 , 308 :223～2361
Makropaulos C. , Drakopoulos J . and Latoussakis J . A revised and extended earthquake catalogue for Greece since 1900
[J ] . Geophys . J . Int . , 1989 , 99 :305～306.
Mercier , J . , Sorel , D. , Simeakis , K. Changes in the state of stress in the overriding plate of a subduction zone : the
Aegean Arc from the Pliocene to the present[J ] . Annales Tectonicae , 1987 , 1 :20～391
Nishenko S. P. , Buland R. A generic recurrence interval distribution for earthquake forecasting[J ] . Bull . Seism . Soc .
Am . , 1987 , 77 :1382～13991
Papadimitriou E. E. Mode of strong earthquake recurrence in the central Ionian Islands ( Greece) : possible triggering due
to Coulomb stress changes generated by the occurrence of previous strong shocks[J ] . Bull . Seism . Soc . Am . ,
2002 , 92 (8) :3293～33081
Papadimitriou P. , Kariris G. and Makropoulos K. The MW = 613 2003 Lefkada earthquake ( Greece) and induced stress
transfer changes[J ] . Tectonophysics , 2006 ,423 :73～82.
Papadopoulos G. A. , Karastathis V. K. , Ganas A. , et al. The Lefkada , Ionian Sea ( Greece) , Shock ( Mw 612) of 14
August 2003 : Evidence for the Characteristic Earthquake from Seismicity and Ground Failures[J ] . Earth Planets
Space , 2003 , 55 (11) :713～7181
Papazachos B. C. A time2predictable model for earthquake generation in Greece [J ] . Bull . Seism . Soc . Am . , 1989 ,
79 :77～841
Papazachos B. C. Seismicity of the Aegean and surrounding area[J ] . Tectonophysics , 1990 , 178 :287～3081
rd
Papazachos B. , Papazachou C. The Earthquake of Greece (3 edt. ) [M] . Thessaloniki : Ziti Public , 2003. P. 286.
Papazachos C. , Kiratzi A. A detailed study of the active crustal deformation in the Aegean and surrounding area [J ] .
Tectonophysics , 1996 , 253 :129～1541
Pavlides S. B. , Papadopoulos G. A , Ganas A. P. , et al. The 14 August 2003 Lefkada ( Ionian Sea) earthquake. In :
Proceedings of the 5 th International Symposium on Eastern Mediterranean Geology , Thessaloniki , Greece , 2004 ,
14220 April 2004 , T52341
Perkins D. M. Contagious fault rupture , probabilistic hazard , and contagion observability. In : Directions in
Palaeoseismology (A. J . Crone and E. M. Omdahl eds. ) , U. S. Geol. Surv. Open File Report 8720673 , Denver
: Federal center , 1987 , 428～4391
Rikitake T. Statistics of ultimate strain of the earthπs crust and probability of earthquake occurrence[J ] . Tectonophysics ,
1975 , 26 :1～211
Rikitake T. Recurrence of great earthquakes at subduction zones[J ] . Tectonophysics , 1976 , 35 :335～3621
Rikitake T. Current Research in Earthquake Prediction I , Reidel Publ. Co , Dordrecht , The Netherlands , 1982 , 3831
Rundle J . B. , Rundle P. B. , Donnellan A. , et al. A simulation2based approach to forecasting the next great San
Francisco earthquake. Proc . Nat . Acad . Sci . , 2005 , 102 :15 ,363～15 ,3671
Savage J . C. , Cockerham R. S. Quasi2periodic occurrence of earthquakes in the 1978～ 1986 Bishop2Mammoth lakes
sequence , eastern California[J ] . Bull . Seism . Soc . Am . , 1987 , 77 (4) :1347～13581
Scordilis E. M , Karakaisis G. F. , Karacostas B. G. , et al. Evidence for transform faulting in the Ionian Sea 2 the
Cephalonia island earthquake sequence of 1983[J ]1 Pure and Applied Geophysics , 1985 , 123 (3) :388～3971
Serpelloni E. , Anzidei M. , Baldi P. , et al. Crustal velocity and strain2rate fields in Italy and surrounding regions : new
results from the analysis of permanent and non2permanent GPS networks[J ] . Geophys . J . Int . , 2005 , 161 ( 3) :
861～8801
Shimazaki K. , Nakata T. Time2predictable recurrence model for large earthquakes. Geophys Res Lett . , 1980 , 7 :279～
2821
Sorel D. Etude neotectonique dam Iπarc Pgeen externe occidental : les iles ioniennes de Kephallinia et Zakinthos , et
IπElide occidentale. These 3e cycle , Lab. de Gtol. Dyn. Interne. Bit. 5091 Centre Universitaire , 91 504 Orsay ,
France. 19761
Stein R. S. , Barka A. A. and Dieterich J . H. Progressive failure on the North Anatolian fault since 1939 by earthquake
stress triggering[J ] . Geophysical Journal International ,1997 , 128 :594～604.
Taymaz T. , Jackson J . , McKenzie D. Active tectonics of the north and central Aegean Sea [J ] . Geophys . J . Int . ,
1991 , 106 :433～4901
Utsu T. Estimation of parameters for recurrence models of earthquakes[J ] . Bull Earthquake Res lnst Univ Tokyo , 1984 ,
59 :53～661
Wen Xueze. Recurrence behaviors of segment2rupturing earthquakes on active faults of the Chinese mainland[J ] . Acta
Seismologica Sinica , 1999 , 12 (4) : 457～4651
Yi Guixi , Wen Xueze , and Xu Xiwei. Study on integrated recurrence behaviors of strong earthquakes along entire active
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net
Volume 21 , Number 4
419
fault zones in the Sichuan2Yunnan region[J ] . China. Earthquake Research in China , 2003 , 17 (2) :145～1591
Yi Guixi and Wen Xueze. Earthquake recurrence on whole active fault zones and its relation to that on individual fault2
segments[J ] . Acta Seismologica Sinica , 2000 , 13 (5) :563～5741
About the Author
Qi Cheng , born in 1979 , graduated in solid geophysics from the Earth and Space Sciences
Department of the University of Science and Technology of China in 2001 , and is a post doctorate in
the Institute of Geology and Geophysics , Chinese Academy of Sciences. He got his doctorπs degree in
the same institute in 20061 He mainly works on seismic tomography and seismic activity. E2mail :
qicheng @mail . iggcas. ac. cn
© 1994-2008 China Academic Journal Electronic Publishing House. All rights reserved.
http://www.cnki.net