Content uploaded by Christoforos Benetatos
Author content
All content in this area was uploaded by Christoforos Benetatos on Jan 17, 2018
Content may be subject to copyright.
Strike-slip motions in the Gulf of Siğaçik (western Turkey):
Properties of the 17 October 2005 earthquake seismic sequence
Christoforos Benetatos
a
, Anastasia Kiratzi
a,
⁎, Athanassios Ganas
b
, Maria Ziazia
b
,
Areti Plessa
b
, George Drakatos
b
a
Department of Geophysics, Aristotle University of Thessaloniki, 54124, Thessaloniki, Greece
b
National Observatory of Athens, Geodynamics Institute, Athens 118 10, Greece
Received 15 March 2006; received in revised form 25 July 2006; accepted 3 August 2006
Available online 26 September 2006
Abstract
The October 2005 series of earthquakes that occurred in the Gulf of Siğaçik (western Turkey) reveal the operation of pure
strike-slip faults, as evidenced from the 49 focal mechanisms we determined, in a region dominated by N–S extension and bounded
by well-documented graben structures. The sequence is characterized by the occurrence of three moderate size events (17 October
2005, 05:45 UTC, Mw 5.4; 17 October 2005, 09:46 UTC, Mw 5.8; and 20 October 2005, 21:40 UTC, Mw 5.8) with an eastward
propagation and close spatial separation (b6 km). We relocated over 200 aftershocks, combining phases from the Greek and
Turkish seismological networks, which align roughly in a NE–SW cloud, but considerably spread after the first day of the
sequence, indicating the simultaneous activation of multiple structures nearly orthogonal to the main rupture. It is hard to relate the
occurrence of the events to any of the previously mapped faults in the region. The region of occurrence is a well-known geothermal
area which implies that it is in a very unstable state, with the fault systems close to rupture and very sensitive to stress perturbations.
Here we showed that the sequence is adequately explained by static stress triggering. It is worth noting that this sequence, though
moderate in magnitudes, provides stronger evidence for the operation of sub-parallel strike-slip faults in the central Aegean Sea–
western Turkey, north of the volcanic arc, which seem to be optimally oriented in the regional stress field and facilitate the Anatolia
motion into the Aegean Sea.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Seferihisar; Siğaçik Gulf; Western Turkey; Aegean Sea; Focal mechanisms
1. Introduction
In October 2005 intense seismic activity burst in the
Gulf of Siğaçik(district of Seferihisar) in western Turkey
(rectangle in Fig. 1). The two strongest events of the
sequence both had magnitudes equal to Mw 5.8, masking
the identification of a classic mainshock, and occurred on
17 and 20 October, 2005 at 09:46 and 21:40 UTC, res-
pectively. These events were preceded on 17 October
2005 at 05:45 by a Mw 5.4 earthquake. The nearby Chios
and Samos Islands (Greece) and the cities of Izmir and
Urla in Turkey were shaken, but no life loss was reported.
Most of the damage was reported in the districts of
Seferihisar and Urla.
The region of occurrence of these events is famous for
its historical heritage, (Dewey and Sengör, 1979;
Tectonophysics 426 (2006) 263–279
www.elsevier.com/locate/tecto
⁎Corresponding author. Tel.: +30 2310 998486; fax: +30 2310 998528.
E-mail address: kiratzi@geo.auth.gr (A. Kiratzi).
0040-1951/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.tecto.2006.08.003
Papazachos and Papazachou, 2002; Altinok et al., 2005)
located in the western part of the Gediz and Menderes
graben systems, and contains several normal faults, very
prominent in the morphology, with an ∼E–Wstrike
(Dewey and Sengör, 1979). Of equal importance are the
NE–SW and NW–SE trending faults which additionally
play significant role on the tectonics (Yilmaz, 1997). In
terms of plate motion models, the region belongs to the
transition zone between two plates, of Anatolia plate in the
east and of the Aegean plate in the west (Papazachos,
1999; McClusky et al., 2000). South of the North
Anatolian Fault the plate motion of Anatolia relative to
Europe can be adequately described by a simple rigid
body rotation with a dominant E–W component and an
average velocity of ∼24 mm/yr (McClusky et al., 2000).
The westward escape of Anatolia is facilitated by the low
elevations in the Aegean Sea. The Aegean plate moves
almost uniformly in a SSW direction with an average
velocity of 35 mm/yr in the southern parts (McClusky et
al., 2000). The change of Anatolia rotation to the Aegean
translation occurs along the central and southern parts of
western Turkey (Papazachos, 1999). The stress field is
mainly extensional in a NNE–SSW direction combined
with considerable strike-slip motions (Kiratzi, 2002;
Koukouvelas and Aydin, 2002; Kiratzi and Louvari,
2003 and references therein).
Fig. 1. Main features of the Aegean Sea and surrounding lands. Dashed lines mark the eastern and northern edges of the Aegean plate (as in
Papazachos, 1999), and the rectangle indicates the region of occurrence of the 2005 in western Anatolia, located in the transition zone from the
Anatolia to the east to the Aegean plate to the west, while arrows indicate the motion of the plates relative to Eurasia (McClusky et al., 2000).
264 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
Here we study the characteristics of the 2005
sequence, focusing on the distribution of aftershocks
and of focal mechanisms in terms of the regional stress
field, we discuss about the main fault ruptures, and we
examine how the strongest events of the sequence
affected the stress field in view of the general consensus
that even small static stress changes due to the coseismic
displacement can trigger off-fault earthquakes (Harris,
1998 and references therein).
2. Regional tectonics
Fig. 2 summarizes the most prominent tectonic struc-
tures which are marked on morphology (from Ocakoğlu
Fig. 2. Focal mechanisms of previous strong events (Table 1), together with those of the strongest events of the 2005 sequence (Table 2). Main
structures clearly seen on morphology (from Ocakoğlu et al., 2004, 2005) are also plotted (KF = Karaburun Fault, UF = Urla Fault, TF = Tuzla Fault
and IF = Izmir Fault). The focal mechanisms show pure strike-slip motions in the Gulf of Siğaçik and normal faulting combined with considerable
strike-slip motion in the north (Karaburun peninsula) and in the south (near Samos Island).
Table 1
The parameters of the focal mechanisms of previous strong events in the area (CMT Harvard solutions)
No Date Time Lat°
N
Lon°
E
Depth
km
Mw Nodal Plane 1 Nodal Plane 2 Paxis Taxis
Yr/m/day Hh/mm/s Strike Dip Rake Strike Dip Rake Az Dip Az Dip
1 790614 11:44:46 38.74 26.50 15 5.8 253 59 −120 121 42 −50 112 63 4 9
2 790416 18:41:59 38.69 26.54 10 5.3 255 58 −124 127 45 −48 112 61 8 7
3 921106 19:08:10 38.09 27.19 25 6.0 238 85 −167 147 77 −5 103 13 12 6
4 940524 02:05:39 38.71 26.49 21 5.5 258 54 −135 138 55 −46 107 55 198 1
5 960402 07:59:24 37.92 26.67 15 5.4 262 41 −127 127 58 −62 87 65 198 9
6 971114 21:38:53 38.80 25.87 15 5.8 242 80 −156 148 67 −10 107 23 13 9
7 030410 00:40:16 38.17 26.76 15 5.7 250 76 −159 155 70 −15 114 24 22 4
8 030417 22:34:22 38.19 26.90 15 5.2 256 79 −139 156 50 −15 124 36 20 19
265C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
et al., 2004, 2005) together with the focal mechanisms of
previous events (see Table 1 for details). Close spatial
separation of strong contrasts in topography is observed,
with the roughly north-south Bozdag mountain range
(1212 m high), and the steep slopes falling directly into the
sea. Focusing in the Seferihisar region the most prominent
features are the Karaburun Fault (KF), the Urla Fault (UF),
the Izmir Fault (IF) and the Tuzla Fault (TF). Recently
(Ocakoğlu et al., 2004, 2005), based on multi-channel
reflection data, identified: a) KF as an active reverse
faulting, responsible for the raise of Karaburun peninsula;
b) UF as an N–S trending reverse fault; c) stressed out the
importance of strike-slip faulting in the region and d)
mapped the continuation of active structures off-shore
(also shown in Fig. 2 just to indicate the cross-cut pattern
of faulting). The Izmir Fault is a normal fault bounding the
southern Gulf of Izmir, and its activity has been well
documented with the occurrence of strong events (e.g. 178
(6.5), 1040 (6.8), 1654 (6.4), 1680 (6.2), 1688 (6.8), 1723
(6.4), 1778 (6.4); Papazachos and Papazachou, 2002). The
Tuzla Fault (∼40 km in length) is a dextral strike-slip fault
based on the offset of river channels and the focal
mechanisms of recent events (Ocakoğlu et al., 2005 and
references therein).
The most devastating earthquake in Urla district, from
the historical period, is reported on 15 October 1883;
Mw 6.8 (38.4° N, 26.6° E) which produced the destruc-
tion of 3600 houses and the death of 120 people in Urla
and Çesme (Papazachos and Papazachou, 2002). For the
instrumental period previous seismic activity in the
broader Seferihisar region includes the earthquakes of 16
November 1992 (Mw 6.0) and of 10 April 2003 (Mw
5.7) which was followed on 17 April 2003 with an Mw
5.2 event. Both the 1992 (Scordilis et al., 1994) and 2003
events are connected with right-lateral strike-slip
motions, based on the distribution of aftershocks which
align in an ENE–WSW direction. The more distant, to
the region of study, events, north of Karaburun peninsula
Fig. 3. Location of the stations whose records we used a) to locate epicentres (all stations) and b) to invert for the focal mechanism solutions (only
records from NOA broad band stations marked with light coloured squares). (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)
266 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
and north of Samos Island, have focal mechanisms that
show normal faulting with strike-slip components, and
the north dipping planes are probably the fault planes in
all three cases (Papazachos and Papazachou, 2002).
3. Distribution of epicentres and focal mechanisms
of the 2005 sequence
To locate the epicentres we combined P and S phases
derived from the Greek and Turkish seismological
stations (Fig. 3), e.g. the networks operated by the
National Observatory of Athens (NOA) and by the
Kandilli Observatory of Turkey. To determine the focal
mechanisms of the strongest events of the sequence we
used regional waveforms as recorded from the broad
band sensors of NOA network.
3.1. Distribution of epicentres
We hand-picked P and S phases from the digital records
of 3-component instruments from the NOA network,
covering the period 17October to 22 November 2005. We
added the phases from the Turkish network as have been
provided to us. The average number of phases used was
29, which was adequate enough for the location process,
and we ended with 235 bestlocated events with 12 or more
phases. The closest seismological station of the NOA
network is located on Samos Island (code SMG) at a
distance of ∼50 km from the sequence, while the closest
station from the Turkish network is BLCB at a distance of
∼35 km, contributing significantly to focal depth esti-
mates. We used HYPOELLIPSE (Lahr, 1999) and after
several tests with regional models we preferred to use the
velocity model of Panagiotopoulos (1984) which has as
follows: V
g
=6 km/s, d
g
=19 km; V
b
=6.6 km/s,
d
b
=12 km, overlying a half-space with V
n
=7.9 km/s
(Fig. 4). This model has been tested in the past and has
been extensively used for the routine earthquake locations
in the broader Aegean Sea region. Here, it produced the
smallest uncertainties in the locations compared to the
model of Novotny et al. (2001) which we prefer to use for
moment tensor inversions, based on previous experience
(Roumelioti et al., 2003).
Fig. 5 (a to d) shows the distribution of the best located
events, (distribution of location uncertainties also shown),
and the evolution of the sequence in selected time win-
dows. In the first 5 h following the occurrence of the 17
October Mw 5.4 earthquake, (Fig. 5a) the few aftershocks
form a small cluster trending NE, where in its easternmost
end the second strong Mw 5.8 event occurred almost 4 h
later. (Fig. 5a). Approximately 19 h after the occurrence of
the first event (Fig. 5b) the sequence develops in a NE–SW
trend, while the activation of other structures orthogonal to
this main trend is evident. The same pattern is also ob-
served during the first 3 days (Fig. 5c), with the last strong
event to occur again at the easternmost end of the activated
area. The entire sequence (Fig. 5d) indicates that the ac-
tivity is concentrated along the northern margin of the Gulf
of Siğaçik, the trend of the aftershocks is NE–SW, and the
activation of nearby small faults is evident. This pattern of
off-fault aftershock activity is more clearly seen in the
work of Aktar et al. (submitted for publication) who have
installed a local network to monitor aftershocks. The
accuracy of their locations clearly shows the activation of
two main structures, one trending NW–SE bounding the
western coast of the Gulf of Siğaçik and the other trending
NE–SW, along the central coast of the Gulf. These two
aftershock clouds are orthogonal to each other, forming
conjugate strike-slip faulting.
The first Mw 5.4 event has a clearly defined directivity
towards ENE (Benetatos et al., in preparation), which
implies that the NE–SW nodal plane is the fault plane, and
the sense of strike-slip motion is right lateral. The two
Mw5.8 events occurred very close in space, have very
similar focal mechanisms (Table 2) and their waveforms
are strikingly similar (Fig. 6). Based on the directivity of
the first event of the sequence, the relocation of the three
strongest events, the evolution of the sequence mainly
along a NE–SW trend and the regional tectonics, we
conclude that the 2005 sequence implies the activation of
Fig. 4. Velocity models used for epicenter location (model of
Panagiotopoulos, 1984) and for the regional waveform modelling
(Novotny et al., 2001).
267C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
aENE–WSW trending dextral strike-slip fault. With the
previous reasoning we also connect the three strongest
events with the same rupture (ellipse in Fig. 5) which for
an Mw 5.8 event the along strike-length is of the order of
10 km and the width is ∼6.0 km (Wells and Coppersmith,
1994).The plethora of aftershocks observed off the main
aftershock cloud can be treated as off-fault aftershocks,
usually observed in aftershock sequences. Actually this
part of western Turkey is a well-documented geothermal
area (Drahor and Berge, 2006) and these areas are known
to be unstable regions, with a fault system in a mechanical
equilibrium state very close to rupture, as suggested by
their high sensitivity to transient stress perturbations. This
fact probably explains the diffuse pattern of aftershock
activity and the simultaneous activation of many struc-
tures. The fault mapping of Ocakoğlu et al. (2004, 2005)
already suggests a complex conjugate net of faults.
3.2. Distribution of focal mechanisms
We used the regional moment tensor inversion tech-
nique (Dreger and Helmberger, 1990, 1991, 1993)to
determine the focal mechanisms for 49 events of the
sequence. The method is routinely applied now and we
do not go into detailed analysis (the reader can refer to
Dreger, 2002 and references therein). Green's functions
Fig. 5. Evolution of aftershock activity (circles) in selected time windows, in relation with the occurrence of the strongest events (stars) of the 2005
sequence, its approximate dimensions marked by the ellipse (for a Mw 5.8 earthquake). a) For the first 4 h until the occurrence of the 17 October
09:45, M 5.8 event b) of ∼the first day of 17 October, c) For 17 to 20 October 2005, d) All one-month aftershocks. Note the simultaneous activation
of off-fault structures within the first 24 h. The uncertainties in epicentre locations are also shown in the lower part (Root Mean Square (RMS),
Horizontal (ERH) and vertical (ERZ)).
268 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
were computed using the FKRPROG code (Saikia,
1994) using the velocity model of Novotny et al. (2001),
(see also Fig. 3), which have been previously tested and
proved adequate for most paths in the Aegean Sea
(Benetatos et al., 2002, 2005). Original waveforms were
cut into segments long enough to ensure resolution of
Table 2
Parameters of the focal mechanisms of the October 2005 sequence that we determined using regional moment tensor inversion applied to broad band
waveforms
No Date Time Lat° N Lon°
E
Depth
km
Mw NVR
(%)
Nodal Plane 1 Nodal Plane 2 Paxis Taxis
Yr/m/day hh/mm/s Strike° Dip° Rake° Strike° Dip° Rake° Az° Dip° Az° Dip°
1 051017 04:31:27 38.150 26.632 10 3.9 4 71 157 70 −25 256 67 −158 116 31 207 2
2 051017 05:45:19 38.153 26.620 15 5.4 10 78 156 85 −8 247 82 −175 111 9 202 2
3 051017 06:16:07 38.168 26.660 12 3.9 2 69 156 74 −31 255 60 −162 112 33 208 9
4 051017 07:05:49 38.158 26.629 18 4.1 5 74 172 80 −17 265 73 −170 128 19 219 5
5 051017 07:49:52 38.164 26.615 6 3.6 1 47 166 58 −95 355 32 −82 62 76 259 13
6 051017 08:07:30 38.166 26.653 16 3.8 3 70 175 80 −12 267 78 −170 131 16 221 1
7 051017 08:28:54 38.177 26.654 18 4.3 6 76 161 81 −24 255 67 −170 116 23 210 10
8 051017 08:34:45 38.138 26.632 11 4.2 4 70 152 85 −43 247 47 −173 100 33 207 25
9 051017 08:50:04 38.166 26.640 5 3.4 1 35 221 85 −165 130 75 −5 86 14 355 7
10 051017 09:46:57 38.178 26.663 7 5.8 7 92 136 81 −11 228 79 −171 92 14 182 1
11 051017 09:55:32 38.147 26.635 21 5.2 4 63 164 85 −17 255 73 −175 118 15 211 8
12 051017 10:57:31 38.140 26.640 5 3.7 3 59 327 83 −10 59 80 −173 283 12 13 2
13 051017 11:20:27 38.180 26.684 6 3.9 4 77 145 82 −16 237 74 −172 100 17 192 6
14 051017 11:48:48 38.187 26.705 16 3.7 1 61 351 82 71 239 21 157 97 34 240 49
15 051017 12:02:23 38.152 26.753 5 3.4 1 62 141 86 18 49 72 176 274 10 6 16
16 051017 12:05:42 38.132 26.673 7 3.2 1 41 47 79 −24 142 66 −168 2 25 96 9
17 051017 12:09:54 38.175 26.681 4 3.5 2 63 153 81 −37 250 54 −168 105 32 207 18
18 051017 12:22:33 38.136 26.674 7 3.7 4 67 239 85 164 330 74 5 286 8 193 15
19 051017 12:32:03 38.146 26.657 6 3.2 1 56 251 72 −143 148 55 −22 115 39 16 11
20 051017 12:34:02 38.392 27.011 10 3.6 1 67 142 81 −39 239 52 −169 93 33 196 19
21 051017 12:43:30 38.146 26.650 8 4 3 88 319 83 10 227 80 173 93 2 183 12
22 051017 13:16:43 38.179 26.657 5 3.9 2 96 151 85 −29 244 61 −174 104 24 201 16
23 051017 13:22:51 38.173 26.638 11 3.8 4 59 155 83 −8 246 82 −173 110 11 201 1
24 051017 14:53:22 38.126 26.689 17 3.4 1 45 38 90 12 128 78 0 84 8 352 8
25 051017 23:13:17 38.154 26.642 10 3.9 5 78 249 88 174 340 84 2 295 3 204 6
26 051018 05:04:52 38.134 26.664 5 3.5 2 43 52 74 −164 317 75 −41 274 22 5 1
27 051018 16:00:49 38.174 26.635 22 4.2 4 66 167 82 35 71 55 103 294 18 35 30
28 051018 22:49:28 38.159 26.705 19 3.9 4 65 77 81 171 168 81 164 123 0 33 13
29 051019 05:51:24 38.129 26.612 17 3.6 2 51 176 88 −19 267 71 −86 130 15 223 12
30 051019 10:11:31 38.181 26.670 15 4.6 8 79 244 89 178 334 88 1 289 1 199 2
31 051019 10:22:34 38.168 26.632 7 3.7 2 63 332 86 36 239 54 175 100 21 202 28
32 051020 09:10:53 38.157 26.717 19 4 4 73 71 89 −179 341 89 −1 296 1 206 0
33 051020 21:40:04 38.183 26.694 7 5.8 7 92 133 73 −25 231 66 −162 90 30 183 5
34 051021 00:34:15 38.112 26.608 20 3.8 3 64 353 90 19 83 71 0 40 13 306 13
35 051021 11:47:38 38.178 26.599 10 4.3 7 77 244 88 −175 154 85 −2 109 5 19 2
36 051021 16:34:40 38.200 26.839 13 3.7 4 59 44 81 162 137 72 10 92 6 359 19
37 051022 07:21:03 38.157 26.765 12 3.7 4 50 151 87 −17 242 73 −176 105 14 198 10
38 051022 15:35:26 38.184 26.628 19 3.7 3 63 80 89 179 170 89 1 125 0 35 1
39 051022 18:00:08 38.185 26.560 20 3.6 2 46 338 82 46 240 45 169 101 24 210 37
40 051022 19:05:07 38.149 26.715 18 3.5 2 40 341 86 −10 72 80 −176 296 10 27 4
41 051023 14:59:38 38.110 26.594 17 4 7 80 79 88 171 169 81 2 124 5 34 8
42 051024 16:55:39 38.155 26.573 12 3.6 4 60 169 87 17 78 73 177 302 10 35 14
43 051024 17:03:17 38.092 26.525 18 3.6 2 60 239 81 142 336 53 11 293 18 191 32
44 051024 21:15:38 38.174 26.607 18 3.9 4 79 155 89 −4 245 86 −179 110 4 200 2
45 051025 08:58:27 38.178 26.702 20 3.8 3 75 163 86 23 71 67 175 295 13 29 19
46 051026 17:48:08 38.127 26.578 13 3.8 4 69 349 89 −17 79 73 −179 303 13 35 11
47 051029 14:48:42 38.128 26.610 14 4.2 4 70 162 70 −25 261 67 −158 121 31 212 2
48 051031 05:26:40 38.138 26.644 16 4.9 5 83 231 75 156 328 67 17 281 5 188 27
49 051031 06:48:22 38.127 26.650 12 4.2 4 79 340 88 −370 87−178 295 4 25 1
Nrefers to the number of stations used for the calculation of the focal mechanism and VR (%) is the total variance reduction of the inversion.
269C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
Fig. 6. Broad band records at stations (shown in Fig. 3) for the two strongest Mw 5.8 events of 17 October 2005 (09:46 UTC) (dashed lines) and of 20
October 2005 (21:40 UTC) (straight lines). All records are filtered between 0.02 and 0.1 Hz. The similarity of waveforms is striking both in shape and
amplitude which favours their close location in space, their magnitude equality, and their focal mechanism similarity that we obtained. The
comparison to the waveforms of the first Mw 5.4 was not as good, implying a different location and mechanism.
270 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
Fig. 7. Moment tensor inversion results (dashed line = synthetics, continuous lines = observed waveforms) obtained for the three strongest events of
the sequence as marked. On the left part the variance reduction is presented for each station and on the right part the strike, dip and rake of the solution
along with the seismic moment and the percentage of double couple (DC) and the percentage of the Compensated Linear Vector Dipole (CLVD).
271C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
the low frequencies, carefully examined to enclose only
one event in each data segments and avoid interference
between two or more consequently earthquakes that
occurred very close in time. Then, the waveforms were
corrected for the instrument response, re-sampled at a
sampling rate of 1 Hz, integrated to displacement and
pass band filtered between 0.05 and 0.08 Hz. For
smaller magnitude events we sometimes filtered be-
tween 0.05 and 0.1 Hz to allow for higher frequencies.
The optimum depth for each event was found with a grid
search (in the range of 2 to 30 km with a step of 2 km)
and the selection was based on the variance reduction
that the inversion routine was returning at each run and
the percent double couple of the focal mechanism. We
expect from a reliable solution to have high variance
reduction and high percentage of double couple force
acting at the source. The focal mechanism solutions
proved to be stable during the tests at different depths
(we indicatively show in Fig. 7 the solutions for the
three strongest events), although it was necessary in
many cases to manually align the synthetics relative to
the observed waveforms in order to achieve the best fit.
When a satisfactory solution was obtained we were
refining our depth search with a step of 1 km. The
method of moment tensor inversion can be used with
single station records (Dreger, 2002). We tested this, for
Fig. 8. Examples of single (SMG) and two-station inversion (SMG and PRK) in order to show the stability of the focal mechanism when using few
stations. At the upper part of each plot the focal mechanism obtained with all available stations is presented and just below the focal mechanism using
SMG or SMG and PRK for various depths with its corresponding variance reduction (VR%).
272 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
six earthquakes, using only one or two close stations
(SMG or SMG and PRK). In all cases the similarity of
our final solution with the one obtained using single or
two stations records in the inversion was proved (Fig. 8).
Moreover, from the stations used in the inversions
(Fig. 3), SMG and PRK at distances of 50 km and
125 km, respectively from the centre of the seismic
sequence, provided very stable solutions, further valid-
ating the velocity model used.
Earthquake focal mechanisms (Table 2 for para-
meters) with magnitudes ranging from 3.2 ≤Mw≤5.8,
clearly show (Fig. 9) extensive strike-slip motions
along the entire activated area. The average Paxis
trends at N104±22° E and the average T-axis trends at
N200± 24° E in accordance with the regional stress
field (Kiratzi, 2002; Kiratzi and Louvari, 2003).
Strike-slip motions were also observed by Melis and
Konstantinou (2006) in a set of 15 focal mechanisms
of the sequence that they determined applying moment
tensor inversion.
4. Coseismic Coulomb stress changes
Strong earthquakes can trigger earthquakes at short
distances from the epicentre by transferring static or
dynamic stresses (e.g. Harris et al., 1995; Caskey and
Wesnousky, 1997; Harris and Simpson, 1998; Gomberg
et al., 2001; Ganas et al., 2005a). The earthquake induced
stress changes are applied to neighbouring receiver faults
(astermedinReasenberg and Simpson, 1992) using the
Coulomb Failure Function:
DCFF ¼DsþlVðDrnÞð1Þ
where Δτis the coseismic change in shear stress on the
receiver fault and in the direction of fault slip, Δσ
n
is the
Fig. 9. Distribution of the focal mechanisms here determined (numbers as in Table 2). The operation of steeply dipping strike-slip faulting in the
region is evident.
Table 3
Slip models for the source faults of the three strongest events used in the stress changes calculations
# Event name; Time Epicentre Lon° N, Lat° E Length (km) Width (km) Strike (°) Dip (°) Rake (°) U
s
(m) U
d
(m) Mw
1 17.10.2005; 05:45 UTC 38.153, 26.620 6 5 247 82 −175 −0.139 0.012 5.4
2 17.10.2005; 09:46 UTC 38.178, 26.663 10 6 228 79 −171 −0.275 0.043 5.8
3 20.10.2005; 21:40 UTC 38.183, 26.694 10 6 231 66 −162 −0.264 0.086 5.8
Fault length and width according to Wells and Coppersmith (1994).U
s
and U
d
are the average strike-slip component and the average dip slip
component, respectively.
273C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
change in normal stress acting on the receiver fault (with
tension positive), and μ′is the effective coefficient of
friction,
lV¼lð1−Dp=DrnÞð2Þ
where μis the static coefficient of friction and Δpis the
pore pressure change within the fault.
To resolve stress on individual fault planes we used
the code DLC (written by R. Simpson, based on the
subroutines of Okada (1992)), to calculate changes in
the stress tensor at points along a specified receiver fault
surface caused by slip on a source fault in an elastic half
space. Table 3 summarizes the fault slip models we use
(as previously obtained) extending from mid-upper
crustal levels to a depth of 15 km and Table 4 gives the
values of the parameters used for the stress change
calculations. All source faults are modelled as, oblique-
slip, inclined rectangular dislocations, ignoring local
fault complexities.
At first we evaluated the triggering capability of the
source fault of the first Mw 5.4 event (17-10-2005
05:45 UTC), and we calculated the amount of Coulomb
stress which was transferred to the region of occurrence
of the 17-10-2005 and 20-10-2005 events. The target
(receiver) planes are dextral strike-slip faults with slip
models identical to that of the 17-10-2005 09:46 UTC.
We sampled ΔCFF (Coulomb stress change between the
initial stress and the final stress) on a horizontal section
at 7 km depth on a 100×100 km grid surrounding the
source event epicentre, with 1 km grid spacing. The
depth of horizontal plane was chosen to comply with the
hypocentre depth of the two Mw5.8 events of 17
October and 20 October 2005. We used the program
ELFGRID to calculate the stress tensor grid at 7 km
depth. Then we applied the program STROP which uses
that tensor to calculate the tractions at that depth on
planes of specified orientation. STROP outputs a ΔCFF
file that does the calculation in Eq. (1) above on the
planes of interest for the friction value specified. We
interpret a positive value of ΔCFF to indicate that a fault
plane occurring within this stress lobe has been brought
closer to failure; when ΔCFF is negative, the fault is
brought further from failure (i.e. relaxed).
For the Mw 5.4 event and its triggering capability we
used this method to additionally examine if the NE–SW
plane is the fault plane, further to support our directivity
observations. Thus, we examined two source models, at
two different depths (considering the uncertainty in the
depth estimate): a) a NW–SE trending rupture plane
associated with left lateral strike-slip motion at a depth of
10 km (Fig. 10a) and of 15 km (Fig. 10b) and b) a NE–SW
trending rupture associated with right lateral strike-slip
motion at a depth of 10 km (Fig. 10c) and of 15 km
(Fig. 10d). Both models predict that the epicentres of two
Mw 5.8 events are in the regions where stress was
transferred. The shallower sources (10 km) for the Mw 5.4
event predict positive stress levels ΔCFFb0.25 bars for a
NW–SE trending plane (Fig. 10a) whereas for a NE–SW
trending plane the loading levels are considerably larger
and of the order of 0.25 to 0.75 bars (Fig. 10c). This is an
indication for the operation of the NE–SW plane, but not
strong evidence, taking into account that ΔCFF
levelsN0.1 bar are associated with earthquake triggering
on neighbouring faults (Harris et al., 1995) and that a
triggering threshold has not yet been established (e.g. Ziv
and Rubin, 2000). For the deeper sources (at 15 km depth)
the stress loading levels (Fig. 10b, d) are smaller by a
factor of three, as compared to the shallow sources. The
hypocentre depth of the Mw 5.4 event is crucial for the
triggering hypothesis if a triggering threshold exists at
∼+0.1 bars, that is why we examined ΔCFF at two depth
ranges. From the rupture models examined for the first
event we prefer the one of NE–SW rupturing (Fig. 10c),
because it is in accordance with the directivity pattern, the
Table 4
Input parameters used for stress transfer modelling
Poisson's ratio 0.25
Shear modulus, μ3×10
10
Pa (3×10
5
bars)
Map projection UTM zone 35
Depth of ΔCFF calculation 7 km (Fig. 10), 12 km (Fig. 11)
Grid size 1 km
Effective Friction Coefficient (μ′) 0.4 (based on Harris and
Simpson, 1998).
Target (Receiver) Planes: NE–SW
trending associated with
right lateral strike-slip motion
Strike/dip angle-direction/rake 228/79NW/-171
Fig. 10. Coulomb stress changes at 7-km depth associated with the 17-10-2005 (05:45 UTC); Mw 5.4 earthquake. Palette of stress values is linear in
the range −2 to + 2 bar. White colour indicates area where transferred stress N2 bar and black colour the area where stress reduction was N−2 bar. Stars
show the three mainshock epicentres and ellipses their estimated location error. Target fault planes are also shown as straight lines. Blue areas indicate
unloading, red areas indicate loading, respectively. Effective coefficient of friction is 0.4 in all cases. Slip modelsare shown in Table 3. Colour scale in
bar (1 bar =100 KPa). a) Source slip model assuming rupture of a NW–SE plane with sinistral strike-slip motion at a hypocentre depth of 10 km; b) as
in a) but at a hypocentre depth of 15 km; c) Source slip model assuming rupture of a NE–SW plane with dextral strike-slip motion at a hypocentre
depth of 10 km; d) as in c) but with hypocentral depth at 15 km. In b) and d) note the reduction of transferred stress near the two strongest Mw
5.8 epicentres by a factor of three.
274 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
275C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
distribution of aftershocks, and it clearly predicts that the
epicentres of the two Mw 5.8 events, taking into account
their epicentre uncertainties, are clearly in the regions
where stress was most increased.
We have to comment that, as the epicentre of the
20.10.2005 Mw 5.8 event is located 2.7 km to the east of
the 17.10.2005 Mw 5.8 event, our static stress transfer
modelling could not document triggering because this area
is relaxed after the combined stress field calculations of the
first two events. Perhaps the uniform-slip model adopted
here is inadequate to simulate the actual rupture process
and associated stress transfer in this case because slip
irregularities can cause local peaks in stress.However, we
note that a positive lobe where ΔCFFN1 bar exists 2 km to
the east of the error margin of the 20.10.2005 epicentre, no
matter which rupture combination we adopt for the first
two events.
Fig. 11 shows the stress change throughout the crust,
combining the effects of the three strongest events of the
sequence (Table 3), along planes oriented to maximize the
Coulomb stress change (Stein et al., 1992), at a depth of
12 km (average depth of all aftershocks that we analysed).
For the calculations we assume the regional stress field at
200 bars (20 MPa) consisting of uniaxial E–W compres-
sion (T-axis azimuth at N200° E). The theoretical optimal
failure angle for a horizontal axis of maximum compres-
sion is 34.1° for μ′=0.4. Coulomb stress increase of
∼5 bars is predicted in lobes following a NE–SW trend,
well explaining the occurrence of the two strongest events.
We also observe that there is a satisfactory, positive corre-
lation between sites of aftershocks and stress increase.
Many aftershocks were triggered at distances of 20–30 km
away from the Gulf of Siğaçik where ΔCFFb0.5 bar. The
“stress shadow”effect is also visible to the north and to the
Fig. 11. Stress change calculated on optimal planes to regional compression (remote stress at 200 bars). Palette of stress values is linear in the range
−5 to +5 bar. Blue areas indicate unloading, red areas indicate loading, respectively. Circles represent aftershocks after 20 October 2005
(21:40 UTC). The occurrence of the two Mw5.8 events is well documented in the loaded regions.
276 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
south of the Gulf of Siğaçik where ΔCFFb−1.5 bar. A
few aftershocks occur inside these relaxed areas and
within a distance of b10 km from the epicentres of the
three strongest events. We infer that our stress change
calculations have been adequately validated by the ob-
served distribution of aftershocks.
5. Discussion and conclusions
The moderate-size earthquake sequence of October
2005 occurred in western Turkey (Gulf of Siğaçik) close
to the Greek islands of Samos and Chios. The sequence is
interesting because it indicates rupture of dextral strike-
slip faults in a region dominated by ∼N–S extension and
well developed mainly east-west trending normal faults.
Three were the strongest events of the sequence. The first
Mw 5.4 event occurred on 17 October 2005 (05:45 UTC)
at the western tip of Gulf of Siğaçik, and was followed 4 h
later (09:46 UTC) by an Mw 5.8 event, and 3 days later,
on 20 October by another Mw 5.8 event, both occurring to
the east of the first event, along the central part of Siğaçik
Gulf.
We combined phases from the Greek and Turkish
networks to relocate more than 200 aftershocks. A general
Fig. 12. Map showing the orientation of Pand Taxes in the Aegean Sea and the surrounding lands. Focal mechanisms are also plotted with beach-ball
sizes that scale with earthquake magnitude. Note the distributed strike-slip motions in the central Aegean Sea that facilitate the transfer of Anatolia
motion to the normal faulting system of continental Greece. The question marks denote areas, which still lack focal mechanisms and where future data
are expected to shed light on the seismotectonics (figure modified from Kiratzi and Louvari, 2003).
277C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
alignment of the best located epicentres along a general
NE–SW trend is observed, however a characteristic of
this sequence is the considerable spread of aftershocks
implying simultaneous activation of other structures in the
region that mainly have a NW–SE trend (also observed by
Aktar et al., submitted for publication), resembling the
compound earthquake sequence that occurred in Novem-
ber 1987, near Superstition Hills in California (Hudnut
et al., 1989; Scholz, 1994). Such a complex fault structure
has been identified in this area on the basis of reflection
data (Ocakoğlu et al., 2004, 2005 and references therein).
Moreover, the presence of geothermal fields in the region
(Drahor and Berge, 2006), and the unstable condition of
the fault systems they imply, probably explains the diffuse
pattern of aftershock activity off the main rupture, shortly
after the occurrence if the first two strong events.
We used moment tensor inversion applied to regional
broad band records to determine the focal mechanisms
of 49 events of the sequence. All (except one) of the
focal mechanisms are connected with pure strike-slip
motions along very steep planes, extending in depth
from 4 km to 22 km.
Based on the similarity of waveforms, the distribution
of aftershocks, the clear directivity towards ENE of the
first event (Benetatos et al.,2006 in preparation), the stress
transfer analysis and the previous knowledge of regional
tectonics we conclude that all three strongest events of the
2005 sequence ruptured a steeply dipping right lateral
strike-slip fault trending NE–SW with an average
mechanism of strike 235°, dip 75° and rake −170°. The
event propagation towards east is well explained by the
stress transfer modelling.
Fig. 12 presents the distribution of focal mechanisms
in the broader Aegean Sea area from a database we
constantly update (figure modified from Kiratzi and
Louvari, 2003), together with the direction of Pand T
axes. If we focus our attention in the central Aegean Sea,
north of the volcanic arc, we observe that as new data
are collected there is evidence of distributed strike-slip
motions along mainly NE–SW striking planes exhibit-
ing right lateral sense of motion. This distributed strike-
slip faulting, with slip vectors parallel to the GPS
vectors, facilitates the westward motion of Anatolia
plate. These strike-slip faults often sub-parallel, operate
as transfer fault systems that connect two extensional
regimes: the extensional regime of western Anatolia, in
the east, and the extensional regime of continental
central Greece, in the west. In this pattern the existence
of NW–SE striking strike-slip faults is also evident.
Evidence for the operation of this NW–SE striking
sinistral strike -slip faults in the Aegean Sea was first
derived from the occurrence of the July 2001 sequence,
north of Skyros Island (Roumelioti et al., 2003, 2004;
Ganas et al., 2005b). The sequence studied here,
intensifies the operation of strike-slip faulting in the
central Aegean Sea–western Turkey, in regions that are
well-away from the North Anatolian Fault system,
indicating that strike-slip motions are the most optimal
to the stress-field to transfer the motion from Anatolia
into continental Greece.
Acknowledgments
Thanks are due to two anonymous reviewers and the
editor for their constructing comments. Dogan Kalafat,
from Kandilli Observatory, is gratefully thanked for his
immediate response in providing phase data from the
Turkish network. We also thank our friends and
colleagues Mustafa Aktar, and Hayrullah Karabulut,
from Kandilli Observatory, for providing results prior to
publication and useful discussions. George Karakaisis
and Spyros Pavlides, from the Department of Geology of
the Aristotle University of Thessaloniki, are also thanked
for stimulating discussions. This work was financed in
part by the General Secretariat of Research and
Technology (Ministry of Development) of Greece and
in part (C.B.) by the Ministry of Education and Religious
Affairs (Project Pythagoras II). Most of the figures were
plotted using the GMT software (Wessel and Smith,
1995). The software RAKE (Louvari and Kiratzi, 1997)
was also used to handle the focal mechanism database.
References
Aktar, M., Karabulut, H., Ozalaybey, S., Childs, D.M., submitted for
publication. A conjugate strike-slip fault system within the
extensional tectonics of western Turkey.
Altinok, Y., Alpar, B., Özer, N., Gazioglu, C., 2005. 1881 and 1949
earthquakes at the Chios–Cesme Strait (Aegean Sea) and their
relation to tsunamis. Nat. Hazards Earth Syst. Sci. 5, 717–725.
Benetatos, C., Roumelioti, Z., Kiratzi, A., Melis, N., 2002. Source
parameters of the M 6.5 Skyros island (North Aegean Sea)
earthquake of July 26, 2001. Ann. Geophys. 45, 513–526.
Benetatos, Ch., Kiratzi, A., Roumelioti, Z., Stavrakakis, G., Drakatos,
G., Latoussakis, I., 2005. The 14 August 2003 Lefkada Island
(Greece) earthquake: focal mechanisms of the mainshock and of
the aftershock sequence. J. Seismol. 9, 171–190.
Caskey, S.J., Wesnousky, S.G., 1997. Static stress changes and earthquake
triggering during the 1954 Fairview peak and Dixie valley earth-
quakes, central Nevada. Bull. Seismol. Soc. Am. 87 (3), 521–527.
Dewey, J.F., Sengör, A.M.C., 1979. Aegean and surrounding regions:
complex multiple and continuum tectonics in a convergent zone.
Geol. Soc. Amer. Bull. 90, 84–92.
Drahor, M., Berge, M., 2006. Geophysical investigations of the Sefer ihisar
geothermal area, western Anatolia, Turkey. Geothermics 35, 302–320.
Dreger, D., 2002. Manual of the Time-Domain Moment Tensor Inverse
Code (TDMT_INVC), Release 1.1. Berkeley Seismological
Laboratory, p. 18.
278 C. Benetatos et al. / Tectonophysics 426 (2006) 263–279
Dreger, D., Helmberger, D., 1990. Broadband modelling of local
earthquakes. Bull. Seismol. Soc. Am. 80, 1162–1179.
Dreger, D., Helmberger, D., 1991. Complex faulting deduced from
broadband modelling of the 28 February 1990 Upland earthquake
(ML=5.2). Bull. Seismol. Soc. Am. 81, 1129–1144.
Dreger, D., Helmberger, D., 1993. Determination of source
parameters at regional distances with single station or sparse
network data. J. Geophys. Res. 98, 8107–8125.
Ganas, A., Shanov, S., Drakatos, G., Dobrev, N., Sboras, S., Tsimi, C.,
Frangov, G., Pavlides, S., 2005a. Active fault segmentation in
southwest Bulgaria and Coulomb stress triggering of the 1904
earthquake sequence. J. Geodyn. 40 (2–3), 316–333.
Ganas, A., Drakatos, G., Pavlides, S., Stavrakakis, G., Ziazia, M.,
Sokos, E., Karastathis, V., 2005b. The 2001 Mw =6.4 Skyros
earthquake, conjugate strike-slip faulting and spatial variation in
stress within the central Aegean Sea. J. Geodyn. 39, 61–77.
Gomberg, J., Reasenberg, P.A., Bodin, P., Harris, R.A., 2001.
Earthquake triggering by seismic waves following the Landers
and Hector Mine earthquakes. Nature 411, 462–466.
Harris, R., 1998. Introduction to special section: stress triggers, stress
shadows, and implications for seismic hazard. J. Geophys. Res.
103, 24,347–24,358.
Harris, R.A., Simpson, R.W., 1998. Suppression of large earthquakes
by stress shadows: a comparison of Coulomb and rate-and-state
failure. J. Geophys. Res. 103, 24439–24451.
Harris, R.A., Simpson, R.W., Reseanberg, P.A., 1995. Influence of
static stress changes on earthquake locations in southern
California. Nature 375, 221–224.
Hudnut, K.W., Seeber, L., Pacheco, J., 1989. Cross-fault triggering in
the November 1987 Superstition Hills earthquakes sequence,
Southern California. Geophys. Res. Lett. 16, 199–202.
Kiratzi, A., 2002. Stress tensor inversions along the westernmost north
and central Aegean Sea. Geophys. J. Int. 106, 433–490.
Kiratzi, A., Louvari, E., 2003. Focal mechanisms of shallow earthquakes
in the Aegean Sea and the surrounding lands determined by
waveform modelling: a new database. J. Geodyn. 36, 251–274.
Koukouvelas, I., Aydin, A., 2002. Fault structure and related basins of
the North Aegean Sea and its surroundings. Tectonics 21 (5), 1046,
doi:10.1029/2001TC901037.
Lahr, J.C., 1999. HYPOELLIPSE: A Computer Program for
Determining Local Earthquake Hypocentral Parameters, Magni-
tude and First-Motion Pattern, Y2K Version, USGS Open-File
Report 99-23.
Louvari, E., Kiratzi, A., 1997. Rake: a Window's program to plot
earthquake focal mechanisms and stress orientation. Comput.
Geosci. 23, 851–857.
McClusky, S., Balassanian, S., Barka, A., Demir, C., Georgiev, I.,
Hamburg, M., Hurst,K., Kahle, H., Kastens, K., Kekelidze, G.,King,
R., Kotzev, V., Lenk, O., Mahmoud, S., Mishin, A., Nadariya, M.,
Ouzounis, A., Paradissis, D., Peter, Y., Prilepin, M., Reilinger, R.,
Sanli, I.,Seeger, H., Tealeb, A.,Toksoz, M.N.,Veis, G., 2000. Global
Positioning System constraints on plate kinematics and dynamics in
the eastern Mediterranean and Caucasus. J. Geophys. Res. 105,
5695–5720.
Melis, N., Konstantinou, K., 2006. Real-time seismic monitoring in the
Greek region: an example from the 17 October 2005 east Aegean
Sea earthquake sequence. Seismol. Res. Lett. 77, 364–370.
Novotny, O., Zahradnik, J., Tselentis,G.-A., 2001. North-western Turkey
earthquakes and the crustal structure inferred from surface waves
observed in Western Greece. Bull. Seismol. Soc. Am. 91, 875–879.
Ocakoğlu, N., Demirbağ, E., Kuşçu, İ., 2004. Neotectonic structures in
the area off shore of Alaçati, Doğanbey and Kuşadasi (western
Turkey): evidence of strike-slip faulting in the Aegean extensional
province. Tectonophysics 391, 67–83.
Ocakoğlu, N., Demirbağ, E., Kuşçu, İ., 2005. Neotectonic structures in
İzmir Gulf and surrounding regions (western Turkey): evidences of
strike-slip faulting with compression in the Aegean extensional
regime. Mar. Geol. 219, 155–171.
Okada, Y., 1992. Internal deformation due to shear and tensile faults in
a half-space. Bull. Seismol. Soc. Am. 82, 1018–1040.
Panagiotopoulos, D., 1984. Travel time curves and crustal structure in
the southern Balkan region. PhD thesis, University of Thessalo-
niki, 159 pp. (in Greek).
Papazachos, C.B., 1999. Seismological and GPS evidence for the
Aegean–Anatolia interaction. Geophys. Res. Lett. 26 (17),
2653–2656.
Papazachos, B.C., Papazachou, C., 2002. The Earthquakes of Greece.
Ziti Publ., Thessaloniki, Greece. 317 pp.
Reasenberg, P.A., Simpson, R.W., 1992. Response of regional
seismicity to the static stress change produced by the Loma Prieta
earthquake. Science 255, 1687–1690.
Roumelioti, Z., Kiratzi, A., Melis, N., 2003. Relocation of the July 26,
2001 Skyros Island (Greece) earthquake sequence using the
double-difference technique. Phys. Earth Planet. Inter. 138,
231–239.
Roumelioti, Z., Kiratzi, A., Dreger, D., 2004. The source process of the
July 26, 2001 Skyros Island (Greece) earthquake. Geophys. J. Int.
156 (3), 541–548.
Saikia, C.K., 1994. Modified frequency-wavenumber algorithm for
regional seismograms using Filon's quadrature; modelling of Lg
waves in eastern North America. Geophys. J. Int. 118, 142–158.
Scholz, C., 1994. The Mechanics of Earthquakes and Faulting.
Cambridge Univ. Press, p. 439.
Scordilis, E., Kiratzi, A., Panagiotopoulos, D., 1994. A study of the
earthquake of November 6, 1992 in Izmir (Turkey). Bull. Geol.
Soc. Greece XXX/5, 243–249.
Stein, R.S., King, G., Lin, J., 1992. Change in failure stress on the
southern San Andreas fault system caused by the 1992
Magnitude=7.4 Landers earthquake. Science 258, 1328–1332.
Wells, D.L., Coppersmith, K.J., 1994. New empirical relationships
among magnitude, rupture length, rupture width, rupture are, and
surface displacement. Bull. Seismol. Soc. Am. 84, 974–1002.
Wessel, P., Smith, W.H.F., 1995. New version of the Generic Mapping
Tools released. Eos, Trans. - Am. Geophys. Union 76, 329.
Yilmaz, Y., 1997. Geology of Western Anatolia. In: Schindler, C.,
Pfister, M. (Eds.), Active Tectonics of NW Anatolia —the
Marmara Poly-Project. VDF, ETH, Zurich.
Ziv, A., Rubin, A.M., 2000. Static stress transfer and earthquake
triggering: no lower threshold in sight? J. Geophys. Res. 105 (B6),
13631–13642.
279C. Benetatos et al. / Tectonophysics 426 (2006) 263–279