Aftershock Activity of the 1999 I˙zmit, Turkey, Earthquake Revealed

Bulletin of the Seismological Society of America, 92, 1, pp. 418–427, February 2002
Aftershock Activity of the 1999 İzmit, Turkey, Earthquake Revealed
from Microearthquake Observations
by Akihiko Ito, Balamir Üçer, Şerıf Barış, Ayako Nakamura, Yoshimori Honkura, Toshio Kono,
Shuichiro Hori, Akira Hasegawa, Rıza Pektaş, and Ahmet Mete Işıkara
Abstract
An accurate aftershock distribution of the 1999 İzmit, Turkey, earthquake was obtained by using the data from a local seismic network, IZINET, and 10
temporary seismic stations. More than 2000 aftershocks were relocated for the period
of about 2 months following the mainshock. From this aftershock distribution we
obtained several pieces of information on the characteristics of the mainshock. First,
the mainshock initiated fault rupture from a place adjacent to an active swarm area
where many microearthquakes had been occurring for more than 20 yr prior to the
mainshock. Second, the aftershock region extended in the east–west direction along
the North Anatolian Fault Zone (NAFZ). This confirms that the mainshock was
caused by a slip on the NAFZ. Third, the western end of the rupture caused by the
mainshock is likely to have reached up to about 29.2⬚ E in the İzmit Bay, and hence
the total length of the fault rupture caused by the mainshock amounts to about 150
km, as long as the estimate of the fault rupture length is based on the aftershock
distribution. This information is important for the discussion on the possibility of
future large earthquakes in the west of the source region of the İzmit earthquake. We
also found a clear tendency that aftershocks occur in clusters, which implies strong
heterogeneity in both the rupture process and the medium along the fault zone.
Introduction
NET seismic network for microearthquake observations
(Üçer et al., 1985), this network could not cover the whole
seismic gap region. According to previous studies, microearthquake activity in the seismic gap region has been very
low in general, except for some swarm areas (Crampin et
al., 1985; Evans et al., 1987; Tsukuda et al., 1988; Lovell
et al., 1989; Nishigami et al., 1990; Iio et al., 1991), but
detailed characteristics have not been very clear without a
dense network covering the seismic gap region.
In view of this, we have attempted, since 1992, to install
a routine seismic network, IZINET, in the westernmost part
of the NAFZ. IZINET was the only local seismic network
in operation when the 1999 İzmit earthquake occurred, and
it recorded the aftershock sequence following the mainshock. In this article, we present detailed features of the aftershock activity associated with the 1999 İzmit earthquake,
as revealed from microearthquake observations.
On 17 August 1999, a disastrous earthquake with a moment magnitude of Mw 7.4 (Kandilli Observatory, Boğaziçi
University; U.S. Geological Survey) took place near İzmit
City in the northwestern part of Turkey. The surface rupture
that appeared with the mainshock followed the North Anatolian Fault Zone (NAFZ) from the eastern end of İzmit Bay
to about 31⬚ E (Barka et al., 1999). The source region of this
earthquake has been well known as a seismic gap; in fact,
large earthquakes have occurred successively along the
NAFZ from the east to the west (Toksöz et al., 1979). In
this sense, this earthquake is the one that many seismologists
have been afraid of for a long time.
The NAFZ is an extremely active tectonic line, and its
tectonics and geology had been well investigated prior to
the 1999 İzmit earthquake (Allen, 1969; Ambraseys, 1970;
Barka, 1996; Ikeda, 1988). Large and medium earthquakes
have also been studied (Dewey, 1976; Jackson and McKenzie, 1988). We have also undertaken multidisciplinary
research in the western part of the NAFZ since 1981 in order
to increase our understanding of this seismic gap (Honkura
and Işıkara, 1991). Seismological observations, however,
have not been sufficient for monitoring microearthquake activity in the seismic gap region. Although the Kandilli Observatory, Boğaziçi University, has established the MAR-
Microearthquake Observations in the Western Part
of the NAFZ
Two major branches have been identified in the westernmost part of the NAFZ. One is the southern branch,
named the Iznik–Mekece Fault, which extends to the south418
419
Aftershock Activity of the 1999 İzmit, Turkey, Earthquake
west from the branching point at about 30.7⬚ E. The other
is the northern branch, the Sapanca–İzmit Fault, which runs
through the Sapanca Lake and İzmit City and extends to
İzmit Bay (Honkura and Işıkara, 1991). We installed a seismic network in 1992 (Barıs et al., 1996) to monitor microearthquakes along both the branches, as shown in Figure 1.
The network was named IZINET after the name of the city,
Iznik, where the Kandilli Observatory has established a
branch observatory for various geophysical observations.
All the stations of IZINET, except for the base station,
IZZ in Figure 1, are equipped with vertical-component seismometers of 1 or 2 Hz. Only the base station, located on the
top of a mountain for effective radio wave transmission, is
equipped with a 2-Hz three-component seismometer. Signals
are transmitted to the base station from each satellite station,
using a radio telemeter device, and are recorded in digital
form onto a PC-based data acquisition system. The number
of stations of IZINET was six initially, but this increased
year by year. When the 1999 İzmit earthquake occurred, 13
stations were in operation.
The IZINET covers only the western part of the seismic
gap region, and the surface rupture zone that appeared in
association with the mainshock extends further east beyond
the coverage of the IZINET. So we installed 10 temporary
seismic observation sites along the surface rupture immediately after the mainshock (Nakamura et al., 2002). Solid
triangles in Figure 1 show the locations of these temporary
stations. Each temporary station is equipped with threecomponent seismometers of 2 Hz, a digital data logger with
a Digital Audio Tape as a recording medium, a Global Positioning System clock, and a battery. The data logger can
record the ground motion continuously for about 3 weeks
without maintenance. We could install all the temporary observation stations by 24 August 1999. Therefore, the resolution and detectability of aftershocks were improved re-
markably after 24 August, especially for the eastern part of
the source region.
Crustal Structure and a Technique for Reliable
Hypocenter Determination
For hypocenter determination, we used a spherical Earth
model consisting of two layers corresponding to the crust
and mantle, respectively, with a gradual change in velocity
with depth within each layer. The parameters of the velocity
structure are shown in Table 1. Although we referred to the
recent studies on the crustal structure in the western part of
Turkey (Ergin et al., 1997; Horasan et al., 1997; Gürbüz et
al., 1998), none of them investigated the velocity structure
beneath the source region of the İzmit earthquake. In general, therefore, fairly large arrival-time residuals remain in
the hypocenter determination. Furthermore, arrival-time residuals at each station seem to show large deviations depending on the locations of the hypocenter. Large residuals
should be ascribed to uncertainty and spatial heterogeneity
in the seismic velocity structure. Therefore, we divided the
source region roughly into three parts, the western, the cen-
Table 1
Velocity Structure Model Used in the Present Study*
VP
VS
Layer
R ⳮ r0
V0
f
V0
f
Crust
Mantle
0 km
39
5.40 km/sec
8.00
40.0
2.3
3.12 km/sec
4.62
40.0
2.3
*The seismic velocity is represented by V(r) ⳱ V0(r/r0)ⳮf, where V(r)
is the seismic velocity at the radial distance r, r0 is the radius of the top
of each layer, and V0 is the seismic velocity at r ⳱ r0. In the table, R
represents the curvature radius of the earth at the latitude of 40⬚, that is
6361.1 km.
Figure 1. Distribution of the seismic observation stations in the westernmost part of
the NAFZ. Solid squares represent the seismic
stations of IZINET and solid triangles the temporary observation stations. Open circles show
some major cities referred to in the text. Solid
lines with a tick are active faults suggested by
Barka (1992, 1997).
420
A. Ito, B. Üçer, Ş. Barış, A. Nakamura, Y. Honkura, T. Kono, S. Hori, A. Hasegawa, R. Pektaş, and A. M. Işıkara
tral, and the eastern parts, and determined appropriate station
corrections for each part in order to avoid large residuals.
The average root mean square (rms) of residuals decreased
to 0.09 sec when this procedure was used, whereas it was
larger than 0.2 sec without these station corrections.
We also attempted to obtain a three-dimensional Pwave velocity structure in and around the source region of
the İzmit earthquake using data from IZINET, temporary
observations, and other temporary network installed by Iio
et al. (2002) and Nakamura et al. (2002). Although we obtained more accurate hypocenter distribution by applying the
three-dimensional velocity structure, the effect of the threedimensional velocity structure on the hypocenter relocation
was not significant. This is natural because we already took
the effect of the lateral heterogeneity of the medium into
account by using different station corrections in the three
regions of the source area. In this article, we use hypocenter
locations obtained by the procedure mentioned previously
for the simple calculation.
The events we refer to hereafter show an rms of arrivaltime residuals of 0.05 sec typically; in the worst case it is
0.3 sec. This means that the typical hypocenter error is 0.9
km, horizontally, and 1.5 km, vertically. By means of this
higher accuracy of hypocenter locations compared with
those of teleseismic studies, we can obtain detailed information about the seismicity before and after the 1999 İzmit
earthquake.
Seismic Activity Prior to the Mainshock
Figure 2 shows the distribution of microearthquakes in
the westernmost part of the NAFZ, as derived from IZINET
during 7 yr prior to the 1999 İzmit earthquake. Generally
speaking, the seismicity in this area has been low as mentioned previously, and the number of events for which hypocenters could be determined by IZINET is about 600 for
the 7-yr period. Figure 2 shows three areas of high microearthquake activity in this region. The most active area
is recognized near İzmit in the northern branch of the NAFZ.
This region has been known as an active swarm area since
1975 (Crampin and Üçer, 1975; Crampin et al., 1985; Üçer
et al., 1985; Evans et al., 1987). We can also recognize another swarm activity near Gemlik. Although this swarm is
not as active as the İzmit swarm, it is noteworthy because it
is located by the southern branch of NAFZ. On the other
hand, rather scattered seismic activity can be seen in the
south of the Iznik–Mekece Fault. In this area, a large earthquake with a magnitude of MS 5.0 occurred on 21 October
1983 (Barış et al., 2002). This would suggest the existence
of another active tectonic line.
Figure 3 shows a space–time plot of the microearthquakes for the region shown in Figure 2. Some quiescence
is apparent in Figure 3; this is obviously artificial and is
caused by a system-down of IZINET. In this figure, we can
clearly see that the swarm activity recognized in Figure 2
near the hypocenter of the 1999 İzmit earthquake has been
taking place rather continuously for a long time. Although
the activity is enhanced, for example in February and March
of 1997, it is hard to say from this figure whether there were
anomalous changes in the microearthquake activity prior to
the İzmit earthquake.
Precise Locations of the Mainshock and Aftershocks
When the 1999 İzmit earthquake occurred, a group from
our project happened to be carrying out magnetotelluric
measurements just across the source region of the mainshock
(Honkura et al., 2000). Their magnetometers and electric
sensors recorded large signals caused by strong ground motion of the mainshock. Unfortunately, IZINET did not record
the mainshock itself because the system happened to be on
the stored records transfer mode. Instead, we could determine the precise hypocenter of the mainshock by using the
electromagnetic signals recorded by our group together with
the routine data of the Kandilli Observatory.
The hypocenter was thus relocated at the end of the
İzmit Bay, as shown by an open star in Figure 2. Parameters
of the hypocenter are also listed in Table 2. The location of
this hypocenter is almost the same as that given in the announcement urgently issued by the Kandilli Observatory.
The uncertainty of the hypocenter is about 2 km horizontally
and 3 km vertically. Although the location of the hypocenter
depends on the velocity structure assumed, it does not deviate by more than 5 km even if we use another velocity
structure model. We can see from Figure 2 that the 1999
İzmit earthquake initiated fault rupture from a place adjacent
to the İzmit swarm area, as mentioned in the section Seismic
Activity Prior to the Mainshock. This would be one important piece of information on the potential risk assessment of
large earthquakes that may occur in the future.
Although IZINET missed the recording of the mainshock, it successfully recorded the aftershock sequence from
10 min after the mainshock. Figure 4 shows the aftershock
distribution until the end of September 1999. The number
of aftershocks for which the hypocenters were determined
during this period is about 2200. The aftershocks took place
approximately linearly along the NAFZ. This clearly indicates that the 1999 İzmit earthquake was caused by the slip
of the NAFZ. Aftershocks occurred in an approximately
170-km-long zone from about 29.2⬚ E to 31⬚ E in longitude.
Most of the aftershocks occurred at depths less than 20 km.
The lower limit of the focal depths is deeper in the eastern
part. Figure 5 shows the space-time plot of the aftershocks
projected on the east–west cross section. As shown in Figure
5, there are some portions where the aftershock data are
lacking, because of the malfunctioning of the acquisition
system. However, we could record the aftershock sequence
continuously for the first 36 hr after the mainshock.
Temporal Changes of Aftershock Activity
The distribution of the aftershocks shows a very simple
strike-slip faulting in the east–west direction. In detail, how-
Aftershock Activity of the 1999 İzmit, Turkey, Earthquake
Figure 2. Distribution of the microearthquakes observed by IZINET for the period
from 1993 to the occurrence of the 1999 İzmit earthquake. The star represents the
hypocenter of the 1999 İzmit earthquake obtained from the magnetotelluric data, together with seismic data provided by the Kandilli Observatory (Honkura et al., 2000).
Figure 3. Space–time plot of microearthquakes in the area shown in Figure 2 during
the period from 1993 to 1999. Horizontal bars represent periods when IZINET was
down for more than 1 month continuously.
421
422
A. Ito, B. Üçer, Ş. Barış, A. Nakamura, Y. Honkura, T. Kono, S. Hori, A. Hasegawa, R. Pektaş, and A. M. Işıkara
Table 2
The Hypocenter of the 1999 İzmit Earthquake Obtained from the Magnetotelluric Data
and the Routine Seismic Data at the Kandilli Observatory
Origin Time (UTC)
(dd/mm/yy)
17/08/99 0 hr 1 min 38.37 Ⳳ 0.53 sec
Latitude
(⬚)
Longitude
(⬚)
Depth
(km)
RMS*
(sec)
40.745 Ⳳ 0.023
29.961 Ⳳ 0.029
17.1 Ⳳ 3.4
0.13
*RMS is the root mean square of the arrival-time residuals.
Figure 4. Hypocenters of the aftershocks obtained by the end of September 1999.
The solid line represents the surface trace of fault rupture that appeared in association
with the mainshock.
ever, we can recognize several significant features in the
distribution of the aftershocks. The most marked feature is
that the aftershocks tend to occur in clusters. In order to
clarify the distribution of the aftershock in more detail, we
have shown the epicenter distribution for five periods in
Figure 6.
Period I represents the first 12 hr after the mainshock.
Aftershocks occurred mainly in four regions during this period, between (a) 29.2⬚ E and 29.5⬚ E, (b) 29.6⬚ E and 29.8⬚
E, (c) 29.9⬚ E and 30.1⬚ E, and (d) 30.2⬚ E and 30.7⬚ E in
longitude. Hereafter we will refer to these regions as (a), (b),
(c), and (d), respectively. In regions (a) and (b), aftershocks
occurred along a very narrow vertical plane. In region (c),
however, aftershocks were not distributed along a simple
east–west plane, but they rather seemed to align in the northeast–southwest direction. Among these three active areas
there lay clear gaps where seismicity was very low. Such a
feature suggests strong heterogeneity in the rupture process
of the mainshock.
For region (d), we must note that both the resolution
and detectability are not sufficient, as there were no observation stations in the eastern part during this period. In fact,
temporary observations had not yet started, and the easternmost IZINET station, GUL (see Fig. 1), was not in a proper
state of operation during the periods I and II because of
power cut in the earthquake area. Nonetheless, the characteristics of the aftershocks that occurred in region (d) were
quite different from those in regions (a)–(c). Aftershocks did
not occur along the surface rupture but were scattered
widely. Period II corresponds to 1 day reckoned from 12
a.m. of 17 August. The characteristics of the aftershock that
occurred were very similar to those of period I.
423
Aftershock Activity of the 1999 İzmit, Turkey, Earthquake
Figure 5.
Space–time plot of the aftershocks of the 1999 İzmit earthquake. Horizontal bars represent periods when IZINET was down for more than 1 day.
Period III consists of 2 days from 20 August. During
this period, a swarm activity started in the southwest of region (a). Hereafter we will refer to this swarm area as region
(e). Thereafter the number of aftershocks in regions (a) and
(b) decreased very quickly. In region (d), the center of activity seems to have migrated to the south and the number
of events along the surface rupture trace decreased as in
regions (a) and (b). We may therefore claim that aftershocks
did not take place along the surface rupture in region (d),
but were rather aligned to a preexistent active fault that was
broken by the 1967 Mudurnu earthquake (Ambraseys and
Zatopek, 1969). It should be noted here that the detectability
and the resolution in region (d) increased during this period
because the easternmost observation point, GUL, was resumed on 20 August. In this context, the aftershock distribution in region (d) must be considered carefully.
As the temporary observation stations were installed at
the beginning of period IV, the accuracy in the determination
of the hypocenters of the aftershocks, especially in region
(d), became much higher in period IV than in the previous
periods. The detectability of the aftershocks, however,
changed very little even during this period, as we analyzed
only the aftershocks that were detected by IZINET. Therefore, the number of events in region (d) actually increased,
compared with other regions. In region (d), the tendency of
the aftershocks to concentrate along a linear trend extending
in the southeast direction became clearer in period V than
in period IV.
From these temporal changes in the aftershock activity,
we can conclude that the western end of the rupture caused
by the mainshock was at the longitude of about 29.2⬚ E in
the İzmit Bay. The remarkable activity that appeared in region (e) is considered as a swarm activity triggered by the
mainshock, because the activity started more than 2 days
after the commencement of the mainshock. We suggest that
the southern portion of region (d) was also activated by the
mainshock.
Figure 7 shows a space–time plot of microearthquakes
that occurred from the beginning of 1998 in a region including both the northern and southern branches of the
NAFZ. This clearly shows that the activity of the microearthquakes along the southern branch also increased after the
1999 İzmit earthquake.
Discussion and Conclusions
The size of the aftershock area within a few days after
a large earthquake is generally considered to correspond to
the dimension of the rupture zone, and then the aftershock
area expands with time (Utsu, 1969; Tajima and Kanamori,
1985). As IZINET was established several years prior to the
1999 İzmit earthquake and had been in operation when it
occurred, we could obtain aftershock data within a critical
period of a few days after the mainshock. The aftershock
distribution is the information essential to estimate the western end of the rupture zone, because we cannot, at present,
confirm the surface rupture under the sea. From the aftershock distribution obtained in our study, we concluded that
the western end of the rupture is at the longitude of about
29.2⬚ E. On the other hand, the eastern end of the rupture
would be at about 31.0⬚ E, as inferred from both of the aftershock distribution and the surface rupture that appeared
in association with the mainshock. Therefore, the total length
of the rupture zone of the 1999 İzmit earthquake would be
about 150 km.
This estimate of the rupture length is quite different
from the solutions of waveform inversion as discussed by
Yagi and Kikuchi (2000). The point up to which the rupture
424
A. Ito, B. Üçer, Ş. Barış, A. Nakamura, Y. Honkura, T. Kono, S. Hori, A. Hasegawa, R. Pektaş, and A. M. Işıkara
Figure 6. Time-dependent distributions of
the aftershocks. Periods are defined in the
figure.
of the main shock extended is an important issue for estimating the risk of future disastrous earthquakes. From the
physical point of view, for the estimation of the rupture area,
the result of the waveform inversion would be more reliable
than the aftershock distribution because we do not know
exactly what the aftershock occurrences represent. However,
such a claim holds good only when we have sufficient
knowledge about the detailed underground structure, but in
reality, there is some ambiguity about the underground structure. For example, several fault models were proposed from
various points of view, based on a variety of data sets, and
the results are somewhat different from each other (for instance, Toksöz et al., 1999; Yagi and Kikuchi, 2000; Sekiguchi and Iwata, 2002). In this sense, the distribution of the
aftershocks is still an important piece of information, because the aftershock distribution is based only on the velocity structure. Moreover, the velocity structure has little affect
on the result of our present study.
Aftershocks were distributed sharply along a narrow
vertical zone in regions (a) and (b), as shown in Figure 6.
They occurred immediately after the mainshock and decayed
rapidly, which may indicate a small amount of energy re-
425
Aftershock Activity of the 1999 İzmit, Turkey, Earthquake
Figure 7. Space–time plot of the microearthquakes that occurred during the period
from 1 January 1998 to 30 September 1999. Horizontal bars represent periods when
IZINET was down for more than 15 days continuously.
lease in these segments during the mainshock. We can see
from Figures 4 and 6 that the active and the inactive parts
are in clear contrast to each other. Such a contrast indicates
that the source region is separated into several segments.
Furthermore, aftershocks have tended to occur not in the
east–west direction as expected, but with a linear trend in
the northeast–southwest direction near the hypocenter of the
mainshock. These features of aftershock occurrences imply
a complex rupture process of the mainshock; it is quite possible that the slip in regions (a) and (b) along the segment
in the İzmit Bay was rather smooth without strong asperities,
so that no strong seismic waves were radiated from there.
In the eastern part of the source region, most of the
aftershocks did not occur along the surface rupture. As
shown in Figure 6, the center of activity in region (d) moved
southwards 1 day after the mainshock, which suggests that
the occurrence of the mainshock increased the stress along
the fault of the 1967 Mudurnu earthquake, especially at the
deeper portion of the fault. A similar mechanism might have
been operative in the western part of the source region. In
fact, an earthquake with a magnitude of M 6.3 occurred near
region (a) in 1963 (Toksöz et al., 1979), and hence the
swarm activity in region (e) might be related to this past
event.
It is very important to note that the 1999 İzmit earthquake initiated the rupture in the area adjacent to the active
swarm area. Generally speaking, there is no clear relation
between the swarm activity of the microearthquakes and the
occurrence of a large earthquake. Although Michael and
Toksoz (1982) showed statistically that many swarm activities in western Turkey are an aftermath of large earthquakes,
the distances between the swarms and the corresponding
earthquakes were very large in most cases. The 1999 İzmit
earthquake is the first large earthquake in the western part
of the NAFZ since a modern seismic observation network
has become available. This earthquake demonstrated the
possibility that the swarm activity of microearthquakes
might be related to the occurrence of a large earthquake in
the NAFZ. Although we cannot derive any definitive conclusions and we need more examples, we should pay more
attention to the swarm activities along the NAFZ.
On 12 November, another large event with Mw 7.1 occurred near Düzce. Milkereit et al. (1999) showed that the
source region of this November event is adjacent to that of
the İzmit earthquake. Therefore, we should pay more attention to the west of the source region of the İzmit earthquake
that could perhaps be the region where the next large earthquakes may occur, because the İzmit earthquake is likely to
have increased the stress accumulation in the western area
(Parsons et al., 2000). At the same time, we should also take
care of the southern branch of the NAFZ, the Iznik–Mekece
Fault, because we do not know to what extent the stress
accumulation in the seismic gap area suggested by Toksöz
et al. (1979) was released by the İzmit earthquake. In fact,
the activity of the microearthquakes along the southern
branch increased remarkably after the İzmit earthquake as
shown in Figure 7, in spite of the decrease in the Coulomb
failure stress (Parsons et al., 2000). In any case, we should
intensify seismic observations as soon as possible for monitoring the microearthquake activity possibly related to earthquakes that may perhaps occur in the future in northwestern
Turkey.
426
A. Ito, B. Üçer, Ş. Barış, A. Nakamura, Y. Honkura, T. Kono, S. Hori, A. Hasegawa, R. Pektaş, and A. M. Işıkara
Acknowledgments
We would like to convey our heartfelt thanks to Dr. Ali Pinar, Nafiz
Kafadar, and other members of the Seismological Department, Kandilli
Observatory for their cooperation. We thank Cengiz Çelik, Niyazi Sarıkaya,
and other staff of the Iznik Branch of the Kandilli Observatory, Boğaziçi
University, for their valuable help in the fieldwork. Yukiko Ishikawa, Utsunomiya University, contributed substantially to the phase reading of the
aftershocks. Stuart Crampin and Andy Michael gave us valuable suggestions as reviewers, and their comments improved the manuscript very much.
The Turkish Airlines (THY) and Ataturk Airport Customs have always
helped us with the transportation of equipments between Turkey and Japan.
This study was partly supported by the grant-in-aid for Scientific Research
(Nos. 04041046, 07044071, 11694063) from the Ministry of Education,
Science, Culture and Sports, Japan, by funding from the Kagami Foundation, Japan, and by the research fund of the Boğaziçi University (99T202).
References
Allen, C. R. (1969). Active faulting in northern Turkey, Contribution Number 1577, Div. Geol. Sci., Calif. Inst. Tech. 32.
Ambraseys, N. N. (1970). Some characteristic features of the North Anatolian fault zone, Tectonophysics 9, 143–165.
Ambraseys, N. N., and A. Zatopek (1969). The Mudurnu Valley, West
Anatolia, Turkey, Earthquake of 22 July 1967, Bull. Seism. Soc. Am.
59, 521–589.
Barıs, Ş., A. Ito, S. B. Üçer, Y. Honkura, N. Kafadar, R. Pektaş, T. Komut,
and A. M. Işıkara (2002). Microearthquake activity before the İzmit
earthquake in eastern Marmara region, Turkey (1 January 1993–17
August 1999), Bull. Seism. Soc. Am. 92, no. 1, 394–405.
Barıs, Ş., S. B. Üçer, A. Ito, N. Kafadar, H. A. Alçık, A. Pınar, C. Gürbüz,
Y. Honkura, and A. M. Işıkara (1996). A Local micro-earthquake
network IZINET in the western part of the North Anatolian Fault
Zone, Turkey, Presented at Japan Earth and Planetary Sciences Joint
Meeting, 1996 Osaka, Japan, E31–01.
Barka, A. (1992). The North Anatolian Fault Zone, Ann. Tectonicae 6, 164–
195.
Barka, A. (1996). Slip distribution along the North Anatolian fault associated with the large earthquakes of the period 1939–1967, Bull. Seism.
Soc. Am. 86, 1238–1254.
Barka, A. (1997). Neotectonics of the Marmara region, in Active Tectonics
of Northwestern Anatolia—The MARMARA Poly-Project, C. Schindler and M. Pfisters (Editors), Hochschulverlag AG an der ETH, Zurich, 55–87.
Barka, A., S. Akyuz, E. Altunel, G. Sunal, Z. Çakir, A. Dikbas, B. Uerli,
T. Rockwell, J. Dolan, T. Dawson, R. Hartlet, A. Tucker, T. Fumal,
R. Langridge, H. Stenner, S. Christofferson, R. Armijo, B. Meyer,
and J. B. de Chabalier (1999). The 17 August 1999 İzmit Earthquake,
Northwestern Turkey, Presented at the AGU 1999 Fall Meeting, San
Francisco, S11D-01.
Crampin, S., and S. B. Üçer (1975). The seismicity of the Marmara sea
region of Turkey, Geophys. J. R. Astr. Soc. 40, 269–288.
Crampin, S., R. Evans, and S. B. Üçer (1985). Analysis of local earthquakes: the Turkish dilatancy projects (TDP1 and TDP2), Geophys.
J. R. Astr. Soc. 83, 1–16.
Dewey, J. W. (1976). Seismicity of northern Anatolia, Bull. Seism. Soc.
Am. 66, 843–868.
Ergin, M., M. Aktar, F. Biçmen, A. Yörük, N. Yalçin, and S. Küleli (1997).
Microearthquake investigation in İzmit Bay, in Proc. of the Meetings
of the Research Group on Active Tectonics, Faculty of Mining &
Institute of Eurasya, Istanbul Technical University, Istanbul, 1–11 (in
Turkish).
Evans, R., D. Beamish, S. Crampin, and S. B. Üçer (1987). Turkish dilatancy project (TDP3): multidisciplinary studies of a potential earthquake source region, Geophys. J. R. Astr. Soc. 91, 265–286.
Gürbüz, C., S. Kuleli, T. Bekler, E. Zor, H. Karabulut, and M. Zobu (1998).
Crustal structure studies in western Turkey based on earthquake and
explosion data, in Proc. of the XXVI General Assembly of the European Seismological Commission (ESC), Tel Aviv, 106–111.
Honkura, Y., and A. M. Işıkara (1991). Multidisciplinary research on fault
activity in the western part of the North Anatolian Fault Zone, Tectonophysics 193, 347–357.
Honkura, Y., A. M. Işıkara, N. Oshiman, A. Ito, S. B. Üçer, Ş. Barış, M. K.
Tunçer, M. Matsushima, R. Pektaş, C. Çelik, S. B. Tank, F. Takahashi, M. Nakanishi, R. Yoshimura, Y. Ikeda, and T. Komut (2000).
Preliminary results of multidisciplinary observations before, during
and after the Kocaeli (İzmit) earthquake in the western part of the
North Anatolian Fault Zone, Earth Planet. Space 52, 293–298.
Horasan, G. H., S. Kuleli, and L. Gülen (1997). Crustal and upper mantle
structure in western Turkey, in Proc. of the Meetings on the Research
Group on Active Tectonics, Faculty of Mining & Institute of Eurasya,
Istanbul Technical University, Istanbul, 94–102 (in Turkish).
Iio, Y., C. Gürbüz, K. Nishigami, A. Pınar, N. Aybey, S. B. Üçer, Y.
Honkura, and A. M. Işıkara (1991). Microseismic activity around the
western extension of the 1967 Mudurnu earthquake fault, Tectonophysics 193, 327–334.
Iio, Y., S. Horiuchi, Ş. Barış, C. Çelik, J. Kyomen, B. Üçer, Y. Honkura,
and A. M. Işıkara (2002). Aftershock distribution around the eastern
end of the earthquake fault of the 1999 İzmit, Turkey, earthquake
Bull. Seism. Soc. Am. 92, no. 1, 411–417.
Jackson, J., and D. McKenzie (1988). The relationship between plate motions and seismic moment tensors, and the rates of active deformation
in the Mediterranean and Middle East, Geophys. J. Int. 93, 45–73.
Lovell, J. H., S. Crampin, R. Evans, and S. B. Üçer (1989). Micro-earthquakes in TDP swarm, Turkey; clustering in space and time, Geophys.
J. R. Astr. Soc. 91, 313–330.
Michael, A. J., and M. N. Toksöz (1982). Earthquake swarms as a longrange precursor to large earthquakes in Turkey, Geophys. J. R. Astr.
Soc. 68, 459–476.
Milkereit, C., S. Zunbul, S. Karakisa, Y. Iravul, J. Zschau, M. Baumbach,
H. Grosser, E. Gnther, N. Umutlu, T. Kuru, E. Erkul, K. Klinge,
M. Seht, and A. Karahan (1999). Preliminary aftershock analysis of
the Mw ⳱ 7.6 İzmit earthquake in western Turkey, Presented at the
AGU 1999 Fall Meeting, San Francisco, S12D-03.
Nakamura, A., A. Hasegawa, A. Ito, B. Üçer, Ş. Barış, Y. Honkura,
T. Kono, S. Hori, R. Pektaş, T. Komut, C. Çelik, and A. M. Işıkara
(2002). P-wave velocity structure of the crust and its relationship to
the occurrence of the 1999 İzmit, Turkey, earthquake and aftershocks,
Bull. Seism. Soc. Am. 92, no. 1, 330–338.
Nishigami, K., Y. Iio, C. Gürbüz, A. Pınar, N. Aybey, S. B. Üçer,
Y. Honkura, and A. M. Işıkara (1990). Microseismic activity and
spatial distribution of coda-Q in westernmost part of the North Anatolian Fault Zone, Turkey, Bull. Dis. Prev. Res. Inst. Kyoto Univ. 40,
41–56.
Parsons, T., S. Toda, R. Stein, A. Barka, and J. Dieterich (2000). Heightened Odds of large earthquakes near Istanbul: an interaction-based
probability calculation, Science 288, 661–665.
Sekiguchi, H., and T. Iwata (2002). Rupture process of the 1999 Kocaeli,
Turkey, earthquake estimated from strong-motion waveforms, Bull.
Seism. Soc. Am. 92, no. 1, 300–311.
Tajima, F., and H. Kanamori (1985). Global survey of aftershock area expansion patterns, Phys. Earth Planet. Inter. 40, 77–134.
Toksöz, M. N., M. Edie, C. G. Doll, L. Xu, V. F. Cormier, and M. Bouchon
(1999). Fault rupture process of the 17 August 1999 earthquake in
Turkey, Presented at the AGU 1999 Fall Meeting, San Francisco,
S11D-02.
Toksöz, M. N., A. F. Shakal, and A. J. Michael (1979). Space-time migration of earthquakes along the North Anatolian Fault Zone and seismic
gaps, Pageoph 117, 1258–1270.
Tsukuda, T., K. Satake, Y. Honkura, S. B. Üçer, and A. M. Işıkara (1988).
Low seismicity, low coda-Q, and discontinuities of the upper crust in
the vicinity of the Iznik-Mekece fault, the North Anatolian Fault Zone,
Turkey, Bull. Earthquake Res. Inst. Univ. Tokyo 63, 327–348.
427
Aftershock Activity of the 1999 Izmit, Turkey, Earthquake
Üçer, S. B., S. Crampin, R. Evans, and N. Kafadar (1985). The MARNET
radio linked seismometer network spanning the Marmara Sea and the
seismicity of western Turkey, Geophys. J. R. Astr. Soc. 83, 17–30.
Utsu, T. (1969). Aftershocks and earthquake statistics (I)—some parameters which characterize an earthquake sequence and their interrelations, J. Fac. Sci. Hokkaido Univ. Ser. VII 3, 129–195.
Yagi, Y., and M. Kikuchi (2000). Source rupture process of the Kocaeli,
Turkey, earthquake of August 17, 1999, obtained by joint inversion
of near-field and teleseismic data, Geophys. Res. Lett. 27, 1969–1972.
Faculty of Education
Utsunomiya University
Utsunomiya 321-8505
Japan
ito@cc.utsunomiya-u.ac.jp
(A.I.)
Geophysical Department
Kandilli Observatory and Earthquake Research Institute
Boğaziçi University
Çengelköy
Istanbul 81220
Turkey
(B.Ü., S.B., R.P., A.M.I.)
Research Center for Prediction of Earthquakes and Volcanic Eruptions
Graduate School of Science
Tohoku University
Sendai 980-8578
Japan
(A.N., A.H., T.K., S.H.)
Department of Earth and Planetary Sciences
Tokyo Institute of Technology
Tokyo 152-8551
Japan
(Y.H.)
Manuscript received 7 October 2000.