Shear Model

1. A tomographic image of the shallow crustal structure
in the Eastern Marmara
Hayrullah Karabulut
Bog˘azic¸i University, Kandilli Observatory and Earthquake Research Institute, C¸ engelkoy, Istanbul, Turkey
Serdar O¨ zalaybey
TU¨ BI
:
TAK, Marmara Research Center, Earth and Marine Sciences Research Institute, Gebze, Kocaeli, Turkey
Tuncay Taymaz
Istanbul Technical University, Department of Geophysics, Maslak, Istanbul, Turkey
Mustafa Aktar
Bog˘azic¸i University, Kandilli Observatory and Earthquake Research Institute, C¸ engelkoy, Istanbul, Turkey
Og˘uz Selvi
TU¨ BI
:
TAK, Marmara Research Center, Earth and Marine Sciences Research Institute, Gebze, Kocaeli, Turkey
Argun Kocaog˘lu
Istanbul Technical University, Department of Geophysics, Maslak, Istanbul, Turkey
Received 30 June 2003; revised 16 August 2003; accepted 12 November 2003; published 26 December 2003.
[1] We present a 2-D tomographic seismic velocity image
in eastern Marmara region along a N-S trending 120-km
long seismic refraction profile, which traverses the northern
branches of North Anatolian Fault (NAF) and tectonically
active C¸ narck Basin, in the Sea of Marmara. The 2-D
velocity model, which is constrained down to upper
crustal depth of 7 km, shows significant heterogeneities in
the upper crust displayed by well-constrained velocity
anomalies. The seismicity observed following the
17 August, 1999 I
:
zmit earthquake concentrates in three
distinct zones and takes place near the inferred low velocity
zones and below high velocity anomalies obtained from
tomographic inversion. This correlation is interpreted as a
result of intensely sheared zones of deformation imposed
by strike-slip motion of the northern branches of the
North Anatolian Fault. INDEX TERMS: 0902 Exploration
Geophysics: Computational methods, seismic; 7205 Seismology:
Continental crust (1242); 7230 Seismology: Seismicity and
seismotectonics. Citation: Karabulut, H., S. O¨ zalaybey,
T. Taymaz, M. Aktar, O. Selvi, and A. Kocaog˘lu, A tomographic
image of the shallow crustal structure in the Eastern Marmara,
Geophys. Res. Lett., 30(24), 2277, doi:10.1029/2003GL018074,
2003.
1. Introduction
[2] The North Anatolian Fault (NAF), which displays a
simple character along most of its 1500 km-long course,
splits into several strands in the Marmara Sea Region
(MSR) of western Turkey. The region is a transition zone
between the strike-slip regime of the NAF and the extension
regime of the Aegean Sea area. The area has witnessed
several major historical earthquakes [Ambrasseys and
Jackson, 2000]. The most recent one, 17 August, 1999
I
:
zmit earthquake (Mw = 7.4), caused extensive damage
and high loss of human life. Following this disastrous
event a major scientific effort has been undertaken in
order to obtain a better estimation of the seismic potential
of active faulting within the MSR [Le Pichon et al., 2001;
Hirn et al., 2002].
[3] During the summer of 2001, within the framework of
SEISMARMARA project [see Hirn et al., 2002 for a
detailed description], seismic refraction data were collected
along a 120 km long north-south profile crossing the
I
:
stanbul zone, C¸ narck Basin (C¸ B) and Armutlu Peninsula
(AP) of the MSR (Figure 1). The shots were fired in the sea
and recorded on the land. The seismic profile cuts across
three branches of the NAF. Intense seismic activity was
observed on the northern branches of the NAF following the
August 17,1999 I
:
zmit earthquake (Mw = 7.4) [Karabulut et
al., 2002; O¨ zalaybey et al., 2002]. The diffuse seismic
activity observed west of I
:
znik Lake is related to the
southern branch of the NAF. A tomographic image derived
from microearthquakes after the I
:
zmit earthquake in the
east of the survey area is presented by Nakamura et al.
[2002] but it does not intersect our seismic profile. In this
study, we use active shots to invert for the crustal velocity
across the Marmara Sea. Next, we present a combined
tectonic interpretation for the eastern MSR based on the
knowledge of the inferred velocity structure, surface geol-
ogy, and seismicity of the MSR.
2. Geology and Tectonic Setting
[4] The MSR covers two different tectonic zones, the
I
:
stanbul-Zonguldak zone and Sakarya continent. The
I
:
stanbul-Zonguldak zone is characterized by a thick Ordo-
vician to Carboniferous, sedimentary sequence, which rests
unconformably on a Precambrian metamorphic basement. It
GEOPHYSICAL RESEARCH LETTERS, VOL. 30, NO. 24, 2277, doi:10.1029/2003GL018074, 2003
Copyright 2003 by the American Geophysical Union.
0094-8276/03/2003GL018074$05.00
SDE 10 -- 1

2. is separated in the south along the Intra-Pontide suture from
the Sakarya continent. The AP is a composite entity made
up of sections of the Sakarya continent, Rodope-Pontide
fragment and ophiolitic units [Ylmaz et al., 1997]. They
were assembled following a continental collision between
Gondwanaland and Laurasia during the Late Cretaceous.
The Rodope-Pontide zone and the Sakarya continent are
separated by the North Anatolian Fault Zone (NAFZ).
Although the geometry and the distribution of active faults
of the NAFZ in the MSR is complicated and remains
controversial in the literature, it is proposed that the NAF
began forming in the late Miocene but a major thoroughgo-
ing principal strike-slip displacement zone was established
in the Pliocene [Le Pichon et al., 2001]. The principal
displacement zone of the NAF enters the Marmara Sea as
a single main dextral fault strand [Le Pichon et al., 2001] but
splays into two branches, one following northern escarpment
of the C¸ B and the other along the southern margin of the C¸ B.
3. Seismic Data Acquisition and Preprocesing
[5] A total of 82 Reftek-125 recorders equipped with
4.5 Hz vertical component geophones were installed on land
with 0.5 to 2 km spacing, to record marine seismic shots in
the C¸ B and Gemlik Bay (Figure 1). The seismic source was
a tuned array of 8 airguns providing a total volume of
8000 in3
. The shot interval was 60 s, resulting in an average
shot spacing of 120 m with more than 180 shots. One quary
shot near S¸ile was also used. The data were continuously
recorded with a sampling rate of 100 Hz and played back
and stored as shot gathers and receiver gathers in SEG-Y
format.
[6] Figure 2 shows two examples of shot gathers. The
first arrivals were picked manually from selected shot and
receiver gathers. The receiver gathers were used since signal
noise ratio was high on some of receiver gathers providing
consistent picks. The traveltime uncertainities of the picked
arrivals vary between 20 and 50 ms. However, these
uncertainities can reach values of 100 ms at large distances.
The criterion of assigning uncertainities to the picks was
subjective and based on the signal-to-noise ratio (SNR).
[7] More than 3800 picks from 44 shot gathers and 18
receiver gathers are used in the analysis. Significant varia-
tions on both topography and bathymetry exist along the
profile. The traveltimes were corrected for the topography
along the profile using a constant velocity of 4.0 km/s and
vertical travel path under the receiver. No corrections were
applied for the crooked line geometry of the sources and
receivers and bathymetry.
4. Method
[8] We applied a tomographic inversion method to the
first-arrival travel-times which relies on travel-time compu-
tation using a finite difference algorithms developed by
Vidale [1988] and modified by Hole and Zelt [1995]. The
inversion is based on minimization of data misfit and model
roughness to provide the smoothest model appropriate for
the data error (see Zelt and Barton [1998] for details of the
Figure 1. Location map for the seismic profile overlying
bathymetry in the Sea of Marmara. Blue marks represent
recording stations, while green marks represent shots.
Blue star indicates the location of 17 August 1999 I
:
zmit
earthquake. Red circles show the aftershocks of 17 August
1999 I
:
zmit earthquake from 27 August 1999 to 17
September 1999 [Karabulut et al., 2002]. Active faults in
the Marmara Sea (purple lines) are from Le Pichon et al.
[2001]. Dashed Purple lines show branches of NAF in the
region.
Figure 2. Two shot gathers recorded for one shot in the
Gemlik Bay (upper) and one shot in the I
:
zmit Bay (lower).
The gathers were bandpass filtered between 2–16 Hz.
SDE 10 - 2 KARABULUT ET AL.: SHEARING EFFECT ON SMALL-SCALE CONVECTION

3. method). To assess quality of the inversion, traveltime RMS
residuals and c2
parameters are used. Optimum values of
the free parameters in the inversion, which controls vertical
versus horizontal smoothness and model roughness, were
obtained from a number of runs.
[9] Since tomographic inversions can reliably determine
small perturbations in the velocity field from an initial
model, which represents gross features of the structure, a
1-D starting velocity model was constructed with a subset of
shots (Figure 2). The velocity of water in the C¸ B and Gemlik
Bay were fixed to 1.53 km/s during the inversion. The model
was defined on a uniform 0.4 km grid extending from 0 to
120 km in the x-direction and 0 to 10 km in the z-direction
for all forward calculations. A 1.0 km lateral and 0.5 km
vertical cell size was used in the inverse step, resulting in
2400 independent model parameters. The RMS residuals for
the starting and the final models were 0.150 and 0.022 s
respectively, and c2
was 1.9. Since the data density is quite
uniform along the profile and there is no major geometric
distortion therefore, a uniform resolution would be expected.
5. Results
[10] Figure 3 displays the results of the tomographic
inversion and ray coverage for the final P-wave velocity
model. The velocities at shallow depths (1 km) are not
well constrained due to a lack of near offset data, as a result
these velocities are sensitive to the initial model. The ray
coverage indicates that the final velocity model is well
constrained between 1 and 6 km depth range. Similar results
were obtained from alternate initial models and different
inversion parameters except at shallow depths.
[11] Relatively high velocities (5.8–6.1 km/s) are located
underneath the AP (Figure 3). The peninsula is dominated by
metamorphic assemblages and an overlying late Mesozoic to
late Tertiary succession [Ylmaz et al., 1997]. Granitic
intrusions of Middle Eocene age crop out at western part
of the peninsula. The resulting velocities could be consistent
with both geological units. However, it is more likely that a
localized body of high velocity observed in this region is
related to the granitic intrusions within the metamorphic
body. At the southern end of the profile, in the Gemlik Bay,
velocities vary from 3.1–4.5 km/s down to 4 km depth.
[12] The velocities in the C¸ B increase from 2.5 km/s
below the water to 4.5 km/s at approximately 4 km depth.
The high velocities (6.0 km/s) are observed below the C¸ B
in a localized zone, at a depth of approximately 5 km. This
zone is in the vicinity of the ruptured segment of the NAF
during the Izmit earthquake (Figure 3). This zone is
delimited by two lower velocity regions located between
the northern strands of the NAF.
[13] The high velocities (5.7–6.0 km/s) north of the C¸ B,
over the Kocaeli Peninsula, take place below the Paleozoic
of I
:
stanbul, which consist of a sedimentary sequence of
Ordovician to Carboniferous age. The velocities of this
sedimentary sequence vary between 4.0–5.0 km/s.
6. Discussion and Conclusions
[14] The tomographic velocity model clearly reflects
geological complexities in the eastern MSR. The average
upper crustal velocity is around 5.7–5.9 km/s but laterally
varying in a consistent way with the known surface geology.
Both the Gemlik Bay and C¸ B are characterized by a
velocity gradient down to a depth of 4 km and present a
sharp velocity contrast within the surrounding units. There
is an apparent correlation between the seismicity and the
velocity distribution in the C¸ B. Figure 3 displays a cross
section showing the distribution of seismicity within 10 km
along the profile following the 17 August 1999 I
:
zmit
earthquake with the inferred P-wave velocity structure
resulting from the tomographic inversion. The aftershocks
were located using a 1-D earth model [Karabulut et al.,
2002]. The seismicity is concentrated within three distinct
zones, each representing significantly different character of
the tectonic regime of the region.
[15] The majority of the seismicity in the AP starts at a
depth of 3 km, below the inferred high velocity zone
interpreted as granitic intrusions. The focal mechanisms of
the microearthquakes show normal-fault solutions with
mainly E-W strikes [Karabulut et al., 2002]. The region is
well known for its numerous hot spring sources. It is likely
that the extensional movement tears up the older, deep-
reaching strike slip faults, which creates vertical zones of
high permability [Eisenlohr, 1997]. Therefore, the seismic
activity detected in this region, following the Izmit main
shock, may be a result of an increasing pore pressure.
[16] The seismicity in the C¸ B occurs below the imaged
high velocity body, in the vicinity of the ruptured segment
Figure 3. Upper: Initial P-wave velocity model; Middle:
P-wave velocity model obtained from tomographic inver-
sion and the seismicity within 10 km zone along the profile;
Lower: ray cell hitcount which is an indication for the ray
coverage and therefore for the velocity uncertainty.
KARABULUT ET AL.: SHEARING EFFECT ON SMALL-SCALE CONVECTION SDE 10 - 3

4. of the I
:
zmit earthquake, dominated by strike-slip mecha-
nisms [O¨ zalaybey et al., 2002]. Though we cannot accu-
rately determine the vertical extent of the retrieved anomaly,
our images show that it is located near the branching of the
rupture. We suggest that it could be related to the Intra-
Pontide suture, consisting of fragments of basaltic rocks.
[17] The seismicity observed along the northern segment
of the NAF occurs below a depth of 5 km. The focal
mechanisms in this region show normal faulting with NE-
SW strike direction between 4 and 7 km depth range in the
triggered Tuzla cluster. This zone coincides with the north-
ern escarpment of C¸ B, below a landslide observed on a
high-resolution bathymetry map that has been obtained
recently by Le Pichon et al. [2001]. The deeper parts of
the seismicity are dominated by pure strike slip motion
[O¨ zalaybey et al., 2002]. The two low velocity regions
imaged around the high velocity body are correlated with
strike slip focal mechanisms located between the northern
strands of the NAF (Figure 3). Although these two low
velocity anomalies lie in a poorly constrained regions, we
interpreted this correlation as the result of intensely sheared
zones of deformation imposed by strike-slip motion of the
northern branch of the NAF.
[18] Future studies involve including OBS data collected
during the experiment (they were not available during this
study) and local earthquake tomography to image deeper
part of the structure.
[19] Acknowledgments. We acknowledge the contribution of the
SEISMARMARA working group which include Hirn, A., Singh, S.,
Charvis, P., Geli, L., Laigle, M., Lepine, J.-C., de Voogd, B., Saatc¸lar, A.,
Vigner, A., Bazin, S., Tan, O, C¸ etin, S., Yolsal, S., Galve, A., Sapin,
M., Marthelot, J.-M., Taprdamaz, Boztepe, A., C., Taranc˜og˘lu, A., Diaz,
J., Verhille, J., and Auffret, Y. We thank Esen Arpat for discussions on the
geology of the study area. We thank two reviewers for constructive com-
ments. This research was supported by TU¨ BI
:
TAK, YDBAG - 102Y114.
[20] We thank Prof. Naci Gorur for his support during this project.
References
Ambrasseys, N. N., and J. A. Jackson, Seismicity of the Sea of Marmara
(Turkey) since 1500, Geophys. J. Int., 141, F1–F6, 2000.
Eisenlohr, T., The Thermal springs of the Armutlu peninsula (NW Turkey)
and their relationship to geology and tectonic, Active Tectonics of North-
western Anatolia-The MARMARA Poly-Project, Zu¨rich, 1997.
Hirn, A., S. Singh, P. Charvis, L. Geli, M. Laigle, J.-C. Lepine, B. de
Voogd, R. Saatc¸lar, T. Taymaz, S. O¨ zalaybey, H. Shimamura, O. Selvi,
H. Karabulut, Y. Murai, Y. Nishimura, A. Yamada, A. Vigner, S. Bazin,
O. Tan, S. Yolsal, M. Aktar, A. Galve, M. Sapin, J.-M. Marthelot,
C. I
:
mren, M. Ergin, C. Taprdamaz, A. Tarancog˘lu, J. Diaz, J. Verhille,
Y. Auffret, S. C¸ etin, N. Ocakog˘lu, F. Karakoc¸, E. Klien, A. Ricolleau,
and K. Sarkavak, SEISMARMARA 2001: A Marine Seismic Survey
and Offshore-Onshore Artificial Source and Natural Earthquakes in the
Seismogenic Regions of the Sea of Marmara, 1st International Sympo-
sium of the Faculty of Mines (ITU ) on Earth sciences and Engineering,
Istanbul, Turkey, 2002.
Hole, J. A., and B. C. Zelt, Three-dimensional finite-difference reflection
travel times, Geophys. J. Int., 121, 427–434, 1995.
Karabulut, H., M. Bouin, M. Bouchon, M. Dietrich, C. Cornou, and
M. Aktar, The seismicity in the Eastern Marmara Sea after the 17 August
1999 I
:
zmit Earthquake, Bull. Seism. Soc. Am., 92, 382–393, 2002.
Le Pichon, X., A. M. C. S¸engo¨r, E. Demirbag˘, C. Rangin, C. I
:
mren,
R. Armijio, N. Go¨ru¨r, N. C¸ ag˘atay, B. Mercier de Lepinay, B. Meyer,
R. Saatclar, and B. Tok, The active main Marmara fault, Earth Planet.
Sci. Lett., 192, 595–616, 2001.
Nakamura, A., A. Hasegawa, A. I
:
to, B. U¨ c¸er, S¸ Bars, Y. Honkura, T. Kono,
R. Pektas¸, T. Komut, C. C¸ elik, and A. M. Is¸kara, P-wave velocity struc-
ture of the crust and its relationship to the occurence of the 1999 Izmit,
Turkey earthquake and aftershocks, Bull. Seism. Soc. Am., 92, 330–338,
2002.
O¨ zalaybey, S., M. Ergin, M. Aktar, C. Taprdamaz, F. Bic¸men, and
A. Yo¨ru¨k, The 1999 I
:
zmit Earthquake Sequence in Turkey: Seismologi-
cal and Tectonic Aspects, Bull. Seism. Soc. Am., 92, 376–386, 2002.
Vidale, J. E., Finite-difference calculation of travel times, Bull. Seism. Soc.
Am., 78, 2062–2076, 1988.
Ylmaz, Y., O. Tu¨ysu¨z, E. Yig˘itbas¸, S¸ C. Genc¸, and A. M. C. S¸engo¨r,
Geology and Tectonic Evolution of the Pontides, in Regional and
petroleum geology of the Black Sea and surrounding region: AAPG
Memoir 68, edited by A. G. Robinson, pp. 183–226, 1997.
Zelt, C. A., and P. J. Barton, Three-dimensional seismic refraction tomog-
raphy: A comparision of two methods applied to data from Faeroe Basin,
J. Geophys. Res., 103, 7187–7201, 1998.
ÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀ
M. Aktar and H. Karabulut, Bog˘azic¸i University, Kandilli Observatory
and Earthquake Research Institute, C¸ engelkoy, 81220, Istanbul, Turkey.
(kara@boun.edu.tr)
A. Kocaog˘lu and T. Taymaz, Istanbul Technical University, Department
of Geophysics, Maslak, 80626, Istanbul, Turkey.
S. O¨ zalaybey and O. Selvi, TU¨ BI
:
TAK, Marmara Research Center, Earth
and Marine Sciences Research Institute, Gebze, 41470, Kocaeli, Turkey.
SDE 10 - 4 KARABULUT ET AL.: SHEARING EFFECT ON SMALL-SCALE CONVECTION