The 2005 - 2007 Bala (Ankara, central Turkey) earthquakes: a case study for strike-slip fault terminations

Strike-slip faults. Fault terminations. Tuzgölü fault zone. Neotectonics. Bala earthquake. Anatolia. KEYWORDS


INTRODUCTION
The neotectonics of Turkey was evaluated in the context of plate tectonics (McKenzie, 1972).The neotectonic provinces have been delineated as the East Anatolian contractional province, the Central Anatolian "ova" province, and the West Anatolian extensional province (Şengör, 1979, 1980; Şengör et al., 1985).A causeeffect relationship for the development of the neotectonic provinces has been generally accepted, the triggering event of this relationship was the collision of the Eurasian and the Arabian plates along the Bitlis suture (Dewey et al., 1986).The collision was followed by the development of the North and East Anatolian fault zones, and by the subsequent westward motion of the Anatolian Plate, which created N-S extensional tectonics in western Turkey (Şengör, 1979, 1980; Şengör et al., 1985).However, Seyitoğlu andScott (1991, 1996) argued that the initiation of extensional tectonics in western Turkey is older than the collision and therefore the cause-effect relationship is not a proper explanation for the development of the West Anatolian extensional province.The transition from the West Anatolian extensional province to the Central Anatolian "ova" province is roughly shown as a narrow N-S zone by McKenzie (1972) and Şengör et al. (1985).On the other hand, Barka and Reilinger (1997) indicate that the provinces are separated by the Eskişehir and Fethiye-Burdur fault zones (Fig. 1).The normal faultinduced earthquakes of 15 December 2000 Sultandağ (Mw=6.0)and 03 February 2002 Çay (Afyon) (Mw=6.3)(see Fig. 1 for location) (Taymaz and Tan, 2001;Başokur et al.., 2002;Emre et al.., 2003) changed the perception of the boundary between the western extensional province and that of central Anatolia (Koçyiğit and Özacar, 2003).The contractional structures previously attributed to the intracontinental convergence in NW central Turkey (Koçyiğit, 1991;Koçyiğit et al., 1995) were now shown as neotectonic structures (i.e.Eldivan-Elmadağ Pinched Crustal Wedge) developed through an interaction between the North Anatolian fault zone and the Kırıkkale-Erbaa fault zone (Seyitoğlu et al., 2000(Seyitoğlu et al., , 2009a)).
The NE-SW trending right lateral Kırıkkale-Erbaa fault zone and the NW-SE trending Tuzgölü fault zone converge in the Bala region, SE of Ankara (Fig. 1), where three significant earthquakes occurred on 30 July 2005(Mw=5.3), 20 December 2007(Mw=5.4), and 26 December 2007 (Mw=5.3).After the 30 July 2005 main shock, intense earthquake activity was observed in the Bala region.This area corresponds to the NW tip of Tuzgölü

A B B
fault zone and the main shocks and their aftershocks provide valuable information on the termination of strikeslip systems (Fig. 1).
The Afşar fault (ARF) and the Karakeçili fault (KKF) are two important active faults of the Bala region (Esat et al., 2009) (Fig. 2).It is easy to envisage the ARF as a continuation of the Tuzgölü fault zone due to its strike, but little is known about the KKF.This fault was previously called the Çatalören fault and interpreted as a left lateral conjugate of the ARF by Koçyiğit, (2003Koçyiğit, ( , 2009) ) and Tan et al. (2010).However, in other studies the KKF is described as a NNE-trending oblique-slip normal fault with a variable strike-slip component, bounding the Karakeçili Neogene-Quaternary basin (Türkmen and Özkul, 1999;Saraç, 2003).The KKF is composed of parallel and sub-parallel fault segments that involve basement rocks (mainly flysch and recrystallized limestone of Eocene age) and basin deposits (Neogene clastic sedimentary rocks, limestone, and gypsum).The slightly oblique normal fault surfaces and their slickensides are observed in the contact between the basement rocks and the sedimentary units (Fig. I, Electronic Appendix available at www.geologica-acta.com),and control the topography from the Bala highlands to the lower altitudes in the east (Fig. II).Minor fracturing, breccias and veins, are common brittle features in the basement rocks in the vicinity of the faults.The fracturing is mainly defined by widespread joints.The joints are subvertical, quite linear and several meters long.Their spacing varies from 0.1 to 1m.They occur at either parallel or variable angles to the normal faults.Some of the fractures contain striae, suggesting a shear mode of fracturing.Veins along faults are evidence of fluid presence during deformation.They are up to 1 m wide and traceable for several meters.Their walls are sub-vertical, sharp and plane.The veins are composed of coarse-grained calcite crystals with blocky and elongate shape, and show brecciated textures.The orientation of veins is approximately N-S, similar to that of the successive normal fault scarps seen in the basin deposits.
In the hanging wall of the KKF at the west of Çatalçeşme village, the Quaternary (?) cover has normal fault scarps (Fig. 3A).Maximum displacement ranges from 0.75 to 1m, with slip dissipating towards the fault tip (Fig. 3B, C).The Çatalçeşme seismic reflection study has been performed in the fault scarps from this area (Fig. 3A, Appendix), and confirms the normal fault nature of the KKF.Most of the displacements seen on the Quaternary (?) cover can be traced in the subsurface (Fig. 4A, B).
The ARF has a right lateral strike-slip character and it is composed of many segments.The fault traces/surfaces cannot be easily recognized in the field because of intense agricultural overprint.However, the existence of the fault can be determined by careful reconnaissance in the field and by the use of satellite images.Most of the NWtrending segments of the ARF cause a variable right lateral displacement on the stream channels (Fig. 2; Fig. 5A, B).After the 2007 main shock, observations on the graveyard wall of Afşar town provided evidence of subsidiary strikeslip shear fractures.Using these fractures, the trend of the master fault can be estimated as N25°W, which is compatible with the en echelon fault segments determined by the deviations of the stream channels in the north of Afşar (Fig. 5C; Fig. 6).These deviations are coherent with the aftershock distribution towards the SE of Afşar.To test the trend of the master fault estimated from the subsidiary fractures on the graveyard wall, a seismic reflection study has been performed based on the morphological expression of the NW trending stream channels (Fig. III, Appendix).In the Üç Ağıl seismic reflection section (Fig. 7A, B), the trace of the ARF is imaged just under the stream (Fig.

III).
A typical positive flower structure is observed because the seismic section passes near the end of the segment that shifts toward NE, creating a local restraining stepover (Fig. III).

SEISMICITY
Three significant earthquakes affected Ankara during the historical period (after 1900AD).The first one (I 0 =IX) occurred in an uncertain location among the cities of Ankara, Çankırı, and Çorum in AD 109 (Soysal et al., 1981).The other two occurred near the city of Ankara on 12 and 15 August 1668.The city center and the Beypazarı town were greatly damaged by these earthquakes (Ambraseys and Finkel, 1995).

Location and re-location process
The waveform data for the 30 July 2005 earthquake and its aftershocks were obtained from the Kandilli Observatory and Earthquake Research Institute of Turkey (KOERI).Nearly 100 events, recorded in high quality by at least 5 stations, were selected from almost 320, and all P and S phases were re-picked before the relocation process.The data of the 20 December 2007 earthquake and its aftershock were collected by the AnkNET.593 well recorded events were selected for phase picking.
The initial location of all events was performed using the double-difference hypocenter location method developed by Waldhauser and Ellsworth (2000) using the HYPODD algorithm of Waldhauser (2001).In the initial location processes, we used a simple P-and S-wave velocity model (Toksöz et al., 2003) corresponding to the central Anatolian crustal structure.Çıvgın and Kaypak (2012) estimated a new 1D velocity model for the region by using these initial locations and 2955 P-and S-seismic wave phases.This new model (Table 2) was used for relocation of the events.
The main shock location and aftershock distribution for the 30 July 2005 and 20 December 2007 earthquakes are shown in Fig. 8A and Fig. 9A, respectively.The aftershock distribution of the 30 July 2005 earthquake has a NNE-SSW trend on the KKF (Fig. 8A), the epicentral locations of the 20 December 2007 earthquake and its aftershocks have a NW-SE trend on the ARF (Fig. 9A).

Depth sections
The aftershock hypocenters of the 2005 main shock are distributed in the upper 35km and they are concentrated in the upper 10km (Fig. 8B).However, the earthquakes in the NE of the region are mostly below 10km depth (Fig. 8B, A-A' and D-D' sections).It can be interpreted that the KKF has an E-SE dipping geometry based on the distribution in the D-D' section which is perpendicular to it (Fig. 8B).
On the other hand, almost all aftershocks hypocenters of the 2007 main shock are above 12km depth (Fig. 9B).The depth sections perpendicular to the ARF show that the hypocenters are aligned in a narrow zone of about 4km (Fig. 9B, F-F' and G-G' sections).This aftershock distribution is compatible with the strike-slip nature of ARF.

Focal mechanism solutions
Focal mechanism solutions for both main shocks and their significant aftershocks were determined by a full waveform moment tensor inversion method which was developed and extended to full waveform data at regional or local distances by Zahradnik et al. (2005), based on Kikuchi and Kanamori (1991).For the computation of the earthquake source parameters, we used MATLAB based software called ISOLA-GUI, developed by Sokos and Zahradnik (2008).During the waveform inversion, the same one-dimensional velocity model was used for all focal mechanism solutions.
The focal mechanism solutions of the 2005 and the 2007 earthquakes and of their large aftershocks are shown in Figure 8A and Figure 9A, respectively.Their source parameters are summarized in Table 1.Besides, the focal mechanism solution of the 4 July 1978 (Mb=4.9)earthquake obtained from P-wave first motion polarities by Kocaefe and Ataman (1982) is given in Figure 8A and Table 1.This event is located on the KKF and indicates oblique-slip normal faulting (Fig. 8A; Table 1).

FAULT SLIP ANALYSIS
Fault slip data obtained from nodal planes of the focal mechanism solutions and from field observations in the Bala region have been analysed (Table 3).One of the two nodal planes of each focal mechanism solutions has been selected as the actual fault, based on the observation of fault orientations in the field.FaultKin software was used for the analysis (see Marrett and Allmendinger (1990); Rowland et al. (2007) for theoretical background of kinematic compatibility).The analysis of the overall data shows that the shortening axis is almost vertical and the extension axis is nearly horizontal.Linked Bingham axes represent these strain axes.According to this strain orientation, the extension axis indicates an extension direction of N81°E in the region (Fig. 10A).Moreover, the uniform direction of the contoured diagrams of the shortening (P) and the extension (T) axes shows that the faults (KKF and ARF) are kinematically compatible (Fig. 10B, C).Kocaefe and Ataman (1982).Solutions of the 30.07.2005 (Mw=5.3)and the 20.12.2007 (Mw=5.4)Bala earthquakes and their large aftershocks are provided by this study D O I : 1 0 . 1 3 4 4 / 1 0 5 .0 0 0 0 0 2 0 7 5 The  9A).These clusters led some researchers to suggest right and left lateral strike-slip conjugate faults are active in the Bala region (Dirik et al., 2008;Kalafat et al., 2008;Çubuk and Taymaz, 2009;Koçyiğit, 2009;Tan et al., 2010).On the other hand, after the field observations following the 30 July 2005 main shock, Kasapoğlu et al. (2005) reported that the earthquake was related to normal faults with a small component of left lateral strike-slip.
The main earthquake and aftershock relocations as well as focal mechanism solutions, together with new

FIGURE 1 .
FIGURE 1. Major neotectonic elements of central Anatolia.The white arrows indicate relative plate motions.The red frame shows the study area.Pink dots represent earthquakes from 1900 to the present.Earthquake data were obtained from the Kandilli Observatory and Earthquake Research Institute of Turkey's catalogue.
FIGURE 3. A) Detailed map of normal fault traces of the Karakeçili fault.Red lines represent the actual fault scarps within the Quaternary (?) units and yellow line shows the location of a seismic reflection study on the KKF.B) Field photo of the normal fault scarps.C) Normal fault scarp and trace of the seismic reflection profile.See Figure 2 for location.
). Six days later, on 26 December 2007 (UTC: 23:47:09.39),a large aftershock (Mw=5.3)shook the region.A time histogram of the events following the 20 December 2007 main shock is shown in Fig. VB.Some large aftershocks that occurred after the 20 December 2007 main shock are given in Table

FIGURE 4 .
FIGURE 4. Çatalçeşme seismic reflection section on the KKF.A) Uninterpreted.B) Interpreted (green arrows represent Quaternary (?) fault scarps' locations); red lines represent the faults.C) Close up view of the area within the frame in B.
2005-2007 Bala (Ankara, central Turkey) earthquakes 78 DISCUSSION AND CONCLUSIONS After the main shock on 30 July 2005 through 2009, earthquake clusters with NNE-SSW and NW-SE directions have been recorded.The aftershocks from the 30 July 2005 main shock have a NNE-SSW direction (Fig. IVB; Fig. 8A), and the 20 December 2007 main shock and its aftershocks have a NW-SE trend (Fig. IVC; Fig.

FIGURE 5 .FIGURE 6 .
FIGURE 5. A) ASTER satellite image of the study area.The black lines represent faults.B) Close up view of the Afşar area; the arrows show displacement directions of the streams.C) Google Earth satellite image; systematic right lateral displacements of the stream are observed north of the graveyard.

FIGURE 9 .
FIGURE 9.A) The main shock and its aftershocks recorded by AnkNET from 20 December 2007 to 10 October 2008.See Table1for the focal mechanism solutions.B) Earthquake depth sections.See A for the section lines.
FIGURE I. Photographs from the Karakeçili fault surfaces.The lithologies are limestone (A and E) and conglomerate (B, C, and D).White circles show the lower hemisphere equal-area stereographic projections of the fault planes.See Figure 2 for the locations.
FIGURE II.Geological cross-section showing the inner structure of the Karakeçili Neogene-Quaternary basin.See Figure2for the cross-section line.

FIGURE V .
FIGURE V. Histograms showing the number and distribution of aftershocks after the A) 2005 main shock and B) 2007 main shock.

TABLE 1 .
Parameters of the focal mechanism solutions related to the Afşar and Karakeçili Faults.Solution of the 04.07.1978 (Mb=4.9)earthquake is provided by

Table 2 .
Crustal 1D velocity model used in relocation (Çıvgın and Kayp

TABLE 2 .
(Çıvgın and Kaypak, 2012) used in relocation(Çıvgın and Kaypak, 2012) The earthquake activity between 30 July 2005 and 31 August 2005.See Table1for focal mechanism solutions.The epicenter locations were re-calculated using data from the Kandilli Observatory and Earthquake Research Institute of Turkey's catalogue.B) Earthquake depth sections.See A for the section line.

Table 3 .
Fault kinematic data obtained from nodal planes of the focal mechanism solutions and from field observations related to the Afşar and Karakeçili Faults in the Bala region (planes are displayed using the right-hand rule).

TABLE 3 .
Fault kinematic data obtained from nodal planes of the focal mechanism solutions and from field observations related to the Afşar and Karakeçili Faults in the Bala region (planes are displayed using the right-hand rule)