Investigation of long period crustal deformation on the inactive branch of the North Anatolian Fault Zone

The western part of North Anatolian Fault (NAF) bifurcates around Mudurnu into two fault segments: northern and southern branch. The latter bifurcates again at west of Pamukova and creates middle strand. This study aimed to analyze crustal movement along the middle strand near Iznik which is considered as inactive fault. We focused on a microgeodetic network called General Command of MappingIstanbul Technical University (GCM-ITU) network around this segment. In order to obtain displacement values, five campaigns performed on the network which were used in the study. The displacements of the stations were estimated relative to the fixed stations located at the south of the network. The coordinates of the stations were calculated from the triangulation measurements realized in 1941 and 1963, trilateration measurements in 1981, and GPS campaigns in 2004 and 2007. Then, mean displacements of the network ranging between 7 mm/yr and 18 mm/yr were obtained for these years. In the second part of the study, the GPS data were reprocessed by adding three stations from Marmara Continuous GPS Network (MAGNET). Details of MAGNET can be found Ergintav et al. (2002). Estimated displacements were ranging between 3 mm/yr and 13 mm/yr for 2004 and 2007. TUBI station of IGS network was taken as stable.


Introduction
North Anatolian Fault Zone as the largest fault system of Anatolia is a natural junction between the Caucasus System and Aegean extension regime.North Anatolian Fault begins Correspondence to: H. Ozener (ozener@boun.edu.tr)from Karliova, elongates through whole Northern Anatolia until the vicinity of Mudurnu where bifurcated into Izmit-Sapanca Fault and Iznik-Mekece Fault (Fig. 1).The belt which is surrounded by these two faults form a middle strand which begins from Mudurnu Valley and goes through Geyve, Iznik, Gemlik and the southwest of the Marmara Region.
The Iznik-Mekece Fault that is located around west of Geyve, goes along approximately 15 km up to Sakarya as an right lateral fault (Barka, 1997) and then bounds the Geyve basin and then leaps over to the right side continuous through Mekece and Iznik Lake in W-SW direction.Many researches were conducted along the fault for investigating seismic potential (Stein et al., 1997;Barka and Reilinger, 1997;Reilinger et al., 2000Reilinger et al., , 2006;;Ayhan et al., 2002;Ergintav et al., 2002Ergintav et al., , 2007;;Meade et al., 2002;Ozener et al., 2009).The GCM-ITU network, that we focused, was established around Iznik region, and the points of network scattered along the middle strand.The study area and the stations are displayed in Fig. 2.
On the other hand, strong seismic events (M>6) have not occurred on this fault nearly 200 years (Straub, 1996).Considering these studies mentioned above and own experiments, Barka (1997) suggested that the appearance interval of large events for Iznik area might have been around 2000 years or more.Thus, the fault is claimed as inactive in general.Figure 3 shows the seismicity of the region.
Geodesy Department of Kandilli Observatory and Earthquake Research Institute (KOERI) started crustal deformation studies on the Iznik region.Between 1990 and 1994, the  Barka and Reilinger, 1997).
The GCM-ITU network was originally implemented by GCM and measured with conventional techniques in 1941, 1963and 1981. In 1941and 1963, the network observed with triangulation method by GCM.These observations actually conducted to connect the stations to the national triangulation network.In 1981, the network was re-observed by Geodesy Department of ITU applying angle and distance measurements.The slope distances were observed by tellurometer, and then reduced to sea level.
After twenty years, the network became a part of the microgeodetic networks of Geodesy Department of KOERI.It was designated to monitor crustal deformation around the western part of NAFZ.In 2004 and 2007, two GPS campaigns were performed on the network.
In addition to this, the study area was extended by adding new stations from MAGNET.This extended network helped us to find the displacements of all points of GCM-ITU network relative to TUBI station of MAGNET.
By this study, both terrestrial and GPS campaign data sets for a long time period along Iznik-Mekece Fault is presented for the first time.

Method for the data analysis
The composition of network geometry had been changed for each campaign.Some of the network stations were not implemented before 1981 and some of them damaged in years.The network stations and the campaign dates are shown in Table 1.
While calculating displacements, the data analyzed into two steps.First, the terrestrial data were processed and analyzed.The inadequate number of control stations to transform one coordinate system to another led us to analyze the all GPS data as if they had been observed by conventional methods.Therefore, GPS baseline results were employed as a trilateration network set into the network adjustment for the first step.
In order to have adjusted coordinates for each observation • A single free network adjustment per epoch, • A single constrained adjustment according to reference stations, •  • Comparison of displacements and their analysis were computed.
The GPS data were processed individually in the second part of the study for GCM-ITU network and the extended network.Therefore, we were able to obtain more recent information about movement around the Iznik-Mekece Fault.
In order to compare the displacements, the stations HDRZ, ARML, and TURB were taken as fixed; so the concept of the adjustment method was based on the idea of having sufficient number of stations located one side of the Iznik-Mekece Fault.As a result, we could determine the relative motions of remaining stations.Control stations also served as datum definition in the study.

The analysis between 1981 and 2007 by terrestrial measurements
Prior to the development of electronic distance measuring equipments and the global positioning system, triangulation was the preferred method for extending horizontal control over long distances.In 1941 and 1963, the GCM-ITU network was monitored by triangulation method.Nevertheless   as a single observation.In adjustment process, a program using least squares, written in FORTRAN, was applied.However, the insufficient geometry of networks had a misleading influence on results and mean square errors were obtained more than 350 mm.Consequently, the 1941 and 1963 observations could not be used in displacement analysis.
In 1981, both trilateration and triangulation methods were applied.Baselines were measured by one of the first generation EDM systems: the tellurometer.Horizontal distances were produced from the raw slope distances by using the reduction to sea level formulas: where S a is the slope distance, H 1 and H 2 are the ellipsoidal heights, R α is the Gauss sphere radius, S E is the ellipsoidal distance, and S 0 is the horizontal distance.Changes in the coordinates of TAVD, ARAN and HDAG stations were evaluated using constrained adjustment relative to the fixed stations (HDRZ, ARML, TURB).Apart from these, the number of redundant observations was adequate in 1981 survey to apply free network adjustment in order to check outliers.
The Pope test which applies a posteriori variances was used upon observations in order to check outliers and determine the confidence regions.Standardized correction for Pope Test was found 3.106.Thus, the observations whose errors exceed that correction limit were eliminated.The maximum positioning error was mp max =±140 mm.
The GPS measurements were carried out in 2004 and 2007 using campaign method.Trimble Geomatics Office Software (TGO, 2007) was employed to process GPS data using precise orbits gathered from International Global Navigation Satellite Systems Service (IGS, 2007) web site.Constrained adjustment applied to only one control station, which was held fixed in the survey network and was employed to evaluate GPS measurements.In order to compare the terrestrial data to the GPS outcomes, baseline components were separated from coordinate results and designating as raw data and then processed again individually as trilateration measurements by constrained adjustment relative to fixed stations (HDRZ, ARML, TURB) in ED-50 datum.Apart from the individual adjustment of each campaign, we combined them and used as a single trilateration network in free network adjustment in order to increase the degrees of freedom.Afterwards, the Two Dimensional Helmert Transformation was applied to coordinates obtained from free network adjustment of combined 2004 and 2007 GPS data.Then, displacements were investigated between the outputs of TAVD, ARAN and HDAG stations from the network adjustment results of 1981 and of combined GPS campaign of 2004 and 2007.Each component of the coordinates and the directions of the displacement vectors are shown in the Table 2 as well as in Fig. 4.

The analysis between 2004 and 2007 by GPS measurements
Geodetic observations were repeated on GCM-ITU Network in 2004 and 2007 using GPS technique by Geodesy Department of KOERI.However, only six of the stations could survive to this time.IGAZ station was added to network, in order to densify the northern part.The method of static GPS measurement was performed in this study.Therefore the campaigns had been planned to monitor the network at least 8 h with 15 s logging interval, but some environmental problems lessen that time.
TGO and GPS Inferred Positioning System/Orbit Analysis and Simulation Software (GIPSY/OASIS II) from JPL (Webb and Zumberge, 1993;Gregorius, 1996) were used in GPS processing.
First, precise coordinates of stable stations were evaluated using precise point positioning (PPP) method in GIPSY/OASIS II.Then, constrained adjustment option in TGO was employed.The analysis strategy was similar to the previous section in order to compare the outcomes.Changes on northing and easting coordinates for each tied stations (HDAG, ARAN, TAVD, ARML, IGAZ) were investigated according to the significance and direction of movement.Changes on northing and easting coordinates between 2004 and 2007 are shown in Table 3.The largest movement is found in TAVD which is the furthest station from the Iznik-Mekece Fault.Accordingly, the minimum movement is found in HDAG, which is the nearest station to the fault.
As displayed on Fig. 5, the direction of movements is different from the terrestrial observation results and all stations are oriented to northeast relative to fixed stations (HDRZ, ARML, TURB) located in Anatolian plate.
Fig. 2. Study area and the stations.Stations shown by the triangles are the stations of the GCM-ITU Network.The filled triangles are the fixed stations and the empty ones are tied stations.The diamond symbols show the stations of MAGNET.

Fig. 2 .
Fig. 2. Study area and the stations.Stations shown by the triangles are the stations of the GCM-ITU Network.The filled triangles are the fixed stations and the empty ones are tied stations.The diamond symbols show the stations of MAGNET.

Fig. 3 .
Fig. 3. Seismicity of the region.Seismic data (between 1900 and 2007) is obtained from National Earthquake Monitoring Center of KOERI.

Fig. 3 .
Fig. 3. Seismicity of the region.Seismic data (between 1900 and 2007) is obtained from National Earthquake Monitoring Center of KOERI.

Fig. 4 .
Fig. 4. The amount of displacements between 1981 and 2004-2007.The displacements of TAVD, ARAN, and HDAG stations were estimated relative to the fixed stations (TURB, HDRZ, and ARML) located in the Anatolian plate.

Fig. 4 .
Fig. 4. The amount of displacements between 1981 and 2004-2007.The displacements of TAVD, ARAN, and HDAG stations were estimated relative to the fixed stations (TURB, HDRZ, and ARML) located in the Anatolian plate.

Fig. 5 .
Fig. 5.The amount of displacements between 2004 and 2007.The displacements of TAVD, ARAN, HDAG, and IGAZ stations were estimated relative to the fixed stations (TURB, HDRZ, and ARML) located in the Anatolian plate.

Fig. 5 .
Fig. 5.The amount of displacements between 2004 and 2007.The displacements of TAVD, ARAN, HDAG, and IGAZ stations were estimated relative to the fixed stations (TURB, HDRZ, and ARML) located in the Anatolian plate.

Fig. 6 .
Fig. 6.The amount of displacement between 2004 and 2007 on the extended network.The network is analyzed relative to TUBI station.

Fig. 6 .
Fig. 6.The amount of displacement between 2004 and 2007 on the extended network.The network is analyzed relative to TUBI station.

Table 1 .
The network stations and the available data in the campaign dates.
both measurements did not contain adequate number of observations for network adjustment.Furthermore, they also failed individually to form proper geometry.Therefore, it was decided to combine both measurements and adjust them

Table 3 .
Station List and Estimated Displacements between 2004 and 2007 in WGS-84 Datum.The displacements of TAVD, ARAN and HDAG stations were estimated relative to the fixed stations (TURB, HDRZ, ARML) located in the Anatolian plate.

Table 4 .
Station List and Estimated Displacements between 2004 and 2007 in WGS-84 Datum.The displacements of the stations were estimated relative to the TUBI station.