Global Positioning System (GPS) estimates of crustal deformation in the Marmara Sea region, Northwestern Anatolia

Global Positioning System (GPS) estimates of crustal deformation in the Marmara Sea region, Northwestern Anatolia

EPSL ELSEVIER Earth and Planetary Science Letters 121 (1994) 495-502 Global Positioning System (GPS) estimates of crustal deformation in the Marmara...

985KB Sizes 0 Downloads 34 Views

EPSL ELSEVIER

Earth and Planetary Science Letters 121 (1994) 495-502

Global Positioning System (GPS) estimates of crustal deformation in the Marmara Sea region, Northwestern Anatolia Christian Straub, Hans-Gert

Kahle

Institute of Geodesy and Photogrammetry, ETH H6nggerberg, CH-8093 Zurich, Switzerland

(Received September 27, 1993; revision accepted November 29, 1993)

Abstract

In the Marmara Sea region the relative motion between the Anatolian and Eurasian lithosphere has been measured by means of the modern techniques of space geodesy using the Global Positioning System (GPS). In order to resolve in detail the kinematic field within the active earthquake belts of Northwestern Anatolia, two GPS campaigns were carried out across a dense network consisting of 45 sites. Preliminary results reveal horizontal westward motion of Anatolia relative to Eurasia of 2.4 cm/y on average that runs along the western end of the North Anatolian Fault Zone. E - W oriented right-lateral strike-slip movement (average 0.06 ppm/y, maximum 0.13 ppm/y) (1 p p m / y = 1 ~strain/y) and NE-SW extension (average 0.06 ppm/y, maximum 0.21 ppm/y) can be obtained from the calculated strain field. In the west strike-slip motion trending WSW-ENE is observed. These findings are compared to the stress pattern derived from seismology and neotectonics.

1. Introduction

As part of the A l p i n e - H i m a l a y a n orogenic belt, Anatolia is a region of high complexity. Its geological and tectonic evolution has been dominated by long-term cycles of openings and closures of Paleozoic and Mesozoic oceans [1,2,3]: continent-continent collision zones, extensional provinces, transform faults and subduction zones have formed an assemblage of lithospheric blocks with a wide range of thicknesses and rheologies. Recent satellite laser ranging (SLR) data [4] confirmed the assumption that current tectonic processes are accompanied by the northerly movement of the Arabian plate towards Eurasia In

[uc]

addition, the results of finite element modelling [5] raise doubts about the assumption that the pushing movement of Arabia is the only driving mechanism. It is, however, generally proposed that most of the deformation in Anatolia is caused by the Caucasus convergent zone crustal escape along major strike-slip faults [6] such as the North Anatolian Fault Z o n e (NAFZ). The N A F is a narrow seismically active fault that extends over more than 1000 km and accommodates the relative motion between the Anatolian and Black Sea microplates. The dominant tectonic structures of our research area are the active fault segments of the western end of the N A F Z and its transition into a complex system of restraining and releasing bends of strike-slip faults within and near the M a r m a r a Sea [7]. The high potential for seismic energy gives cause for considerable concern about

0012-821X/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0012-821X(93)E0227-B

C. Straub, H.-G. Kahle / Earth and Planetary Science Letters 121 (1994) 495-502

496

the earthquake hazard along these faults [8], which are in many cases separated by active pullapart basins that experience high seismicity. In 1990 ETH Ziirich started the MARMARA Poly-Project [9] (MPP), the object of which is the study of the interaction between groundwater circulation, heat flow and seismicity, and the development of tectonic models constrained by space geodetic measurements; the project involves the study of tectonics, geology, seismology, hydrogeology, geothermics and geodesy. The Geodetic Institute of Istanbul Technical University (IT0) and the General Command of Mapping (GCM) in Ankara are participating in the part of the project that is based on satellite geodesy. The main goal of this part of the project is to assess the deformation field around the Marmara Sea and to contribute to the geodynamic modelling of the continuing tectonic processes. Special emphasis is placed on five regions (Fig. 1) in which the GPS was employed for the determination of strain and strain rates. The geology group of the MPP is at the same time carrying out extensive field work in these regions, which are delineated in such a way as to cover on the one hand the local tectonic features proposed by the MPP geologists and on the other the three 27

28

so-called 'strands' of the westward extension of the NAF. All three strands originate near region 2, to the southeast of AdapazarL The southern strand follows the line from the Mudurnu Valley through Bursa to Edremit and is covered by regions 2, 3, 4 and 5. Since 1855 at least four strong earthquakes (M = 6-7) have occurred along this belt. The middle strand extends from the Mudurnu Valley along Iznik and Bandlrma towards Ezine and is also covered by regions 2, 3, 4 and 5. The northern strand lies for the most part offshore, beneath the northern half of the Marmara Sea. It can be traced from near Adapazarl along the Gulf of Izmit to Tekirdag and further on to the Gulf of Saros (region 1). This segment ruptured in 1912 (M = 7.3).

2. Measurement campaigns

The GPS [10] is based on the simultaneous determination of relative distances between several ground receivers and a number of GPS satellites by means of phase measurements of their radiowave signals. In 1990, 45 GPS sites were selected and installed in order to meet the various geological and neotectonic objectives. In or-

29

¢'.. --r~...~.___.~L,:,.?.,

-

-

~

q;!A~a.Pla,oI

1

..a_

Graben Structur~ ~

S t o k e Slip Fault

Fig. 1. T h e five a r e a s o f i n v e s t i g a t i o n a r e s u p e r i m p o s e d o n t h e m a j o r t e c t o n i c s t r u c t u r e s [7,9]. L o c a l n e t w o r k s : 1 = K u r u D a g - M f i r e f t e s e i s m i c z o n e [26]; 2 = K u z u l u k s u b s i d e n c e a r e a [27]; 3 = E z i n e , T u z l a [28], G u l f o f E d r e m i t a n d B e r g a m a ; 4 = I z m i t - I z n i k a n d B u r s a [29]; 5 = G 6 n e n - Y e n i c e s e i s m i c z o n e . S L R s t a t i o n Y l g d c a is s h o w n in t h e inset. Ad = A d a p a z a n ; Ba = B a n d l r m a ; Ed = E d r e m i t ; Ez = E z i n e ; Izm = Izmit; Izn = Iznik; Mu = M u d u r n u ; Te = T e k i r d a g .

C. Straub, H.-G. Kahle /Earth and Planetary Science Letters 121 (1994) 495-502

der to gain insight into the regional deformation field the baseline lengths within this network range between 2 km and 169 km. The first period of measurement was carried out in 1990 [11], and in 1992 the whole network was remeasured [12,13]. In both campaigns WM102 receivers were applied. All the stations were observed in at least two sessions of 5 (1990) and 6 (1992) hours each. For a detailed description of the GPS sites and extensive documentation on the campaigns see [11,12,13]. In 1992, four sites were occupied throughout the entire campaign in order to strengthen the solution and to assess the daily repeatability. Apart from the regions shown in Fig. 1, an additional baseline leading to the closest SLR station, Ylgtlca (see inset in Fig. 1), was measured in 1992 in order to establish a link with the Global International Terrestrial Reference Frame of 1991. This tie to a global reference frame allows us to obtain more precise relative coordinates [14]. The GPS campaign of 1992 was coordinated with the Massachussetts Institute of Technology (MIT) [15] and the Institute for Applied Geodesy (IfAG) [16] who were measuring their GPS networks in Western Anatolia at the same time as us. This coordination enabled link-

26 °

27 °

28 °

497

age of all the data collected by the different groups.

3. Analysis The analysis of the GPS data was carried out with version 3.4 of the Bernese software package [17]. The following ephemeris of the GPS satellite orbits were used: US D M A precise ephemeris for the 1990 data and I G S - C O D E orbits [18] for the 1992 observations. In the first step of the analysis daily solutions were calculated for the individual observation sessions. For each day a set of baselines was chosen which would connect the observed sites in such a way that for each region one station with very good data coverage could be selected as a main reference site and tied to Istanbul. All the other sites of each individual region were then connected to their corresponding reference station. Ionospheric disturbances were eliminated by processing a linear combination (L3) of the dual GPS microwave frequencies. Refractive effects of the troposphere were corrected for using the model of Saastamoinen. Before evaluating the phase measurements cycle-

29 o

30 °

31 °

41 °

40*

39 °

26 °

27 °

28 °

29 °

30 °

31 °

Fig. 2. Displacement estimates derived from GPS measurements between 1990 and 1992. The displacements are given relative to the reference site in Istanbul. Dots on the arrows indicate observed sites; ellipses on the point of each arrow show the individual formal errors multiplied by a factor of 20.

498

C. Straub, H.-G. Kahle / Earth and Planetary Science Letters 121 (1994) 495-502

slips within the data were removed, and the ambiguities (integer number of carrier wavelengths) were subsequently determined. Finally, a solution taking into acount all the data collected throughout the entire campaign was calculated. The obtained formal errors of the horizontal components of coordinates relative to Istanbul range between 0.3 mm and 1.3 mm (site-dependent). These formal errors are statistical error estimates and must not be regarded as actual coordinate errors. It is generally known that the calculated formal errors are too small because they do not account for systematic effects. In order to obtain a better estimate of the accuracy of the daily solutions, they were compared with the final solution derived from all observation sessions. These relative comparisons are called the daily repeatabilities. The analysis resulted in daily repeatabilities of up to 4 mm (lo- rms) for the horizontal components of the coordinates, and of up to 21 mm (lo- rms) for the vertical component. The maximum deviation in position was 17 mm (this value was reached in one out of 270 site occupations). A multiplication factor of 10 is generally assumed as a good estimate for the coordinate accuracies. However, the error ellipses shown in

26 °

27 °

28 °

Fig. 2 reflect a multiplication of the formal error by a factor of 20, which seems to be a realistic value for the indication of the accuracy achieved.

4. Results The displacements (Fig. 2) derived from the comparison of the two datasets (1990 and 1992) refer to the site in Istanbul as a fixed point. A dextral drift of the Anatolian plate to the west can be clearly seen. Using an assumed stable Black Sea microplate (Eurasia) as a reference, the rates of displacement reach 2 c m / y in the eastern part, and a little more than 3 c m / y in the western part. In addition to the westward movement of Anatolia a N E - S W trend can be seen in the results. In order to visualize these findings, the strain field was calculated by using the method of two-dimensional collocation [19,20] for interpolation of the horizontal components of the displacement vectors. Collocation can be regarded as a general method of least-squares adjustment including parameter estimation, filtering and prediction. The displacements, the signal to noise ratio of the displacements and a covariance

29 °

30 °

31 °

41"

40*

39 °

26 °

27 °

28 °

29 °

30 °

31 °

Fig. 3. Normal strain rates derived from the biannual observations of displacements. White arrows indicate extension; black arrows show compression. The length of the arrows indicate the strength of the strain rate (1 ppm/y = 1/.tstrain/y). The maximum normal strain rate is 0.21 ppm/y. The inset shows an azimuthal plot of the tr3 axes for the area west of 27.7°E. The average strike of the major axes of extension is N41°E.

C. Straub, H.-G. Kahle / Earth and Planetary Science Letters 121 (1994) 495-502

function are used as input. The covariance function correlates the interpolated displacement values with the actual measurements. For the calculations we used a covariance function that is the reciprocal of the square of the distance. The correlation length of the covariance function was set to 64 km (signal to noise ratio = 5.7). In Fig. 3 the results are presented in terms of the normal strain rates of compression and extension, and in Fig. 4 in terms of shear strain rates. These calculations enable us to obtain more detailed information on the deformation field within the five regions. The distribution of normal strain rates indicates N E - S W oriented extension regimes over the northern strand of the NAF, and N - S extension on the Biga Peninsula (western part of the network). The average compressional component is 0.06 p p m / y (maximum = 0.20 p p m / y ) (1 p p m / y = 1/zstrain/y), while the m e a n extensional component is 0.06 p p m / y (maximum = 0.21 p p m / y ) . The field of shear strain rates reveals dextral motions throughout the northern strand of the N A F ' s westward extension and a distinct southwesterly bend in the west (Fig. 4). The average shear strain rate is also 0.06 p p m / y (maximum = 0.13 p p m / y ) .

26 °

27 °

28 °

499

The results of previous GPS measurements of regional deformation carried out in Turkey in 1988 and 1990 by the M I T group yielded a rightlateral motion accross the N A F Z of 2.5 c m / y . These results were based on an extensive network covering a much larger area (only three stations of the M I T network lie within the M a r m a r a Sea region) [15]. The results obtained from the more local and therefore dense M a r m a r a Sea network with its 45 sites presented here confirm the overall findings of the M I T group. In addition, this network facilitates the identification of the fine structure of the strain field, which shows significant N E - S W extension in the eastern and middle part of our investigation area. To the west of the M a r m a r a Sea region a bend in the extension towards N - S was detected.

5. C o m p a r i s o n

with seismological

data

The deformation field described above accords with maps of earthquake epicentres. Such maps clearly reveal how the N A F trifurcates [21] as it approaches the M a r m a r a Sea from the Mudurnu Valley area (region 2 in Fig. 1) towards Saros

29 °

30 °

31 °

n

39 °

26 °

27 °

28 °

29 °

30 °

31 °

Fig. 4. Shear strain rates derived from the biannual observations of displacements. The length of the arrows indicates the strength of the strain rate (1 ppm/y = 1 /~strain/y). If the strain is assumed to accumulate linearly the maximum shear strain rate is 0.13 ppm/y.

500

C. Straub, H.-G. Kahle / Earth and Planetary Science Letters 121 (1994) 495-502

Bay, Iznik and the B u r s a - G 6 n e n - Y e n i c e seismic zone (see regions 1, 4 and 5 in Fig. 1). Large-scale interpolated maps of the current strain field in the Eastern Mediterranean and the Caucasus region presented by Rebai et al. [22] also indicate extensional processes in Western Anatolia. We have also compared our results with focal mechanism solutions for various strong earthquakes within the study area. The last large event in region 2 was the 1967 Mudurnu earthquake ( M - - 7 . 1 ) , which was interpreted as due to dextral E - W oriented (N95°E) strike-slip motion [23]. Recent estimates of deformation along the N A F Z have been made by Kiratzi [24] based on the moment tensors of numerous earthquakes. It was calculated that in the east of our study area, close to region 2, the deformation was mainly taken up by shearing with an azimuth of N86°E, with an average motion of 16 m m / y . The average strike of the shear motion calculated from our GPS data in region 2 (to the west of the Mudurnu Valley) is N84°E, and the average E - W component of slip rate is 15 m m / y (relative to Istanbul), which is compatible with the earthquake focal mechanism solution. Another example is the 1963 Cmarctk earthquake ( M = 6.4) in the eastern Marmara Sea. This was associated with normal faulting, with the 0-3 axis (extension) oriented N56°E [23]. In the same area around Cmarclk (the Gulf of Izmit) we observed extensional processes with a strike of N44°E. In the northwestern Marmara Sea (region 1) the 1912 and 1975 Saros earthquakes ( M = 7.4 and 6.6) occurred on a fault system striking N68°E [23]. Based on repeated GPS observations the average strike of the shear in the Gulf of Saros is N75°E. An analysis of the earthquake focal mechanisms in the southern part of region 3 and to the southeast of it has recently been conducted by Zanchi and Angelier [25], who conclude that the dominant trend of the current strain field in Western Anatolia is a N N E - S S W extension. This was deduced from the analysis of 66 earthquake mechanisms which also included earthquakes with magnitude M < 6. The average strike of the principal axis of extension in this part of our area (regions 1, 3 and 5) calculated from the GPS-derived strain field is N41°E (a polar plot of the azimuthal

distribution of the o"3 axes is shown in the inset in Fig. 3). This value clearly lies within the range of the seismotectonic solutions ( N N E - S W to N E W), which depend on the depth of the earthquakes [25] and on the reactivation of older faults. The mean extensional strain rate derived from the GPS for this western part is 0.06 p p m / y .

6. Conclusions and outlook

Two GPS campaigns were successfully conducted in 1990 and 1992. The displacements and the calculated strain field for this two-year interval provide a detailed view of the current tectonic motions in the Marmara Sea region. Evaluation of the GPS data resulted in a deformation field which confirms the results of earlier investigations in this region. Comparisons with large-scale strain fields, earthquake epicentres and focal solutions show an overall agreement; the focal mechanisms of strong earthquakes that have occurred around the Marmara Sea are comparable with the principal strain axes derived from GPS data. It is interesting to note that, although no very strong earthquakes occurred in the investigation area between the two campaigns, the result of our observations indicates the same sense of movement that the available focal solutions derived from seismic events exhibit. Global plate motion models average velocities over a span of several million years. It is therefore remarkable that our measurements, which were carried out with an interval of merely two years, show the same direction and amount of displacement that is suggested by neotectonic studies which take geological time scales into consideration. This demonstrates once more that regional and local measurements with the GPS are of considerable importance and great value for the study of tectonic problems as complex as those in the Marmara Sea region. In view of the fact that only one repetition of the zero epoch measurements has been done so far, we refrain at this stage from an in-depth interpretation of the comparison between the GPS data and those provided by seismotectonics and neotectonics.

C. Straub, H.-G. Kahle / Earth and Planetary Science Letters 121 (1994) 495-502

Further investigation using a combination of three observational methods as varied and independent of each other as seismology, geology and geodesy opens up new avenues for studying the mechanisms of long-term tectonic processes and can therefore shed some light on the forces at lithospheric plate boundaries. In order to obtain a more detailed deformation field, the GPS measurements will be repeated again in autumn 1994. In the future our work will also comprise more extensive analysis of the obtained deformation in local and regional frameworks in terms of strain and stress analysis which will be based on collocation and finite element modelling. Furthermore, an investigation into the active faults will be carried out on a more local scale in collaboration with the MPP geology, hydrogeology, seismology and geothermic groups.

7. Acknowledgements We thank all our partners who are working with us on the MARMARAProject. These include a large number of scientists from the ETH in Zurich (C. Schindler (leader), W. Balderer, T. Eisenlohr, E. Greber, T. Imbach, C. Jeckelmann, D. MayerRosa, St. Miiller, N. Pavoni, M. Pfister, L. Rybach and S. Sellami) and our Turkish partners at the ITI3 (A. Aksoy) and GCM (I. Klmk). We also appreciate the close cooperation with the MIT (R. Reilinger), IfAG (H. Seeger) and Durham University (G. Foulger), who are also doing GPS work in Anatolia. We thank too TU Berlin, UniBw Mfinchen, Uni Stuttgart, FHT Stuttgart, TH Karlsruhe, TU Wien and EPF Lausanne for generously putting receivers at our disposal and for their participation. We also thank T. Imbach (ETH Zfirich), A. Geiger (ETH Ziirich) and three reviewers for critically reading the manuscript and for their helpful suggestions. Thank you to K. Blarer for her invaluable help in editing the English. The MARMARA Project is financed by the ETH Ziirich, and we are especially grateful to R. Hfitter, ETH Vice-President of Research and Development, for his continuing help and support in the many aspects of this project.

501

8. References [1] J.F. Dewey and A.M. Seng6r, Aegean and surrounding regions: Complex multiplate and continuum tectonics in a convergent zone, Geol. Soc. Am. Bull. 90, 84-92, 1979. [2] J.A. Jackson and D.P. McKenzie, The relationship between plate motions and seismic moment tensors, and the rates of active deformation in the Mediterranean and the Middle East, Geophys. J.R. Astron. Soc. 93, 45-73, 1988. [3] A. Seng6r and Y. Yilmaz, Tethyan evolution of Turkey: A plate tectonic approach, Tectonophysics 75, 181-241, 1981. [4] R. Noomen, B. Ambrosius and K. Wakker, Crustal motions in the Mediterranean region determined from LASER ranging to LAGEOS, A G U Crustal Dyn. Proj. Monogr., in press, 1993. [5] K.E. Kasapoglu and M.N. Toks6z, Tectonic consequences of the collision of the Arabian and Eurasian plates: Finite element models, Tectonophysics 100, 71-95, 1983. [6] D.P. McKenzie, Active tectonics of the AlpineHimalayan belt: The Aegean and surrounding regions, Geophys. J.R. Astron. Soc. 55, 217-354, 1978. [7] A. Barka and K. Kadinsky-Cade, Strike-slip fault geometry in Turkey and its influence on earthquake activity, Tectonics 7(3), 663-684, 1988. [8] M.N. Toks6z, A.F. Shakal and A.J. Michael, Space-time migration of earthquakes along the North Anatolian fault zone and seismic gaps, Pure Appl. Geophys. 117, 12581270, 1979. [9] C. Schindler et al., The MARMARA Poly-project: tectonics and recent crustal movements revealed by spacegeodesy and their interaction with the circulation of groundwater, heat flow and seismicity in Northwestern Turkey, Terra Nova 5, 164-173, 1993. [10] G. Seeber, Satellite Geodesy, 531 pp., De Gruyter, Berlin, 1993. lll] U. Marti, H.-G. Kahle, A. Aksoy and R. Deniz, GPS-Projekt Marmara: Dokumentation der ersten Messkampagne September 1990, Inst. Geod. Photogramm. (IGP), Publ. 179, 47 pp. and Append., 1991. [12] C. Straub and H.-G. Kahle, GPS Projekt MARMARA: Punktprotokolle Stand 1993, Inst. Geod. Photogramm. (IGP) Publ. 227, 56 pp., 1993. [13] C. Straub, H.-G. Kahle, A. Aksoy and R. Deniz, GPS Projekt MARMARA: Report on the second campaign August/September 1992, Inst. Geod. Photogramm. (IGP) Publ. 208, 45 pp. and Append., 1992. [14] G. Beutler, I. Bauersima, W. Gurtner, M. Rothacher, T. Schildknecht and A. Geiger, Atmospheric refraction and other important biases in the GPS carrier phase observations, Sch. Surv., Univ. South Wales, Aust., Monogr. 12, 1988. [15] B. Oral, R. Reilinger, A. Barka and M.N. Toks6z, Preliminary Results of 1988 and 1990 GPS measurements in

502

[16]

[17]

[18]

[19]

[20] [21]

[22]

C. Straub, H.-G. Kahle / Earth and Planetary Science Letters 121 (1994) 495-502 Western Turkey and their tectonic implications, AGU Crustal Dyn. Proj. Monogr., in press, 1993. Y. Altiner, SLR- und GPS-Messungen in Anatolien und erste Resultate, Vermess. Raumordnung 54(8), 393-399, 1992. M. Rothacher, G. Beutler, W. Gurtner, E. Brockmann and L. Mervart, Bernese GPS software version 3.4: documentation, May 1993. G. Beutler, The 1992 IGS test campaign, Epoch'92, and the IGS pilot service: an overview, Proc. 1993 Bern IGS Workshop, in press, 1993. G. Danuser, A. Geiger and M. Miiller, Modellierung yon Verschiebungs- und Verzerrungsfeldern, Inst. Geod. Photogramm. (IGP) Publ. 218, 176 pp., 1993. H. Moritz, Advanced Physical Geodesy, 500 pp., Wichmann, Karlsruhe, 1980. S. Crampin and S.B. l~lcer, The seismicity of the Marmara Sea region of Turkey, Geophys. J.R. Astron. Soc. 40, 269-288, 1975. S. Rebai, H. Philip and A. Taboada, Modern tectonic stress field in the Mediterranean region: evidence for variation in stress directions at different scales, Geophys. J. Int. 110, 106-140, 1992.

[23] T. Taymaz, J. Jackson and D.P. McKenzie, Active tectonics of the north and central Aegean Sea, Geophys. J. Int. 106, 433-450, 1991. [24] A. Kiratzi, A study on the active crustal deformation of the North and East Anatolian Fault Zones, Tectonophysics 225, 191-203, 1993. [25] A. Zanchi and J. Angelier, Seismotectonics of western Anatolia: regional stress orientation from geophysical and geological data, Tectonophysics 222, 259-274, 1993. [26] C. Schindler, Die St6rungszone von GSl~iik (Tiirkisch Thrazien) und ihr Einfluss auf die Sedimentation des Altterti~irs, Eclogae Geol. Helv. 52(2), 695-707, 1960. [27] E. Greber, Das Geothermalfeld von Kuzuluk/Adapazan, Ph.D. Thesis 9984, ETH, Zurich, 1992. [28] S. Miitzenberg, Westliche Biga-Halbinsel ((~anakkale, Tiirkei): Beziehung zwischen Geologie, Tektonik und Entwicklung der Thermalquellen, Mitt. Geol. Inst. ETH und Univ. Ziirich 287, 160 pp., 1991. [29] T. Imbach, Thermalw~isser von Bursa. Geologische und hydrogeologische Untersuchungen am Berg Uludag (NW-Tiirkei), Ph.D. Thesis 9988, ETH, Zurich, 1992.