A first deep seismic survey in the Sea of Marmara: Deep basins and whole crust architecture and evolution

Abstract

Increased source strength, streamer length and dense spatial coverage of seismic reflection profiles of the SEISMARMARA Leg 1 allow to image the deep structure of the marine North Marmara Trough (NMT) on the strike-slip North Anatolian Fault (NAF) west of the destructive Izmit 1999 earthquake. A reflective lower crust and the Moho boundary are detected. They appear upwarped on an E-W profile from the southern Central Basin eastwards, towards more internal parts of the deformed region. Thinning of the upper crust could use a detachment suggested from an imaged dipping intracrustal reflector that would allow upper crustal material to be dragged from beneath it and above the lower crust, accounting for the extensional component but also southwest motion of the southern margin of the NMT. Sections across the eastern half of the NMT, crossing the Cinarcik and Imrali basins, reveal several faults that are active reaching into the basement and have varying strike and proportions of normal and strike-slip displacement. They might be viewed as petals of a large scale negative flower-structure that spreads over a width of 30 km at surface and is rooted deeper in the lithosphere. Under the Central Basin a very thick sediment infill is revealed and its extensional bounding faults are active and imaged as much as 8 km apart down to 6 km depth. We interpret them as two deep-rooted faults encompassing a foundering basement block, rather than being merely pulled-apart from a jog in a strike-slip above a décollement. The deep-basin lengthening would account for only a modest part of the proposed 60 km finite motion since 4 Myr along the same direction oblique to the NMT that sidesteps the shear motion from its two ends. Thus differential motion occurred much beyond the deep basins, like subsidence involving the NMT bounding faults and the intracrustal detachments. The complex partitioned motion localized on active faults with diverse natures and orientations is suggested to represent the overburden deformation induced from horizontal plane simple shear occurring in depth at lithospheric scale, and in front of the North Anatolian Fault when it propagated through the region.

Introduction

The origin, evolution and present activity of the northern Sea of Marmara that formed on the North Anatolian Fault (NAF) is a matter of debate. It has been brought into focus in the discussion of the corresponding seismic hazard, after the 1999 destructive earthquakes that occurred just east of it (Fig. 1a). These Izmit and Düzce M > 7 events occurred in a westward progression of major earthquakes since 1939 reaching from the eastern end of the NAF to the eastern tip of the Sea of Marmara. The segment of the NAF located in the West of the Sea of Marmara, which crosses the peninsula of Gallipoli, has also slipped in the 1912 Ganos earthquake. The Sea of Marmara appears as a seismic gap since the previous earthquake sequence of the eighteenth century.

As a geologically active fault, the NAF propagated into the region, and crossed it towards the Aegean in the west in lower Pliocene, some 4-5 Myr ago (Armijo et al., 1999, Hubert-Ferrari et al., 2003). It grossly straddles the intra-Pontide suture zone where the oceanic domain between Eurasia and the Sakarya continent closed. This occurred diachronously from the East, along the Armutlu Peninsula to North of Marmara Island (Yilmaz et al., 1996), with the Neogene-Eocene Thrace basin being regarded as a forearc basin (Görür and Okay, 1996).

In the SEISMARMARA French-Turkish program (Hirn et al., 2001/2002; Hirn, 2003, Bécel et al., 2004, Carton et al., 2007, Bécel, 2006), the N/O Le Nadir acquired in August 2001 multichannel seismic reflection profiles (MCS) in the North Marmara Trough (NMT). This acquisition has been restricted to the deep part of the Sea of Marmara, in water depths larger than 100 m (Fig. 1b). The profiles have an unprecedented depth-penetration due to the 4.5 km length of a 360-channel streamer and the strength of the low frequency sources, a 8100 cu. in., or 2900 cu. in., 14-airgun array. Efficient penetration was obtained with the original "single-bubble" method, which results from the modification of conventional airgun array shooting. We developed this method for lithospheric probing with rather modest academic marine seismic facilities, in the Streamers-Profiles surveys (Avedik et al., 1995, Avedik et al., 1996), later adopted by others (White et al., 2002). With its strong source signal at low-frequency the effects of attenuation are limited and its still short source signal duration allows a reasonable resolution.

We focus here on Leg 1 of the multichannel survey which comprised a network of 4 E-W lines and 30 cross-lines with a total of about 2000 km encompassing the whole NMT. This provided a coverage and spatial resolution of the main structures of regional significance. Thirty-seven ocean-bottom seismometers (OBS), and a larger number of land seismometers recorded coincident wide-angle reflection-refraction, as well as local earthquakes (Bécel, 2006).

Previous marine reflection seismic surveys (Okay et al., 1999, Okay et al., 2000, Imren et al., 2001, Parke et al., 2002) provided stacked time-sections which were limited in depth by the sea-bottom multiple. Together with the detailed in-plane view at the sea-bottom, from a comprehensive sea-beam survey (Le Pichon et al., 2001, Armijo et al., 2002), these previous studies have captured essential elements of the general picture. Despite this penetration limitation into greater depths, different aspects in their interpretations have been emphasized in terms of contrasting simple end-member models of present activity (e.g. Yaltirak, 2002). Among those models: plate-junction like geometry of structures (Okay et al., 2000), ongoing pull-apart at different scales (Armijo et al., 2002), important extensional faulting (Parke et al., 2002, Aksu et al., 2000, Yaltirak, 2002), recent dominance of strike-slip localization (Le Pichon et al., 2001) do not appear to dominate from the new observations.

As a major unprecedented result of the tuning of the SEISMARMARA Leg1 survey's acquisition parameters, the image of the crust is resolved here through its whole thickness. This is for instance the case of Profile SM 36 (Fig. 2a) that crosses the eastern basin of the North Marmara Trough, or Cinarcik Basin and its southern flank. On this record-section, primary reflections are obtained from the thick infill of the deep basin, its basement, and the whole crust over 30 km depth. This opens new volumes to imaging, thus providing new perspectives to the description and understanding of the structure and its evolution.

Indication of times in seconds will hereafter always refer to echo times read on MCS profiles and which correspond to two-way travel times to reflectors. All the profiles have not been depth-migrated, and we prefer thus to keep all the sections in time to discuss the structural features. Thanks to the streamer length (4.5 km), we have a precise idea of the velocity range for the shallow sedimentary layers. The P-wave velocities are very low for the sea-bottom deposits, especially in the deep bathymetric troughs (1.6 to 1.8 km/s), and are gradually increasing from sea-bottom down to the pre-kinematic basement where they can reach values of 4 to 4.2 km/s. These values are consistent with the joint modeling of the shots recorded by the OBS array. These data allow especially to constrain velocities for the deeper structural units, which are ranging between 5.7 to 6.3 km/s for the upper crust and on the order of 6.7 km/s for the lower crust (Bécel, 2006).