Preliminary Results of Full Seismic Waveform Tomography for Sea of Marmara Region (NW Turkey)

Abstract:

The Marmara and Northwestern Anatolia regions are known to be a transition zone from the strike-slip tectonics to the extensional tectonics. Although, the Sea of Marmara has been subjected to several active and passive seismic investigations, the accurate knowledge on the heterogeneity in the crust and upper mantle beneath the study area still remains enigmatic. On small-scale tomography problems, seismograms strongly reflect the effects of heterogeneities and the scattering properties of the Earth. Thus, the knowledge of high-resolution seismic imaging with an improved 3D radially anisotropic crustal model of the Northwestern Anatolia will enable better localization of earthquakes, identification of faults as well as the improvement of the seismic hazard assessment. For this purpose, 3D non-linear full waveform inversion methodology has been used to obtain an accurate image of the lithosphere and the upper-most mantle structure over an area of 37.5˚-42˚ N and 25˚-32˚ E and down to a depth of 471 km. The earthquake data were principally obtained from the Kandilli Observatory and Earthquake Research Institute (KOERI) and Earthquake Research Center (AFAD-DAD) database. In addition to this, some of the seismic waveform data extracted from the Hellenic Unified Seismic Network (HUSN) stations that are located within our study region were also used in this study. We have selected and simulated the waveforms of earthquakes with magnitudes Mw ≥ 4 occurred in the period of 2007-2014. In total, 3002 three-component regional seismograms from 95 events were used. The initial 3D earth model for the study region has been implemented from the multi-scale seismic tomography study of Fichtner et al. (2013). The synthetic seismograms were computed with forward modeling of seismic wave propagation by using spectral elements method (SEM). The complete waveforms were filtered at 8-100 seconds. The adjoint method is used to compute sensitivity kernels. The differences between observed and synthetic waveforms determined using time-frequency misfits. We minimize the data misfit using the conjugate gradient approach in an iterative manner. Findings of the future work of this study are expected to illuminate the previously unresolved shallow and deep features of Northwestern Turkey with better resolution and higher accuracy.