Numerical modelling of ground motions in Eskişehir basın

Abstract

Eskişehir basin is located at the boundary of central and western Anatolia tectonic regions. Between two active faults it extends in EW direction with two open ends. So far deep velocity structure of the basin has not been well constrained however, average shear wave velocity for the top 30 m and sedimentary thickness estimations are available at various locations of the basin (e.g.,T¨un et al. (2016); Yamanaka et al. (2018); Ozel et al. (2020)). Number of strong motion recordings is rather limited due low seismicity of the region. The largest magnitude event that has ever been recorded within 150 km is the 2011 Simav Earthquake (Mw 5.9). Eskişehir city, with a popula tion close to a million people, has been expanding towards to this sedimentary basin. Long period ground motion is the concern of large scale structures that will be built at this region. Here we first present observed features of strong ground motions of this event recorded in the Eski¸sehir basin. Firstly, we observed that ground motion from the 19.05.2020 Mw 5.9 earthquake is governed by Rayleigh waves at periods longer than 0.5 s. Retrograde motion is visible almost at all basin- recordings. Among recorded waveforms, PGA and PGV of a basin- edge station (#2610 AFAD station) are formed by Rayleigh waves at periods 1 s. The longest significant duration of recordings is as high as 53 sec. Recorded spectral acceleration for 5% damping at spectal periods longer than 1 s is much higher than the one predicted by region specific ground motion prediction models. In the second phase, we showed formation of an experimental basin geometry utilizing linear interpolation of predominant frequencies at 95 measurement points. Dimensions of the model are 43 km \ 27 km \ 15 km. Basin layer continues across the entire model in EW direction, but bordered by northern and southern hills to mimic the geographical environment. Maximum depth is about 600 m. In the last phase we investigated the 3D wave propagation of small magnitude events, 17.01.2015 Mw 4.3 and 18.09.2015 Mw 3.7, occurred at northwestern part of the region and center of the basin, and compared with observed recordings for a possible validation of the velocity model. The computer code utilized in simulation relies on a finite difference modelling using staggered grids with nonuniform spacing. Ground motion simulation of the Mw 4.3 event reveals that the current velocity model overestimates the velocities in the eastern part of the basin in the NS direction, where E-W direction synthetics are generally smaller than the observed ones. On the other hand, synthetic velocities agree with observed ones at basin-center stations in the west. These findings suggest that more careful definitions of basin boundaries are necessary for the future models. Comparison of 1D and 3D simulation results also suggest that a 3D velocity model may produce longer and -more realistic- duration ground motions. The final step is to perform a blind simulation for the 20 February 1956 Mw 6.5 earthquake. The source was modeled by considering the ambiguities in the source parameters. The previous research was compiled to deal with unknown information about the mechanism and location of this event for consensus. We have compared the simulation outcomes with GMPEs models. The numerical simulation results yielded higher outcomes than estimated spectral ordinates by GMMs.