Abstract: This study proposes a method to simulate spatially correlated multi-point ground motions, incorporating local site effects using a frequency-dependent equivalent linearization model. The finite element-boundary element coupling method is employed to calculate the nonlinear seismic wave scattering effects in complex sites. Additionally, the amplification effects of ground motions in mountain-valley coupled sites are simulated and analyzed. The results are compared with those obtained from the traditional equivalent linearization method. The frequency-dependent equivalent shear modulus and damping ratio are determined from a standard strain spectrum. The results indicate that the frequency-dependent equivalent linearization method and the conventional equivalent linearization method exhibit similar overall trends. Both reflect significant spatial variations in ground motion within the mountain valley site. Furthermore, compared to linear conditions, both methods show a reduction in the peak ground acceleration and an extension of the predominant period of the response spectrum. However, the frequency-dependent method accounts for the frequency dependence of equivalent shear modulus and damping ratio, effectively improving the high-frequency response estimation that conventional methods often underestimate. Results show that the ground motion accelerations and response spectrum peaks derived from the frequency-dependent approach are markedly higher compared to those from the conventional method, by more than 23%. Additionally, the response spectrum peak shifts toward shorter periods, which better reflects the high-frequency behavior of seismic waves in soil layers. This method can reasonably optimize the high-frequency segment of site response, thereby enhancing the safety of engineering structures.

Key words: ground motion simulation, frequency-dependent equivalent linearization, local site effect, finite element-boundary element coupling method, mountain-valley coupled sites