The seismic effects of complex, deep, and inhomogeneous sites constitute a significant research topic. Utilizing geological borehole data from the Suzhou urban area, a refined 2D finite element model with nonuniform meshes of a stratigraphic crossing the Suzhou region was established. Within the ABAQUS/explicit framework, the spatial inhomogeneity of soils, including the spatial variation of S-wave velocity structures, was considered in detail. The nonlinear and hysteretic stress-strain relationship of soil was characterized using a non-Masing constitutive model. Ricker wavelets with varying peak times, peak frequencies (fp), and amplitudes were selected as input bedrock motions. The analysis revealed the spatial distribution characteristics of 2D nonlinear seismic effects on the surface of deep and complex sedimentary layers. The surface peak ground acceleration (PGA) amplification coefficients initially increased and then decreased as fp increases. The surface PGA amplification was most pronounced when the fp is close to the site fundamental frequency. Additionally, when fp = 0.1 Hz, the surface PGA amplification was found to depend solely on the level of bedrock seismic shaking, with amplification factors ranging from 1.20 to 1.40. Furthermore, the ensemble empirical mode decomposition components of seismic site responses can intuitively reveal the variations in time-frequency and time-energy characteristics of Ricker wavelets as they propagate upward from bedrock to surface.

In this study, the one-dimensional nonlinear Matasovic model was extended to the two- and three- dimensional stress space by replacing the shear strain with the generalized shear strain. The extended Matasovic model was subsequently combined with Dobry's excess pore water pressure generation model, and the reliability of the proposed loosely coupled effective stress analysis model was verified through undrained cyclic triaxial tests and a one-dimensional site response analysis. Finally, this method was applied to an engineering site in China to evaluate the nonlinear seismic response of liquefiable sites. The results show that the proposed effective stress analysis method can capture the dynamic behavior of the soil and the generation of pore water pressure during strong shaking. Moreover, there are differences between the proposed effective stress analysis method and the total stress analysis method for liquefiable sites, which indicates the need for careful selection in practical applications.