A three-dimensional model of the Sichuan basin was established by incorporating the Quaternary Chengdu plain, the near surface sediment, the crystalline basement, the 1D crustal velocity structure, and the realistic surface topography of the basin. Based on this model, and by using the spectral-element method and a parallel computing technique, the low frequency (0.05–0.5 Hz) wave amplification behavior of the Sichuan basin during the 2008 Wenchuan earthquake was simulated, and the frequency-related amplification characteristics were investigated. The results reveal that the Sichuan basin significantly amplified low frequency ground motions, with a maximum amplification factor of 9.6 located at the Zhuwa depression, which has the deepest Quaternary sediments. Compared to the vertical component, the horizontal component of the ground motion demonstrates greater amplification factors. The coupling of the basin-edge effect, the rupture directivity effect, and the interference between the waves from different subfaults results in strong local ground motions at the basin edge areas, such as Dujiangyan and Mianzhu. For the studied frequency band, the Quaternary Chengdu plain mainly amplifies seismic waves above 0.3 Hz, while the basin basement is sensitive to waves below 0.3 Hz. The Quaternary Chengdu plain significantly amplifies wave amplitudes and changes the distribution features of strong ground motions.

A stochastic finite‐fault approach based on corner frequency (EXSIM) is applied to simulate the Tottori M _w 6.2 earthquake. The parameter κ ₀ is calculated based on ground motion recordings. Other parameters, such as quality factor (127 f ^0.61 ) and stress drop (27.97 bars) are taken from our earlier work. The slip distribution refers to the results of Kubo et al. (2017, https://doi.org/10.1186/s40623-017-0714-3 ). The geometric spreading function and ground motion duration are taken from Atkinson and Boore (1995, https://doi.org/10.1785/bssa0850010017 ). The simulated results match well with the observed values in a short period ( T < 1 s). In addition, the effects of hanging wall (HW) and footwall (FW) on the simulated values are discussed. The results show that the simulated results of HW stations are more consistent with the observed values than those of FW stations. The differences between simulated Pseudo Spectral Acceleration (PSA) and observed PSA with epicentral distance and azimuth are also analyzed. The results show that the local site amplification and geometric location of stations influence the simulation results for soft soil sites. Overall, the simulated ground motions obtained by applying EXSIM approach matched well with the observed recordings which could be considered as the basis for earthquake‐resistant design during the post‐disaster recovery and could become a powerful tool for earthquake ground motion prediction.