Recovering accelerograms from response spectra can be helpful in several applications in earthquake engineering, such as hazard assessment and earthquake resistant design. However, this inverse problem is challenging because there is insufficient information in the response spectra to determine ground motion time histories. Supervised learning techniques of neural networks have been employed with wavelet packet transform to address this problem. Nevertheless, such approaches still require human-labeled training data, which can lead to memorizing the training set. An alternative approach to overcome the limitation is the use of the generative adversarial network (GAN), which is an unsupervised learning approach. GAN does not require any paired information like human-labeled data that has been used in computer vision fields to generate new data samples by capturing the underlying characteristics of the training data. The method uses random input in the form of latent vectors, which are processed by the generator model of GAN to produce a new data sample. This paper introduces the GAN-based method to produce accelerograms not only triggered by latent vectors but also conditioned on pseudo-spectral acceleration (PSA) that is additionally provided as input to the generator model. Several error metrics are utilized to compute the differences between the PSAs. The PSAs of generated samples generally match well with the input PSAs on which the generator model was conditioned, suggesting the proposed method is effective in solving the inverse problem.

Investigating the seismic response of earth dams is challenging but often controls decisions when evaluating existing dams. Dam fundamental frequency (f₀) and second mode vibration, key parameters in the dynamic response of a dam, can be assessed using the horizontal-to-vertical spectral ratio (HVSR) and standard spectral ratio (SSR) methods. The HVSR is mostly assessed by performing microtremor measurements in the field. However, when earthquake records are available, HVSR can also be calculated using the recorded motions. Additionally, standard spectral ratio (SSR) is calculated to assess the vibration characteristics of the dam. The overarching goal of this research is to assess vibration characteristics for dam sites in California. This is achieved by using the ground motion data on the California Strong Ground Motion Instrumentation Program (CSMIP) database for dam sites with recordings at different locations (i.e., right crest, left crest, etc.) and to study the local and site-specific features. Signal-to-noise ratios (SNR) were calculated to determine the usable frequency range and assess the quality of the ground motions. Then the earthquake-based HVSRs are evaluated using the following different approaches; i) A Fourier amplitude spectra ratio of the horizontal to vertical components (FAS_H/FAS_V) of the ground motion records at the surface using the intense S-wave portion and using the entire motion. Standard spectral ratio (SSR) which is the ratio of the FAS_H of the crest records over the FAS_H of the abutment records are also calculated. The vibration characteristics are evaluated based on different methods and the resulting estimate of the fundamental frequency and second mode of vibration for Briones Dam are presented.

Multi-point excitations are generally required to analyze the seismic response of spatially distributed structures, such as pipelines or long-span bridges. Simulated or synthetic ground motions, as a surrogate of recorded ones, have been widely used in earthquake engineering practice. When simulating ground motions for multi-point excitations, one of the key challenges is to appropriately represent the spatial correlation characteristics of acceleration-, velocity-, and displacement-time histories at neighboring sites. To address this issue, this study proposes a physical-based ground motion simulation method based on seismic physical process. Within this method, the kinematic finite-fault model is first employed to describe the source process, and the frequency-wavenumber Green's function is then utilized to calculate the propagation of seismic waves. Then, a frequency-domain adjustment is performed to ensure congruence with the design response spectrum. Ground motions at multiple sites can then be simulated under given scenario earthquakes, in which the spatial cross-correlation characteristics of these acceleration-time histories are properly quantified during the modeling of the physical propagation process of seismic waves. Moreover, by taking the 2019 moment magnitude (M_w) 7.1 Ridgecrest earthquake as a demonstrated example, synthetic ground motions are simulated using both the proposed and conventional methods, respectively. The simulation results are then compared with the data recorded from a dense array located at the Nevada National Security Site. It is found that the characteristics of the simulated motions of the proposed method are generally consistent with the recorded ground motions. Additionally, dynamic analyses are performed for a 300-m-span continuous box girder bridge using the simulated and recorded ground motions as the multi-point excitations, respectively. It is indicated that the difference of the dynamic responses by employing the simulated motions and recorded motions is generally smaller than 12%, validating the rationality of the method developed. Therefore, the proposed method can be readily used to simulate a set of ground motions at multiple sites, and to assess the seismic performance of spatially distributed structures by multi-point excitations.

Fueled by the advancement in sensor technology and increased seismicity, repeated observations of earthquake-induced ground shakings are accumulated at an unprecedented number of instrumented sites around the globe. Thus, the repeatable (or average) site responses at those recording stations can be disentangled from other systematic effects underlying ground-motion phenomena using regional-network-based approaches, e.g., generalized inversion technique or mixed-effect regression. This provides an excellent opportunity to gain new insights into the global variations in site response, which is impossible via regional studies only. In addition, site response inferred from weak motions based on Fourier amplitude spectrum (FAS), in comparison to that of response spectrum, can maintain the linearity of site response at all frequencies, facilitating a more straightforward physical interpretation. Capitalizing on the high-quality point-observations of Fourier site response in New Zealand, Japan, Europe, and California, we aim to unravel the physical drivers underlying the regional variations in site response, and infer globally consistent patterns, which are essential for developing predictive models transferable across regions. In this preliminary work, we report the global variations in the scaling of Fourier site response with commonly used site characterization parameters/proxies.

In irregular ground where engineering bedrock is inclined, seismic ground motions are locally amplified due to interference between body waves and surface waves generated secondary. In irregular ground where bedrock inclined in multiple directions, seismic ground motions may be more significantly amplified not only by 2-D effects but also by 3-D effects. In this study, we obtained the seismic observation records on actual irregular ground and tried to clarify the 3-D amplification effect of seismic motions by FEM analyses. Although we observed in a narrow area of about

300 m, we found large differences not only in the duration and maximum acceleration of the seismic records, but also in the peak frequency and peak magnification of their spectral ratios (at each site / bedrock outcrop site). We compared the transfer function of the 1-D ground with spectral ratios evaluated from the 2-D or 3-D analyses and confirmed that the amplification of seismic motions due to the 3-D effect of the irregular ground is significant at the ground.

The seismic ground motion is driven by the equation of motion, which depends on several parameters describing the earthquake source, the wave travel path, and the site of interest. We propose in this study to understand the relative importance of three shear wave velocities through a model combining global geophysical measurements with numerical simulations to estimate synthetic seismic motions sensitivity. Global sensitivity analysis (GSA) involves the identification and quantification of parameters contributions to output variability.

Recently, many earthquakes have occurred frequently in various countries in the world including Japan. In cases where the epicenter is located in mountainous area, a number of seismic slope failures tend to happen. But it is very difficult to analyze detailed mechanisms of each failure due to its wide range of impact. Simple and preliminary assessment of slope failure risk based on geological and geomorphological viewpoint will provide us with useful knowledge for future hazard quantification and countermeasures. It has been expected to develop a numerical system based on simple tools with use of only digital elevation model data and conventional regional geological information. In this study, a case study is demonstrated with use of a wide-area simple seismic slope failure risk evaluation system based on 3D dynamic FEM named as BESSRA, developed by the authors before. In the analysis, the soils in each layer follow the elasto-plastic model with considering cyclic loading effects. The proposed system has inexpensive specifications that do not require an environment such as high-performance parallel computers, owing that the concept of substructures in which computational processing is completed independently at each computational node are utilized. The effectiveness of the proposed system is verified through the simulation of observed slope failure distributions in the 2008 Iwate-Miyagi Nairiku Earthquake in Japan. Finally, a detailed discussion for improvement of the numerical modeling in the proposed system is presented.