This study aims to predict intensity measures (IMs) at sites nearby seismic stations using recorded ground motions at seismic stations. Ground motions attenuate by distance and amplify/de-amplify due to site effects. We estimate IMs at a site by deconvolution-linear scaling-convolution method. The first step of this method is to perform the deconvolution to a seismic record observed at a seismic station at surface. The deconvolved ground motion represents a bedrock-equivalent ground motion by offsetting linear site effect of the site. The second step is scaling the amplitude of the bedrock-equivalent ground motion to the target site considering site-to-source distance. In this step we assumed that bedrock motions within one earthquake event can be linearly interpolated by distance. The third step is convolving the scaled ground motion to obtain surface ground motion including site effect at the target site. Using this method, the prediction of ground motion IMs would be accurate if the station and the site are identical, and the accuracy would be decreased if station-to-site distance increases. We tested this method using seismic stations located closely, and provide the effectiveness per separation distance between a station and a target site.

This paper presents a case history of one-dimensional time-domain nonlinear site response analysis for a major infrastructure project in the San José area, USA. The subsurface material of the study site consists of Holocene alluvial deposits with interbedded and crosscut clays and silts, sands, and gravels. Two levels of design earthquakes were considered herein associated with the return periods of 225 and 975 years. For each design level, site response analyses were performed to propagate a suite of eleven pairs of horizontal input ground motions at the reference soil horizon (a depth of about 75 m or 240 ft; V_S30 of about 490 m/s or 1600 ft/s) through a site-specific ground model. The site-specific ground model also accounts for strain rate effects on clay layers. A series of sensitivity evaluations were also completed for verifying and enhancing understanding of dynamic response of the study site. The site response analysis results show the site exhibits high nonlinearity, especially in clay layers, de-amplifying short-to-intermediate period content (0.01 to 1 s) while amplifying long period content. Also, a comparison of surface response spectra from the site response analysis versus ground motion models is presented herein highlighting the design implications of conducting site response analyses at soft sites, where high nonlinearity is anticipated.

Site terms of conventional ergodic ground-motion models largely rely on unified site proxies, such as the time-averaged shear-wave velocity in the upper 30 m (V_S30) and basin depth, which have been shown to induce large uncertainties at specific sites. Recent ground-motion model developers have used the peak frequency (f_peak) derived from horizontal-to-vertical spectral ratios (HVSRs) as an individual site parameter in ergodic site-effects models for central and eastern North America (e.g., Hassani and Atkinson, 2016, 2018). In this work, we demonstrate that using multiple site terms derived from geospatial data is advantageous in nonergodic ground-motion modeling by better representing site-to-site variability. We integrate nonparametric machine learning techniques and geospatial variables to develop fully data-driven nonergodic ground-motion models (GMMs). A decision tree ensemble method (gradient boosting model, GBM) is employed to predict PGA, PGV, and 5%-damped PSA using the NGA-West2 database. We examine the predictive power of 24 globally-available geospatial proxies for ground-motion modeling.

The 2016 M_w 7.1 Kumamoto earthquake induced widespread landslides, predominantly in the Aso area, approximately 20 to 40 km from the epicenter. Pulse-like ground motion (PLGM) was observed in this area. To understand the landslide causation in this event, we hypothesize that the velocity pulse of PLGM is the critical factor in triggering landslides. Firstly, the PLGM in the widespread landslide area is identified. Secondly, the characteristics of typical PLGM waveforms are analyzed. Finally, the Aso-Bridge landslide is taken as a case study to analyze the influence of PLGM on triggering landslides. The results indicate that the velocity pulses of PLGM have significant velocity amplitudes even though the acceleration is small. Moreover, the velocity pulses make up the major energy of these PLGM. Also, the simulation results of Aso-Bridge landslide show that both the original and extracted pulse motion can initiate landslides, whereas the residual ground motion cannot, thus validating the proposed assumption. Furthermore, high-frequency residual ground motion is difficult to trigger landslides even though the PGA is large, but its energy and PGV are low, indicating the limitation of slope stability analysis only considering the PGA.

The quantitative prediction of the seismic response of basins is difficult to achieve due to the highly two- and three-dimensional phenomena involved. Eurocode 8 parameterizes stratigraphic and topographic site effects by means of multiplicative coefficients. Expeditious methods are not available at present to account for basin effects in seismic design codes. Previous research suggests that two-dimensional aggravation factors may provide a convenient format to model two-dimensional basin effects starting from the simplified assignment of pseudo-spectral acceleration assigned from code-based approaches or one-dimensional site response analysis. Following on a previous study, this paper synthetically describes a work-in-progress approach to the quantitative calibration of two-dimensional aggravation factors for the seismic response analysis in basins. The approach relies on the development, implementation, and validation of a Python-based software suite which allows the generation of a large database of modelling schemes, the application of a set of seismic input motions, the conduction of 1D and 2D numerical analyses using the free OpenSeesPy package which leverages the modelling capabilities of OpenSees, and the subsequent statistical calibration of 2D aggravation factors. The general structure of the process is described, and initial example outputs are provided.

Most dynamic response analyses of rock slopes only consider the vertical incidence of seismic waves. However, in near-field seismic areas, obliquely incident waves are predominant. In the 2008 Iwate-Miyagi earthquake, numerous large-scale landslides occurred, leading to significant loss of life and property. Among them, the Aratosawa landslide is considered one of the largest. This paper evaluates the comprehensive impact of the seismic wave incidence angle on landslides. Based on the Discontinuous Deformation Analysis (DDA) method and combined with viscous boundaries, an Aratosawa landslide model was established. By comparing the seismic waves of vertical and oblique incidence, the occurrence, destruction, and consequences of the Aratosawa landslide were simulated. The results show that obliquely incident seismic waves lead to a larger seismic response. Furthermore, the slope failure simulation, considering the wave incident angle, can be well-analyzed using the DDA method.

Studies focusing on the numerical modelling of geotechnical profiles with potentially liquefiable soils will often focus on acceleration and pore pressure responses. If numerical models are being validated using experimental data, researchers usually need to compare measured and modelled responses in order to assess the modelling assumptions or to select a best-fit model. Early studies typically based this comparison on qualitative judgement that can introduce potential bias and may vary from person to person. Several researchers have subsequently proposed quantitative criteria to minimize the subjective influence and standardize the comparison process, with most focused on the comparison of acceleration time series. This paper presents the development and validation of a quantitative criterion to compare the experimental and numerical pore pressure time series. The criterion evaluates the goodness-of-fit between pore pressure time series as a weighted overall score based on five parameters: average difference, maximum pore pressure, time to reach steady pore pressure, pore pressure integral, and dynamic time warping. A sensitivity analysis is used to investigate how different factors, such as magnitudes and noise, influence the criterion evaluation. The quantitative results are then compared with subjective opinions to validate the criterion applicability. The results indicate that the proposed quantitative criterion evaluates consistent results as the qualitative judgement, and the amplitude misfit is the dominant factor that influences the goodness-of-fit.