There are a number of geotechnical earthquake engineering problems that require predicting the probability of a binary (Yes or No) outcome, typically using logistic regression or similar models. Two relevant examples are liquefaction triggering and surface fault rupture. The datasets used to develop these models often have imbalance in the Yes/No class ratio. The number of yes data points can outweigh the no datapoints by a large fraction. This is because finding true No data points is often very hard and requires careful investigations, whereas Yes data points are obvious and attractive to document and measure. Modelers are often concerned that this class imbalance might lead to biased or skewed results. However, they usually do not explicitly distinguish between class imbalance, the observed Yes/No ratio, and sampling bias due to something that causes potentially observed data to be excluded from observations. This paper examines the problem of sampling bias versus class imbalance and makes recommendations for when it needs to be addressed and when it does not influence the predictive capacity the models.

Using open topographic data before and after the damage caused by the 2018 Hokkaido Eastern Iburi Earthquake, the stability of a wide area slope in Atsuma-cho was evaluated using GIS software. Straight slip method was used to evaluate the stability of the slopes, with a view to linking the results with practical design. Calculation parameters for straight slip were determined based on the results of previous surveys and numerical elevation models before and after the disaster. As a result, although the range of unstable slopes obtained from the analysis covered the actual slope failure area, a wider range of unstable areas were extracted. The horizontal seismic intensity, Kh, was set to a value higher than the maximum value of the back-calculated value to ensure a safe evaluation, and this is assumed to be a contributing factor.

Sulawesi is a large island of Indonesia's archipelago which has experienced major earthquakes. On January 15th, 2021, West Sulawesi was struck by a shallow earthquake Mw 6.2 at 02:28:21 local time. Thousands of structures collapsed and suffered serious damage as a result of the earthquake, killing hundreds of people. Mamuju's seismicity is poorly understood, and the studies that have been done on it have only covered a small portion of it. As a result, the objective of this research is to determine how the city's local geology and soil conditions affect the intensity of ground motion, which is measured in terms of its amplitude, frequency, and duration, and how these factors relate to the city's damaged buildings as a result of the 2021 Earthquake. Therefore, a series of geotechnical drillings consisting of borehole-N SPT measurements were undertaken on the studied area in the city. Shear wave velocity profiles of the ground were derived through MASW measurements. Time histories of the 2021 Mamuju Earthquake were derived from the data of Indonesia’s geophysical agency (BMKG). Site-specific ground response analysis was conducted within equivalent elastic linear (EL) approaches to model ground time histories in the 2021 Mamuju Earthquake. A non-linear time histories FEM modeling of typical structures in the city within three stories buildings was undertaken based on generated ground time histories of the 2021 Mamuju Earthquake. The seismic amplification in the earthquake event and its relation to the numerous cases of buildings collapsing was evaluated. The results of this study reveal that typical ground in the city has a resonance frequency of 0.5 –0.8 Hz, with an amplification factor of 2.3. This seismic amplification has resulted in structural consequences in the typical RC building in Mamuju, in which several columns in the 2^nd and 3^rd stories have beam-column capacity ratios exceeding 1.00. The results may indicate the high case of buildings collapsed during the 2021 Mamuju – Majene Earthquake.

Earthquakes (abbreviated as EQ in the paper) are catastrophic events resulting from the rupture of faults deep below the ground surface. Past earthquakes in the Himalayan region caused severe damage to various sites in terms of high ground acceleration, leading to subsidence and liquefaction. Due to a prolonged locking period despite continuous subduction of the Indian plate under the Eurasian plate, scientists dread future earthquakes in the Central Himalayas that would gravely affect Eastern India. Researchers communicated about repeated liquefaction events at Madhubani, a city in Bihar state of Eastern India, due to earthquakes in the Central Himalayas. This paper takes into account specific parameters like the moment magnitude (M_W) of the earthquake, focal depth (D), stress drop (∆σ), hypocentral distance (R) to Madhubani, and the average fault slip (d) at the source. These factors serve as input variables towards determining the individual and collective variance due to a seismic event triggered in the Himalayas. Computation of parametric influence from such a cluster of parameters tends to be a tedious and complex process. Hence, the current study employs Principal Component Analysis (PCA) to reduce the dimensionality of the data and estimate the contributions of various influencing factors to induce a peak acceleration value at Madhubani. The principal components obtained from the analysis enable understanding the extent of contribution by various input features on the dataset. Reducing the redundancy in the acquired dataset helped propose a relationship between stress drop (∆σ) and average fault slip (d). The developed relationship turns out to precisely accommodate stress drops for earthquakes with magnitudes 2-7.2 .

In most large-scale seismic hazard and risk applications, local site conditions are inevitably taken into account through site models inferred from proxies, such as geology and terrain. In this study we examine the effect of applying proxy-based Vs30 models in PSHA analyses, focusing on the region affected by the M7.8 Pazarcik earthquake of February 2023 in Türkiye. To this end, we compare PSHA results obtained from two different proxy-based Vs30 models for the region with those obtained from a more precise site model developed from Vs30 data from AFAD-TADAS. The results indicate that for large-scale applications, and in exception of prominent near field conditions, the site proxies may provide a valuable tool in PSHA studies. In addition, a comparison of the PSHA results with the recorded ground motions allows for a quantitative estimate on how rare this earthquake event was.

We define the physical processes that control the style and distribution of ground surface ruptures on thrust and reverse faults during large magnitude earthquakes through an expansive suite of geomechanical models developed with the distinct element method (DEM). Our models are based on insights from analog sandbox fault experiments as well as coseismic ground surface ruptures in historic earthquakes. DEM effectively models the geologic processes of faulting at depth in cohesive rocks, as well as the granular mechanics of soil and sediment deformation in the shallow subsurface. We developed an initial suite of 45 2D DEM experiments on dense, 5.0 m thick sediment in a model 50 m wide with a fault positioned 20 m from the driving wall and slipped each model at a constant rate (0.3 m/s) from 0 to 5.0 m. We evaluated a range of homogeneous sediment mechanics (cohesion and tensile strength from 0.1 to 2.0 MPa) across a range of fault dip angles. In addition, we examined various depths of sediment above the fault tip. Based on these experiments, we developed a classification system of the observed fault scarp morphology including three main types (monoclinal, pressure ridge, and simple scarps), each of which can be subsequently modified by hanging wall collapse. After this initial suite of models, we generated an additional 2,981 experiments of homogeneous and heterogeneous sediment in dense, medium-dense, and loosely packed sediment across a wide range of sediment depths and mechanics, as well as a range of fault dips (20 – 70º). These models provide robust statistical relationships between model parameters such as the fault dip and sediment strength mechanics with the observed surface deformation characteristics, including scarp height, width, and dip as well as the tendency for secondary fault splays. These relationships are supported by natural rupture patterns from recent and paleo-earthquakes across a range of geologic settings. In conjunction with these natural examples, our models provide a basis to more accurately forecast ground surface deformation characteristics that will result from future earthquakes based on limited information about the earthquake source and local sediment properties.