Session: Fiber-optic Sensing Applications in Seismology - II

Type: Oral

Room: Key Ballroom 9

Date: 4/15/2025

Presentation Time: 10:30 AM Pacific

Presenting Author: Hilary Chang

Description

We use Distributed Acoustic Sensing (DAS) on the borehole optical fibers to characterize microearthquakes at the Cape Modern geothermal field in Utah. The Cape Modern geothermal site is an enhanced geothermal system. In 2024, a series of stimulations at a 2.5 km depth generated events with M<3 between 1.5 and 3.5 km depth. The downhole fibers provide a dense network that records the wavefield between 0.5 and 2.5 km depth. The body-wave event spectra contain the high-frequency (< 500 Hz) energy required to characterize these small events. We use these spectra to estimate spectral stress drop, source radius, and moment magnitude with Brune’s source spectral model. The spectral-ratio approach for source modeling is advantageous as it accounts for the wave propagation effects and instrument responses using a co-located small event as the empirical Green’s Function (eGF). However, a qualified eGF-target event pair is often not available for microearthquakes. For source modeling using individual spectra, we need the medium properties to estimate the wave propagation effects. We obtain depth-varying attenuation, which was rarely directly observed from seismic data before DAS, by deconvolving event arrivals between DAS channels. The attenuation decreases with depth in the top 1.5 km of sediments. We investigate the source parameters as well as their variabilities between events and channels to estimate the uncertainties and to understand if there is a potential difference in the source physics. We examine the conversion between DAS strain and particle motions, which might distort the shape of the spectra depending on the gauge length and the medium properties. Stress drop for small events has been difficult to measure as shallow attenuating sediments suppress the high-frequency signals and limit the bandwidth crucial for characterizing small events. The downhole DAS, which provides broadband observations and a dense network for sampling the wavefield and the depth-varying medium properties, has great potential to advance our knowledge of the physics of microearthquakes.

Additional Authors

Session: Station Installations and Site Conditions, a Quest for Improved Strong Motion Database [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/16/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Junju Xie

Description

The updating of seismic ground motion parameter zoning map requires more accurate site classification maps to account for adjustments related to site conditions. When site investigations are not feasible, proxy-based estimations of V_S30 are typically used to evaluate site conditions. We compiled a nationwide borehole database from research documents and technical reports, with approximately 82.02% of the profiles having depths greater than 20m. Unlike traditional approaches that primarily consider surface characteristics, we developed a hybrid V_S30 model considering both surface and subsurface site information. Our new model uses surface geology category and subsurface bedrock depth as indicators. The geological units across China were classified into 33 Quaternary deposit units and 5 rock units according to geological age and genesis. Combined with 1:500,000 geological maps and a 100-m resolution depth-to-bedrock map, we generated a nationwide V_S30 map. We made regional adjustments based on the geological environmental zoning and developed a 100-m resolution Chinese site classification map. Validation through borehole data and comparisons with previous models demonstrated that the V_S30 results and Chinese site classification derived from our model effectively classify site conditions on a national scale. Additionally, in plain areas, our model outperforms slope-based ones in distinguishing between Chinese site classes II and III, as well as NEHRP classes C and D.

Additional Authors

Session: Station Installations and Site Conditions, a Quest for Improved Strong Motion Database - I

Type: Oral

Room: Key Ballroom 11

Date: 4/16/2025

Presentation Time: 03:00 PM Pacific

Presenting Author: Silvia Castellaro

Description

Seismic methods based on surface waves established in use about 25 years ago. 20 years ago, the European project SESAME (Site Effects assessment with AMbient Excitation, 2005) had the great merit of standardizing processing and interpretation of the single-station microtremor data. The project ended with the formulation of 9 criteria that established the reliability of an H/V curve and the clarity of a possible peak in the same curve. However, much time has passed since then and new knowledge has been added, to the point that some of the SESAME criteria could be reformulated today.

SURVEY PLANNING & SITE SELECTION. Before any interpretation of an H/V peak, it is necessary to understand the nature of the H/V peak itself, i.e. whether it is linked to surface waves or a true resonance. The two cases are easily distinguished in the microtremor spectra; however, this determines a different approach in the interpretation of H/V peaks.

SITE SELECTION & SENSOR INSTALLATION. Contrary to what the SESAME guidelines suggested, it has been shown that the measurement point influences the output. Rigid artificial layers break down the horizontal spectral components of motion, leading to H/V ratios < 1 even up to disappearance of H/V peaks.

Tremor measurements are easily affected by structures surrounding the investigated site. It is not difficult to distinguish H/V peaks related to nearby buildings from should be introduced in an ideally renewed set of guidelines.

DATA ANALYSIS. Data analysis always requires the removal of spurious measurements (outliers). This is traditionally carried out based on STA/LTA algorithms, but these end up removing too much of the signal than necessary.

DATA INTERPRETATION AND CONCLUSIONS. We will review the issues above and show that there is more information in the individual component spectra (H & V) than in the H/V spectral ratio, to the point that while the former can be interpreted without the latter, the latter is almost not interpretable without the first. We will review the 9 SESAME criteria, suggesting those that can still be considered valid today and those that could be modified.

Additional Authors

Session: The Landscape Record of Earthquakes and Faulting [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/16/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Nadine Reitman

Description

The landscape record of earthquakes is a primary tool to interpret the location, displacement, and recency of active faulting. Landscapes evolve after earthquake rupture from both natural and anthropogenic causes (e.g., precipitation, gravity, dirt bikes), causing erosion and modification of vertical fault scarps and laterally offset features. The 1992 M7.3 Landers, California, earthquake offers an opportunity to quantify the degradation of lateral and vertical displacements that were well documented post-earthquake (Arrowsmith & Rhodes 1994, McGill & Rubin 1999). We focus on an ~500-m-long section of the Emerson fault scarp near Galway Lake Road that had up to 5 m right-lateral and 1.75 m vertical displacement with areas of localized and distributed deformation. We document landscape change over 30 years using fault-parallel and fault-normal trenches and vertical differencing of topography derived from 1992 USGS aerial photographs taken two days after the earthquake and 2022 drone photogrammetry. The 1992 colluvial wedge in the fault-normal trench is 30-50 cm thick at the base of the vertical scarp, diminishing to 10-15 cm, 1.5 m down-trench from the base of the scarp. This wedge overlies a basal rock line, interpreted to be the ground surface prior to the 1992 earthquake, followed by chaotic debris deposits, and topped with finely bedded slope wash, burying ~1/3 of the original free face. The fault-parallel trench exposes a new channel incised since 1992, a channel offset in the 1992 earthquake, and a broad wash containing a thalweg possibly offset in the penultimate event. Vertical differencing and field observations document extensive rills across the scarp (some with apparent left-lateral offset due to flow patterns), headward-eroding channels deeply incised into what were broad washes in 1992, and the formation of post-1992 channels on the downthrown block. Evidence of most distributed deformation (cracks and minor faults) is erased in 30 years. The results of this work will inform how landscapes are interpreted for evidence of active faults, with implications for fault displacement hazard models.

Additional Authors

Session: Accuracy and Variability of Physics-based Ground Motion Modeling - II

Type: Oral

Room: Key Ballroom 11

Date: 4/15/2025

Presentation Time: 04:45 PM Pacific

Presenting Author: Ke Xu

Description

The Statewide California Earthquake Center (SCEC) Broadband Platform (BBP) is an important tool for structural and geotechnical earthquake engineering applications, as well as for probabilistic seismic hazard analysis, with the capability to simulate ground motions up to 25 Hz. The BBP models scenario earthquakes by a combination of physics-based ground motion simulations at low frequencies (typically up to 1 Hz) and stochastic signals for higher frequencies. However, a notable limitation of this approach is the use of 1D velocity models for the low-frequency component, which do not capture the complexities introduced by 3D sedimentary basins and complex topography. These factors may significantly influence local peak ground motions and durations, potentially leading to inaccuracies in seismic hazard assessments.

To overcome these limitations, we have integrated into the BBP the 3D physics-based ground motion simulation code AWP-ODC with support for topography (O’Reilly et al., 2022), frequency-dependent anelastic attenuation (Withers et al., 2015), statistical distributions of small-scale heterogeneities (Savran and Olsen, 2016) and extended the physics-based simulations' maximum frequency to 2 Hz. We extract 3D velocity models for California using the SCEC Unified Community Velocity Model (UCVM) and compare ground motion predictions from 1D and 3D simulations to Ground Motion Models and recordings of the 1989 Loma Prieta, 2019 Ridgecrest and 1994 Northridge earthquakes in northern, central and southern California, respectively. Then, we simulate a series of M7-7.7 scenario earthquakes in California, combining the 3D deterministic simulations with stochastic signals up to 25 Hz. The results will identify and quantify deficiencies in existing hazard models and are expected to contribute to more accurate and comprehensive seismic hazard assessment across California.

Additional Authors

Session: Unusual Earthquakes and Their Implications [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/16/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Harriet Z. Yin

Description

The 2021 Mw7.2 Haiti earthquake was a devastating event which occurred within the Enriquillo Plantain Garden Fault Zone (EPGFZ). It is not well-understood why neither the 2021 nor the prior 2010 Mw7.0 earthquake were simple strike slip events and, instead, ruptured with distinct patches of dip slip and strike slip motion on largely separate fault planes. We develop several 3D dynamic rupture simulations of the 2021 earthquake to analyze which conditions may have controlled the complex rupture. The major characteristics of the earthquake rupture include: the characteristic spatial and temporal separation of strike-slip and dip-slip motion, rupture transfer to the Ravine du Sud Fault (RSF), and a multi-peak source time function. We construct a detailed fault system geometry comprising 17 fault segments, which includes a north-dipping Thrust Fault (TF) and near-vertical RSF, along with surrounding regional and secondary faults. We find that along-strike changes to the frictional strength of the TF are needed to focus the slip to reproduce the scale and pattern of deformation observed with InSAR. Lateral changes in the regional stress shape and orientation are key to reproducing the observed rupture transfer from the TF to the RSF while maintaining the rake required to reproduce the broad InSAR surface deformation pattern and multi-peak source time function. The dynamic rupture modeling results suggest that significant variability in fault stress and strength as well as complexities of the subsurface geometry may have been key controlling factors on the dynamics of the 2021 rupture.

Additional Authors

Session: Earthquake Shaking and the Geologic Record: Triggered Phenomena and Preserved Fragile Geologic Features - I

Type: Oral

Room: Holiday Ballroom 1

Date: 4/17/2025

Presentation Time: 02:45 PM Pacific

Presenting Author: Ramon Arrowsmith

Description

We investigate the 3D geometry and dynamic behavior of precariously balanced rocks (PBRs) for fragility modeling. To accurately map and analyze fields of PBRs, we use a combination of advanced technologies, including unpiloted aerial vehicle (UAV)-based Structure-from-Motion, UAV lidar, and iPhone lidar. We developed a method combining machine learning and deterministic segmentation for precise 3D extraction of PBR geometry. The machine learning component employs a deep neural network to detect PBRs in 2D imagery, facilitating their localization in point clouds. Deterministic segmentation utilizes a region-growing algorithm to separate PBRs from their pedestals, guided by user-provided basal contact curves at their interface. This cascading approach enables high-precision geometry extraction. Using the extracted 3D geometries, we perform PBR overturning analysis with the Virtual Shake Robot v2.0 (VSR2). VSR2 is built on the PyBullet physics engine and the Robot Operating System framework, validated against physical PBR overturning experiments conducted on a large-scale shake table. Our work demonstrates cutting-edge tools for PBR mapping and fragility modeling, providing critical insights into local ground motion constraints inferred from fragile geological features. Ultimately, we aim to offer an automated solution for using PBRs to refine seismic hazard curves and enhance our understanding of regional seismic risks. We apply the toolkit to the PBR fields in the Sierra Nevada, east of Fresno, California, relatively far from known fault sources. The PBRs in this region are glacial erratics, deposited by the Tioga (26–19 ka) and Tahoe (60 ka) glaciations. This recent glaciation has stripped the landscape of typical evidence used to identify active faults, and the existence of the PBRs will hopefully help constrain seismic hazard in the area.

Additional Authors

Session: Earthquakes, Lithospheric Structure, and Dynamics in Stable Continental Region [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Xiaotao Yang

Description

Despite being relatively stable, the interiors of global cratons preserve evidence of structural modification on geologic timescales. The North American midcontinent is the locality of a series of rifts, basins, arches, and domes. Geologic evidence suggests that these features formed throughout the Paleozoic or earlier with limited subsequent deformation. However, the mechanism(s) that drove the uplift and subsidence of these intracratonic features is poorly understood. Linking these surface tectonic features with deep crustal and upper mantle processes is important in understanding the evolution of cratons. Here we present a new 3-D shear-wave velocity model of the crust and uppermost mantle (10-100 km) of the North American midcontinent, using full-wave ambient noise tomography. Our results show a heterogeneous crustal structure across the midcontinent, with scattered low-velocity zones in the uppermost mantle. These low-velocity zones are concentrated below major rifting structures, including the Midcontinent Rift, the central Michigan Basin, the La Salle Deformation Belt within the Illinois Basin, and the Reelfoot Rift. Except for below the Reelfoot Rift, where a localized thin crust is imaged, these low velocities are imaged below a thicker crust than the regional average. These low-velocity anomalies may reflect the remnants of magmatic underplating or frozen hydrated mantle lithosphere, formed during the early rifting stage. These materials were stagnant when the rifting stopped. Below the Michigan and Illinois Basins, the relatively deep Moho may be the consequence of subsidence in the basins. These observations suggest that rifting of the midcontinent created lithospheric weak zones, serving as the pathways for magma or other hydrothermal fluids or partial melts to rise and accumulate below the crust. The findings in this study imply strong modifications within the cratonic interior that are preserved in the present-day lithospheric structure.

Additional Authors

Session: Advancing Time-dependent PSHA and Seismic Risk Assessment: Accounting for Short- to Medium-term Clustering [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/15/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Trey Apel

Description

Although the science has shown for over a century that earthquakes are fundamentally a time-dependent process, hazard and risk analyses have generally been stuck in a time-independent framework. At CoreLogic, we have implemented time-dependent hazard and vulnerability models that provide a better assessment of the risk earthquakes pose. We will present our 2025 time-dependent earthquake model and compare it to the time-independent approach in California and the Pacific Northwest. We will show how hazard and loss change between the two models in regions of high risk, like Los Angeles, the San Francisco Bay area, and the Pacific Northwest. We also show how memory can be incorporated into vulnerability functions to better assess the cumulative damage effects of a mainshock and its subsequent aftershocks.

Additional Authors

Session: Advancements in Forensic Seismology and Explosion Monitoring [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Miro Ronac Giannone

Description

Under certain entry velocities and initial mass conditions, the interaction of a meteoroid with dense regions of the Earth’s atmosphere can produce bolides, which refer to detonating fireballs that explode at resolvable altitudes within the atmosphere. These atmospheric phenomena can generate infrasound at rates which can be processed and subsequently leveraged towards better characterization of meteoroids when other means of observation are not available. Classification of bolide infrasound is regulated by the source geometry, which corresponds to the point along the trajectory which induced a shock wave. Shock waves resulting from a fragmentation event could have spherical or quasi-spherical geometry, whereas infrasound generated via the meteoroid’s hypersonic flight has a characteristic cylindrical line source geometry. Infrasound’s low-frequency content facilitates long-range propagation at global scales. Interpretation of processing results is oftentimes complicated due to the time-sensitive nature of the atmosphere causing irreversible effects on waveform morphology.

In this study, we utilize a data-driven approach to improve characterization of bolides documented within NASA’s Center for Near-Earth Object Studies exclusively with infrasound detected by global arrays. We implement a multi-frequency array processing software, known as Cardinal, to extract directionality estimates and phase velocities, as well as quantify waveform coherence over a broadband frequency range using narrowband spacing. When appropriate, we utilize the adaptive array algorithm to better characterize the signal over sequential frequency bands and subarray configurations. Furthermore, we measure signal delay times using known infrasound propagation velocities and a start time coinciding with bolide peak brightness, typically associated with a fragmentation episode. Our work comprehensively analyzes bolide infrasound as a function of source-receiver distance, signal arrival times, event altitude, and total impact energy.

Additional Information: SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.

Additional Authors

Session: Modern Waveform Processing and Engineering Datasets - Accessibility, Quality Control, and Metadata [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Balakumar Anbazhagan

Description

The horizontal-to-vertical spectral ratio of ambient noise (mHVSR) has been used to develop proxies for site effects for many decades. Recent studies have considered the use of mHVSR as a predictor of site effects and parameters derived from mHVSR (e.g., the frequency corresponding to the peak amplitude) have been implemented into ground motion models for predicting site effects. These advancements create a need for obtaining mHVSR at strong motion recording sites. Recording ambient noise, using temporary broadband seismometers, at each of the strong motion stations would be ideal, but it would require an enormous manual effort. Moreover, studying time-dependent variations in ambient noise requires deployment of the seismometer for an extended duration. Alternatively, continuous noise measurements from the permanent seismic stations can be used for obtaining mHVSR. Using the continuous noise measurements from permanent seismic stations drastically reduces the manual effort in data collection as the noise data are continuously archived for most of the seismic networks. The objective of this research is to compile a database of mHVSR obtained from noise measurements at permanent seismic stations in the United States. In the light of the development of new generation of ground motion models for shallow crustal earthquakes (NGA West 3), the compiled HVSR data could be useful for modeling site response. In this research, we compile the three-component noise recordings from approximately 4,200 permanent seismic stations. Three-hour noise recordings were processed using the open source HVSR processing tool hvsrpy. Quality control checks are necessary to identify cases in which the computed HVSR is affected by non-stationary effects or by potential instrument malfunction. However, the quality control checks are difficult to automate, and often requires manual examination. After comprehensive quality control checks, the resulting database containing HVSR curves and noise time series will be published in DesignSafe data repository. It is hoped that the database will aid in the development of robust site response models.

Additional Authors

Session: Improving the State of the Art of Earthquake Forecasting Through Models, Testing and Communication [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/15/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Leila Mizrahi

Description

The Epidemic-Type Aftershock Sequence (ETAS) model is the state-of-the-art framework used for operational earthquake forecasting worldwide. It models earthquake occurrence as a stochastic point process, where earthquakes trigger cascades of aftershocks according to a set of empirical laws governing the timing, location, and quantity of direct aftershocks, as well as the magnitude distribution. These laws are described analytically and depend on several region-specific parameters that characterize the triggering relationships between earthquakes and their direct aftershocks. However, due to the stochastic nature of the cascading process, the distribution of indirect aftershocks has traditionally been obtained only through extensive simulations—a process that is both time- and resource-intensive.

In our recent study (Mizrahi and Jozinović, 2024), we demonstrated that the mean number of expected indirect aftershocks, given the parameters governing the direct aftershock distribution, can be accurately estimated using a simple deep-learning model. The main advantage of our approach compared to the simulation-based approach is the significant speed-up of the calculations, with the ML model taking only 813 µs on average, without a loss in accuracy. Here, we extend this approach to estimate not only the mean but the full distribution of higher-order aftershock numbers, modeling it as a negative binomial distribution. While estimating the full distribution is more challenging than estimating the mean, it is a crucial step toward enabling the faster machine-learning-based approach to replace simulations without sacrificing valuable information about forecast uncertainty. We demonstrate that this extended model can infer the full distribution of higher-order aftershock numbers accurately orders of magnitude faster than the classical simulation-based approach.

Additional Authors

Session: Advances in Reliable Earthquake Source Parameter Estimation - III

Type: Oral

Room: Key Ballroom 12

Date: 4/16/2025

Presentation Time: 02:00 PM Pacific

Presenting Author: Yongsoo Park

Description

The impulsiveness of an arrival is one of the metrics that are often labeled along when picking seismic phase arrival times. This can be considered as the slope of the seismogram at the arrival, which can theoretically (but not realistically) range between 0 (perfectly emergent or parallel to the time axis) and 1 (perfectly impulsive or perpendicular to the time axis). However, due to the broad frequency contents and the noise in seismograms, the slope cannot be measured with confidence in most cases. Hence, this continuous value has been labeled subjectively by choosing from a set of choices – typically either impulsive or emergent (i.e., binary). Here, we propose a simple approach for converting these binary labels to continuous values. We train neural networks for binary classification using the binary impulsive/emergent labels. We show that the two classes cannot be separated even if we optimize the hyperparameters of the models to maximize the classification accuracy, which demonstrates the measure of impulsiveness is indeed a continuous value. We also show that different neural networks output almost consistent classification scores with most samples having a standard deviation of less than 0.1. We further evaluate the approach by altering the impulsiveness and signal-to-noise ratio of an arrival and show that the changes in classification scores align with our intuition. This classification score (or the ‘impulsiveness’) can serve as a measure of difficulty for phase arrivals, which can be used as weights when training phase picking models or for various geophysical inversion problems that rely on arrival time measurements.

Additional Authors

Session: Mechanistic Insights into Fluid-induced Earthquakes from the Laboratory to the Field [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/15/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Jeremy M. Gosselin

Description

The parameters of the Gutenberg-Richter relation, particularly the b-value, are used to characterize the source behavior of seismicity given an earthquake catalog. Many studies have considered spatiotemporal variations in the b-value, which are interpreted to be linked to the accumulation of stress near active faults in Earth’s crust. The b-value is also a valuable input for earthquake hazard assessment. Most methods for estimating variations in the b-value rely on subjective choices of key parameters, such as the minimum and maximum magnitude values and spatiotemporal windowing, which can significantly affect the results. Recent work has introduced a probabilistic trans-dimensional approach, where windows are self-defined according to the information contained within the earthquake catalog. Existing methods require prior definition of the magnitude of completeness of the earthquake catalog, such that only events above this magnitude threshold are considered in the analysis. Here, we develop a general probabilistic, trans-dimensional approach that uses the entire (incomplete) earthquake catalog to estimate models of the spatiotemporal variations in the b-value as well as catalog incompleteness (i.e., decay of the distribution at high and low magnitudes). We demonstrate our method by analyzing several simulated and real earthquake catalogs, including natural and induced seismicity in northwestern Canada. Our results show that variations in catalog completeness can bias estimates of b-value variations when not accounted for, and vice versa. Our results highlight the effectiveness of a general probabilistic trans-dimensional approach in considering the entire earthquake catalog to obtain robust estimates of spatiotemporal variations in recorded seismicity. This will lead to more reliable assessments of seismic hazards, particularly for regions of frequent induced seismicity, such as in western Canada.

Additional Authors

Session: Fiber-optic Sensing Applications in Seismology [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/15/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Yapo Abole Serge Innocent Oboue

Description

Distributed Acoustic Sensing (DAS) is a powerful tool in seismic monitoring, enabling high-resolution strain detection over long distances with fiber-optic cables. This cost-effective technology has diverse applications in seismology, but DAS recordings are often affected by erratic and horizontal noise, compromising their effectiveness in seismic analyses.

To address these challenges, we propose a hybrid denoising framework that combines traditional signal processing techniques with advanced deep learning. The core of the framework builds on Deep Image Prior (DIP), an unsupervised deep learning method that excels at mitigating random noise. However, DIP struggles with complex, overlapping noise types due to limitations in network design and hyperparameter sensitivity. To overcome these issues, we integrate DIP with a Multi-Head Attention Regression Network (MH-ARN), which incorporates fully connected layers and multi-head self-attention blocks to capture significant details in DAS data.

Our approach starts with dynamic patch-based processing using Non-Local Means (NLM) for high-frequency noise suppression. A variance-based patch selection targets signal-rich regions for training. Horizontal noise is mitigated through a frequency-wavenumber dip (FK-dip) filter applied after the first iteration, with iterative MH-ARN refinements ensuring effective noise suppression.

Applications to synthetic and real-world DAS datasets from the FORGE geothermal field show that our framework significantly improves the signal-to-noise ratio while preserving key details. It outperforms conventional DIP and integrated denoising methods, which combine operators like bandpass, structure-oriented median, and FK-dip filters. This multi-step process enhances signal quality and overcomes the limitations of traditional DIP methods, offering a robust solution for complex noise in DAS applications.

Additional Authors

Session: Data-driven and Computational Characterization of Non-earthquake Seismoacoustic Sources [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/16/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Albert R. Kottke

Description

Random vibration theory (RVT) is a robust probabilistic tool for the bidirectional conversion between Fourier amplitude spectra (FAS) and expected pseudo-spectral acceleration (PSA) without explicit time domain information. In practice, RVT allows for the development of ground-motion models in the FAS domain and the computation of associated time domain peak values (e.g., PSA) for engineering analyses. This approach offers several advantages over traditional methods due to the inherent limitations of PSA, such as oscillator frequency interdependence, which presents challenges in ground motion development, limits adaptability, and restricts the utilization of small-magnitude motion data. However, earthquake ground motions do not meet the assumption of stationarity (i.e., time-invariant statistics), a fundamental assumption of RVT. Various authors have proposed adjustments to RVT to address this deficiency. These adjustments, which are typically developed using simulated time series, rely on additional parameters and often fail to capture the effects of spectral shape due to site response or regional differences. We present a new correction for RVT based on the moments of the FAS of the oscillator response. This approach is based on the theory of transient random vibrations and is calibrated to recorded earthquake ground motions. By using the moments of the FAS in the calculation of the correction factor, the factor changes as the shape of the FAS changes. The calibration is performed using records from the NGA-West2 database at multiple damping levels (0.5 to 30 %), and then tested against the NGA-East and a preliminary NGA-West3 databases. The results show that the proposed correction is robust.

Additional Authors

Session: Recent Advances in Modeling Near-source Ground Motions for Seismic Hazard Applications [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/16/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Grace A. Parker

Description

We develop a model for the near-source saturation of ground motions for shallow crustal earthquakes using the NGA-West3 database of effective amplitude spectra (EAS, the smoothed quadratic mean of two horizontal-component Fourier amplitude spectra). Observations of ground motions show amplitude saturation at the closest source-to-site distances. This is a geometrical effect: at near-fault stations, the relative distance between the closest and farthest portions of the rupture is larger. Large-magnitude earthquakes have larger fault ruptures, and a larger proportion of those ruptures are also farther away from any given near-fault location, causing stronger saturation. Ground-motion models account for this by using a magnitude-dependent modification to the distance metric (here termed h), sometimes referred to as a pseudo-depth term. Most work on this topic has been performed using the distance to the closest point on the rupture surface, R_Rup. Here we use the closest distance to the surface projection of the fault, the Joyner Boore distance, R_JB. However, no magnitude-dependent models for h compatible with R_JB exist. We develop an empirical h model that depends on moment magnitude, M, depth to top of rupture, z_TOR, and frequency, f, assuming a geometrical spreading of 1/R and adjusting near-fault data for site response and Q using fits to far-fault data. We parameterize a function of h with M that has a f-dependent breakpoint, with a constant h value for large M above the breakpoint, where z_TOR is 0. Below the breakpoint for small magnitudes there is a f- and z_TOR-dependent slope. Using a grid-search approach we find parameters that minimize the L1 norm for records with R_JB ≤ 10 km. Our model predicts h values for large M, where R_JB = R_Rup, that are consistent with previous models for response spectra (~15-30 km). However, we find that saturation distances for smaller M depend on f and z_TOR, where shallow earthquakes and high frequencies have shorter saturation distances, a previously unmodeled behavior. Overall, our new model fits the data better compared to existing models without these added input parameters.

Additional Authors

Session: Earth’s Structure from the Crust to the Core - IV

Type: Oral

Room: Key Ballroom 9

Date: 4/16/2025

Presentation Time: 05:15 PM Pacific

Presenting Author: Meichen Liu

Description

The Preliminary Reference Earth Model (PREM) is a widely used seismic reference model in seismology. However, global SS-S and PP-P differential travel times have bimodal ocean/continent distributions with mean values of +3 s of both PP-P and SS-S. This suggests that PREM represents the seismic structure of the continental upper mantle better than the oceanic upper mantle and that, on average, PREM overpredicts the P-wave (v_p) and S-wave (v_s) velocities in the upper mantle.

PP-P and SS-S can be influenced by deep mantle structure, especially when P and S waves propagate through D’’. It is therefore important to find independent constraints on v_p and v_s in the upper mantle. We explore P-wave reverberations in the mantle contained in the six-hour long segment of a vertical-component seismogram following the major-arc Rayleigh wave. We detect top-side reflections (i.e., p410s and p410p) off the 410-km discontinuity in stacks of cross-correlated “late coda” from 83 large earthquakes recorded by the Southern California Seismic Network. The lag time between p410s and p410p is about 39 s for an inter-station distance of 100 km. Together with PP-P and SS-S, we estimate that v_s and v_p in the uppermost 200 km of the mantle are 1% and 5% lower than in PREM, respectively. This is an important constraint for tomographic studies of Earth’s upper mantle, especially studies focused on horizontal v_p variations and radial anisotropy.

Additional Authors

Session: Testing, Testing 1 2 3: Appropriate Evaluation of New Seismic Hazard and Risk Models [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/15/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: JyunYan Huang

Description

Models for the probability of surface rupture are used in probabilistic fault displacement hazard analyses (PFDHA). Most existing empirical modelsfor the probability of surface rupture (PSR) are based on a magnitude and style of faulting. An advantage of empirical models is that they need fewer assumptions about the depth distribution of the top of rupture of large-magnitude earthquakes, but a disadvantage is they do not account for the differences in the seismogenic thickness. Faultswith down-dip widths that are much thinner than average (e.g., 5 km), have a larger chance of surface rupture from small-magnitude earthquakes. Notable examples include the 2010 Mw 5.0 earthquake in Ecuador and the 2019 Mw 4.9 quake in France. Mammarella et al. (2024) proposed a PSR model utilizing a numerical approach that accounts for the seismogenic thickness and dip of the fault. This model has the advantage that it uses the fault-specific hypocenter depth distribution and seismogenic thickness, but it has the disadvantage that it assumes that the depth distribution of the top of rupture can be computed from the distribution of hypocenters from all magnitudes. In this study, we adopted an approach that uses the strengths of the empirical and numerical approaches. An empirical model is developed using the width-rupture ratio (WRR) in place of magnitude for the logistic regression. The WRR is the rupture width divided by the fault width. We compiled an empirical dataset of crustal earthquakes that occurred between 1970 and 2023 with a magnitude greater than 6. About half of the events have fault rupture information from finite-fault inversion. In cases where fault rupture widths were not available, we utilized the source scaling model from Huang et al. (2024) to estimate the rupture widths. By using the WRR, the proposed model accounts for the fault-specific seismogenic thickness and dip, but with empirical constraints on the PSR, avoiding the key assumptions in the numerical approach. This model successfully accommodates both large-and small-magnitude cases, facilitating fault-specific evaluations of the PSR.

Additional Authors

Session: Earthquakes, Lithospheric Structure, and Dynamics in Stable Continental Region [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Chris Justiniano

Description

Southwestern Puerto Rico (SWPR) received significant attention after the 2020-ongoing earthquake sequence, which included events located near offshore Guayanilla Canyon. Previous research suggested that this seismicity migrated onshore along left-lateral strike slip and transtensional faults, including the Punta Montalva and South Lajas faults, respectively. Most of previous geophysical work in this region was conducted through offshore seismic reflection surveys, leaving significant gaps for understanding the subsurface character of onshore faults. Strikes, dips, and depths of potentially active fault planes are critical factors to comprehensively characterize faults in tectonically active regions and for updating future seismic hazard assessments. Seismic reflection profiles were collected in the Spring of 2023 and 2024 across Lajas Valley, SWPR, to produce a new onshore fault characterization of the South Lajas Fault and other associated faults. Two phases of data acquisition were completed using a 216-channel seismic system with a weight-drop source. Profiles were collected perpendicular to the strike of the valley, and spanned the entire valley. Seismic data processing and data analyses were carried out using the Seismic Unix processing system. Ground Penetrating Radar (GPR) surveys were completed in Fall 2024 along the same profile location to characterize structures and potential faulting in the upper 20 m of the valley’s subsurface using the open-source R language processing software RGPR. Preliminary seismic results showed excellent imaging of southward-thickening stratigraphy that record faulting associated with the South Lajas Fault up to 250 m depth along with interpretation of depositional events before, during, and after tectonic activity in Lajas Valley. GPR data displayed near-surface layers and two discontinuous features possibly related to faulting to at least 10 m depth. This information will help illuminate the structure and tectonic evolution of Lajas Valley along the SWPR boundary.

Additional Authors

Session: Accuracy and Variability of Physics-based Ground Motion Modeling [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/15/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Mehran Davatgari-Tafreshi

Description

This study aims to derive adjustment factors for the median of the NGA-East ground-motion models (GMMs) targeting the Gulf Coast and the Coastal Plain of the United States. Within the region collectively referred to as the Coastal Plain, we formulate a set of adjustment factors based on sediment thickness and rupture distance. This study uses sediment thickness contour maps from Boyd et al. (2024) and a comprehensive dataset to compute residuals by subtracting the natural logarithms of observed data from those predicted by the median of the NGA-East GMMs. This dataset merges the NGA-East dataset (Goulet et al., 2014), the Chapman and Guo (2021) dataset, and the newly compiled and verified USGS dataset (Thompson et al., 2023). A mixed-effects regression separates total residuals into between-event and within-event components. Adjustment factors for stations within the Coastal Plain are derived by regressing within-event residuals using a functional form incorporating sediment thickness and rupture distance. The results indicate that the proposed adjustment factors effectively mitigate residual trends related to site and path terms across most periods for stations within the Coastal Plain. These findings apply to seismic hazard and risk assessments for sites within the Coastal Plain.

Additional Authors

Session: Macroseismic Intensity: Past, Present and Future - I

Type: Oral

Room: Key Ballroom 12

Date: 4/17/2025

Presentation Time: 03:00 PM Pacific

Presenting Author: Ava Fathian Sabet

Description

A new data set for macro intensity from crustal earthquakes in the Western US was developed combining the Did you feel it (DYFI) data set from the USGS with the historical intensity data set. Rupture distances were estimated for all events with magnitudes greater than 6. The site conditions are parameterized by the proxy-based VS30 and by the surface geology. The full data set includes 696,000 intensity observations from 5561 earthquakes. In addition to the intensity values, the data set includes the peak acceleration and peak velocity from the NGA-W3 data set for intensity observations near seismic stations. The data set is dominated by the DYFI observations with 96% of the intensity values and 98% of the earthquakes. For magnitudes of 6 or larger at distances less than 30 km, the historical intensity data accounts for 20% of the intensity values. To avoid the impact of censoring on lower MMI values, we establish (1) an Rmax distance based on the distance at which the minus two standard deviation level below the median from a non-parametric fit, and (2) all intensity values less than 2.1 were excluded. This reduced the data set used to develop the intensity prediction equation (IPE) to 358,000 recordings from 3642 earthquakes. The IPE is developed using model forms and regression methods commonly used in modern ground-motion models. In particular, we use a mixed-effects regression and include a magnitude-dependent finite-fault term and a site term, which are not included in the IPEs currently used for the Western US. The new IPE has much stronger saturation in magnitude scaling above M7 at shorter distances and a weaker magnitude dependence M5 at longer distances when compared to currently used IPEs for the western US (e.g., Atkinson et al. 2014). Although the intensity data is correlated to the peak velocity, the site term in the IPE is weak with a ln(VS30) coefficient of only -0.1 compared to -0.67 for peak velocity.

Additional Authors

Session: Exploring Planetary Interiors and Seismology: Observations, Models, Experiments and Future Missions [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Doyeon Kim

Description

Analysis of conversions between compressional and shear waves is a workhorse method for constraining crustal and lithospheric structure on Earth; yet, such converted waves have not been unequivocally identified in seismic data from the largest events on the Moon, due to the highly scattered waveforms of shallow seismic events. We reanalyze the polarization attributes of waveforms recorded by the Apollo seismic network to identify signals with rectilinear particle motion below 1 Hz, arising from conversions across the crust-mantle boundary. Delay times of these converted waves are then inverted to estimate crustal thickness and wave speeds beneath the seismometers. Combined with gravimetric modeling, these new crustal thickness tie-points yield an updated lunar crustal model with an average thickness of 29–47 km. The model provides critical context for future lunar exploration and geophysical studies, predicting a 15-36 km thick crust at Schrödinger basin and 29–52 km at potential Artemis III landing locations.

Additional Authors

Session: Advancements in Forensic Seismology and Explosion Monitoring - I

Type: Oral

Room: Holiday Ballroom 4-6+Corridor

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Arthur J. Rodgers

Description

Advances in the workflow for full waveform inversion (FWI) tomography, improved open broadband waveform data sets and more powerful high-performance computing (including GPU accelerated platforms) now make it possible to easily develop three-dimensional (3D) seismic Earth models for improved simulations of complete waveforms. Regional- and continental-scale FWI models enable physics-based simulations of observed complete waveforms including path-specific propagation effects of wavespeed heterogeneity. Forward waveform simulations in 3D models can provide Greens functions (GFs) for source inversions to estimate the moment tensor, seismic moment (and magnitude) as well as depth and location. This presentation will demonstrate how FWI tomography models can be obtained with modern workflow tools and openly available broadband waveform data. The resulting models improve waveform simulations for earthquakes and nuclear explosions. We show how reciprocity can be used to precompute databases of Strain Greens Tensors for use as receiver-side GFs. Precomputation of databases liberates operational pipelines from dependence on “on demand” HPC for event-specific analysis. Such databases can provide GFs via HDF5 file or webservices queries (e.g. InstaSeis). Rapid access of GFs enables stochastic inversions (e.g. Random Walk Hastings-Metropolis, Hamiltonian Monte Carlo) for source parameters with their improved representation of posterior probabilities (parameter variance and covariance). We propose that FWI models can be developed for nuclear test sites and broad areas enabling moment tensor inversion with improved accuracy provided by better representation of path-specific propagation effects. Stochastic inversions of source parameters with improved parameter uncertainties and covariances support quantitative source-type characterization that compliments traditional identification and magnitude/yield estimation.

Additional Authors

Session: Unusual Earthquakes and Their Implications [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/16/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Zhe Su

Description

The Sichuan Basin, adjacent to the eastern margin of the Tibetan Plateau, has long intrigued geologists and seismologists due to its complex tectonic and seismic activity. While existing models, such as upper crustal thickening and low-to-middle crust channel flow, explain aspects of plateau growth and tectonic extrusion, they fail to fully account for the internal seismicity of the basin and the weak tectonic activity preceding major earthquakes like the 2008 Mw 7.9 Wenchuan event. To address these gaps, we present a novel framework for understanding the tectonic and seismic dynamics of the Sichuan Basin.

Integrating geological evidence, focal mechanism solutions, GPS measurements, and geomorphic analyses, we identify a counterclockwise rotation of the basin’s crystalline basement—initiated ~12 million years ago—driven by the lateral extrusion of the Chuan-Dian fragment at the southeastern margin of the Tibetan Plateau. This rotation is facilitated by interactions between two dominant fault systems: the left-lateral Xianshuihe fault, which accommodates southeastward motion, and the Longmenshan fault belt, responsible for oblique compression and thrusting. These processes lead to strain partitioning, reactivation of NE-SW transpressional faults, and decoupling of the sedimentary cover from the crystalline basement.

Earthquakes predominantly occur at depths of 8–25 km within the rigid basement, as demonstrated by events like the 2008 Wenchuan earthquake. This highlights the role of fault discontinuities and rotational dynamics in accommodating strain. Our findings enhance the understanding of the occurrence of right-lateral shearing during the Wenchuan earthquake and the high frequency of seismic events within the rigid Sichuan Basin block, thereby contributing to the development of improved seismic risk models and early warning systems for tectonically active regions worldwide.

Additional Authors

Session: Earthquake Shaking and the Geologic Record: Triggered Phenomena and Preserved Fragile Geologic Features [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Ann E. Morey

Description

Paleoseismic evidence resulting from past earthquakes can be used to infer shaking parameters using the relationships between PGA and the maximum distance of earthquake-triggered liquefaction and landslide evidence. Although this is useful information, probabilistic seismic hazard models need estimates of local shaking parameters where the potential for damage is the greatest. The most damaging earthquake ground motions are S-waves, and a new type of paleoseismic evidence has the potential to provide information about their frequency and duration. Preliminary results from a deposit sequence representing the 1873 CE Brookings earthquake sequence from lower Acorn Woman Lake suggests that, whereas the lower deposit is the result of a landslide likely formed in response to a crustal earthquake, the upper deposit was formed in response to shaking that lasted longer, supporting a published inference that the upper deposit was formed in response to a southern Cascadia subduction earthquake.

This new paleoseismic evidence is the presence of a long organic-rich deposit tail. Evidence suggests that organic tail deposits may form if liquefaction releases ground water and fine-grained sediment into the lake, which, during sustained shaking, results in flocculation and the rapid settling. Previous laboratory studies have suggested that flocculation can occur when stirring occurs at a frequency of ~4 Hz or less, for a long duration (a minute or longer). If shaking is shorter, flocs may not form, and if shaking is too strong, flocs may break apart. Because the upper deposit of the 1873 CE Brookings earthquake sequence has a tail, it supports the interpretation that this deposit was the result of a southern Cascadia subduction earthquake. These results are preliminary, and experiments are planned to quantify the range of S-wave frequencies and duration required to produce this type of paleoseismic evidence. This type of evidence could be extremely useful in regions where both crustal and subduction earthquakes occur, such as in Cascadia, where little information is known about the amount of shaking inland where most people live.

Additional Authors

Session: Seismology for the Energy Transition [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/16/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Yuan Tian

Description

Large-scale geologic carbon storage (GCS) requires affordable geophysical tools for imaging and monitoring of subsurface CO2 plumes. Compared with active seismic methods, passive seismic surveys are low-cost but need enhanced resolution and event location precision. An improved passive seismic method and a high-resolution basin-scale earth model are crucial for cost-effective monitoring carbon storage. We will explore and test a novel passive seismic approach for basin scale GCS monitoring. This approach will also benefit source characterization by providing improved basin-scale earth models. Our technique uses adjoint waveform tomography, leveraging the long-lasting amplified seismic waves in the basin to passively image carbon plumes. The longer time windows selected for the adjoint waveform inversion contain information about the reverberating waves passing through the carbon plume, enabling high-resolution imaging of the carbon plume using only a few natural earthquakes. Our preliminary adjoint waveform tomography in Alaska's Nenana Basin, utilizing one seismic event and a single iteration, shows that shear wave sensitivity is significantly enhanced in the basin. This approach is expected to fill a critical technology gap in GCS monitoring, utilizing open source automated seismic adjoint tomography tool and data from existing CO2 storage sites.

Additional Authors

Session: Data-driven and Computational Characterization of Non-earthquake Seismoacoustic Sources - I

Type: Oral

Room: Holiday Ballroom 1

Date: 4/16/2025

Presentation Time: 02:30 PM Pacific

Presenting Author: Denizhan Kara

Description

A moving seismic source generates signatures that change both temporally and spectrally. When combined with the seismic propagation due to the laterally varying structure of the shallow crust, we can expect significant variations in the received seismic signal. We present a contrastive learning strategy that uses temporal similarity, derived from time-frequency coherence, to estimate and aggregate consistent seismic signal patterns for complex anthropogenic sources. By leveraging physics-guided frequency-based temporal analysis, our approach captures the non-stationary characteristics of seismic data more effectively than standard methods, and under multiple propagation conditions.
To evaluate our approach under varying noise levels and signal complexities, we use three seismic and acoustic datasets for a variety of ground vehicles: (1) MOD, a self-collected set of seven vehicles plus a dismounted human; (2) ACIDS, containing nine ground vehicles traveling at speeds from 5-40 km/h; and (3) NSIN, featuring eleven commercially available vehicles in three different environments under varying speeds, distances, and directions. Using our physics-augmented embeddings in a lightweight downstream classification model (using just a few neural network layers), we create classifiers that are robust to the inherent non-stationarity in seismic signals. Experimental results show that time-frequency coherence yields more accurate inter-sample similarity assignments and improves downstream tasks with up to 7% higher target classification accuracy while requiring fewer labeled samples than conventional contrastive learning frameworks. These findings demonstrate the strong potential of the Foundation Models for advanced seismic signal understanding under varying noise and propagation conditions.

Additional Authors

Session: Macroseismic Intensity: Past, Present and Future [Poster]

Type: Poster

Room: Exhibit Hall

Date: 4/17/2025

Presentation Time: 08:00 AM Pacific

Presenting Author: Tiegan E. Hobbs

Description

The International Macroseismic Scale (IMS) seeks to improve upon the collection of macroseismic data by detailing and standardizing the process used to assign earthquake intensity values globally. While IMS is intended to be representative of building classes around the world, countries may create their own Annex document. National Annexes are meant to account for the nuances of building types not otherwise captured in the central document, differing vulnerability classes and damage grades of those buildings, and national or regional post-earthquake building code and damage assessment practices that are likely to be mined for macroseismic data.

Canada intends to create a National Annex to the IMS guidelines, leveraging the existing “Did You Feel It?” system from Natural Resources Canada and the related macroseismic evolution ongoing in the United States and New Zealand. This synergy is possible due to the many shared building characteristics and post-event reporting strategies between all three countries. For example, the draft United States Annex (Hortacsu et al., 2024) recommends using a building taxonomy and corresponding vulnerability classes from the Federal Emergency Management Agency's FEMA P-154, combined with damage classifications from the Applied Technology Center's ATC-20. Canadian equivalents from the first-generation Canadian Seismic Risk Model and British Columbia’s Post Disaster Building Assessment program will be evaluated using criteria similar to those applied in the United States and abroad to arrive at a concept of implementation. We will present our preliminary Canadian IMS Annex to facilitate input from the Canadian community of earthquake engineers and seismologists.

Additional Authors

Session: Earthquakes, Lithospheric Structure, and Dynamics in Stable Continental Region - I

Type: Oral

Room: Key Ballroom 10

Date: 4/17/2025

Presentation Time: 08:15 AM Pacific

Presenting Author: Roger Bilham

Description

Railroads that cross active faults can provide a striking visual record of surface rupture in large earthquakes, similar to the offset fences that led Reid (1908) to develop elastic rebound theory. Railroads, once offset, may provide a precise measure coseismic offset long after they are repaired, since repairs are typically confined to the region of damage near the fault. We provide examples from a subset of railroads damaged by historical and recent earthquakes: the M > 7 1886 Charleston earthquake (doi.org/10.1785/0320230022), the ;M 6.8 1892 Chaman and M 7.7 1935 Quetta earthquakes in Pakistan (doi.org/10.1785/0220190148), the M 7.4 1944 Bolu, and the M 7.6 1999 Izmit and M 7.8 2023 Kharamanmaras earthquakes in Turkey. Railroads can, however, provide more than just a measure of coseismic offset. They act as distributed strainmeters in the meizoseismal region of an earthquake (doi.org/10.1785/0120240025). As an astute contemporary geologist deduced following the 1886 Charleston earthquake, railroad tracks respond to the transient strain associated with the outward propagation of surface waves. Prior to the implementation of continuously welded rail, early railroads incorporated expansion gaps sufficient to accommodate ~10^-4 of thermoelastic strain. Transient dynamic strain contraction exceeding this amount following an earthquake results in track buckling, and transient extension results in track separation. Once formed, buckled track cannot be straightened by the tensile phase of transient strain, since the forces required are typically far higher than those needed to create the buckle. Damaged railroad tracks can thus provide a max/min measure of dynamic strain in an earthquake. Because peak dynamic strain is proportional to peak ground velocity and conventional macroseismic shaking intensity at strong shaking levels is proportional to the logarithm of peak velocity, it follows that macroseismic intensity can be estimated from dynamic strain: 2x10^-4 strain is approximately equivalent to MMI 7.5, and 10^-3 strain is equivalent to MMI 9.

Additional Authors