Session: Earthquake Source Parameters: Theory, Observations and Interpretations

Type: Oral

Room: 202A/B

Date: 4/18/2023

Presentation Time: 04:45 PM Pacific

Presenting Author: Margaret Hellweg

Description

North of 40N and west of 122W lies the Mendocino Triple Junction (MTJ), one of California’s tectonically most active and complex regions. Damaging earthquakes can occur on the San Andreas and Mendocino faults, within the Gorda and North American plates, and on the Cascadia subduction interface. Recent earthquake sequences in the MTJ region, starting on 2021-12-20 (2021) and 2022-12-20 (2022), highlight the complex interactions of regional faulting. In the catalog, the 2021 foreshock/mainshock (FS/MS) pair, separated by ~10 s in time and 30 km in distance, have catalog magnitudes of 5.7 and 6.2. Detailed analysis indicates they are more likely to be a doublet, both with M~6.1, with the onshore MS triggered by the S-waves of the FS which occurred to its W on the Mendocino Fault. Finite fault analysis shows the FS ruptured to the W, and the MS to the NW, away from the population centers. The 2022 MS (M6.4) occurred just offshore, to the NW of the 2021 MS and produced aftershocks within the Gorda Plate and beneath the False Cape Shear Zone. Its largest aftershock (AS, M5.4) took place ~40 km to the ESE of the main fault rupture. We present finite fault analysis of the 2022 MS which indicates that it ruptured NE, toward the towns in the area and causing damage, which was exacerbated by the 2022 AS. Over the past few years, the instrumentation in the region has been upgraded as part of the implementation of the earthquake early warning system. We explore the seismicity associated with these earthquakes, looking to better define the faults in the MTJ region, as well as their modes of rupture.

Additional Authors

Session: USGS National Seismic Hazard Models: 2023 and Beyond

Type: Oral

Room: 208A

Date: 4/18/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Mark D. Petersen

Description

We are developing the 2023 U.S. National Seismic Hazard Model for all 50 states by applying new seismicity, fault rupture models, and ground motion assessments. We consider two types of models: (1) a public model that guides hazard assessments intended for building code provisions and policy applications and (2) a research model geared towards less formal applications that may be included in future public models. New data and methods will be introduced in the 2023 earthquake source components: earthquake catalogs - excluding induced earthquakes, alternative declustering methods, spatially smoothed seismicity, new geodetic- and geologic-based fault and deformation models, and earthquake rupture forecast models accounting for a more complete representation of potential earthquakes in Alaska, Hawaii, and the conterminous U.S. Improved ground motion models (GMMs) consider new NGA-Subduction, modified NGA-East and NGA-West2. Some GMMs and earthquake source models contain changes in median and aleatory variability that need to be evaluated and weighted. Amplification models will also consider available 3D simulations to supplement the empirical GMMs, and basin specific data in California Central Valley, Seattle Washington, Los Angeles California, and near Portland Oregon. New stress drop, tomographic, and other geophysical models help refine the complex boundary between the tectonically stable and active regions used in assigning GMMs. The research model may incorporate more geological information on timing of past earthquakes for time-dependent assessments and on regional variations in earthquake shaking - accounted for by a partially non-ergodic ground motion representation. Scenarios and population exposure assessments of these models will help users assess impacts of the earthquakes and facilitate risk mitigation measures.

Additional Authors

Session: Active Faults in the Caribbean and Central America

Type: Oral

Room: 208C

Date: 4/20/2023

Presentation Time: 02:15 PM Pacific

Presenting Author: Belle Philibosian

Description

The earthquake and tsunami potential of the Lesser Antilles megathrust is poorly known, despite the hazard it poses to numerous island populations and its proximity to the Americas. While the fault has not produced large earthquakes in the instrumental era, historical records of great earthquakes in the 19th century and earlier, which were likely megathrust ruptures, suggest that the subduction is not entirely aseismic. We used coral microatoll paleogeodesy to study century-scale vertical deformation in the northern Lesser Antilles. Only one microatoll has been found that is old enough to potentially record 19th-century earthquakes, and no “fossil” microatolls have been found that might record earlier events. However, we found many sites with microatolls recording 20th-century vertical deformation, which demonstrate that the eastern coasts of the forearc islands have been subsiding by up to ∼8 mm/y relative to sites closer to the arc. Modeling this deformation as the result of underlying strain accumulation on the megathrust suggests that a portion of the megathrust interface just east of the forearc islands has been locked during the 20th century. If this model is correct, the accumulated strain detected by microatoll studies will likely be released in future megathrust earthquakes, uplifting previously subsiding areas and potentially causing widespread damage from strong ground motion and tsunami waves.

Reconciling our results with decadal-scale instrumental geodesy has proven challenging. Models of instrumental horizontal deformation suggest little or no strain accumulation anywhere along the Lesser Antilles megathrust, and instrumental vertical deformation shows subsidence averaging 1–2 mm/y across the Lesser Antilles, variable between neighboring sites and with varying degrees of agreement with our data. The discrepancy with our results can potentially be explained by the different time scales of measurement, as recent studies elsewhere have indicated that interseismic coupling patterns may vary on decadal time scales and that century-scale or longer records are required to fully assess seismic potential.

Additional Authors

Session: Seismology's Role in Assessing Volcanic Hazard at Multiple Time Scales [Poster]

Type: Poster

Room: Ballroom

Date: 4/18/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Wendy McCausland

Description

After 110 years of repose, Momotombo volcano began erupting on 1 Dec. 2015. Momotombo has very high temperature fumaroles (~800C) since at least 1978 and magma close to the surface. Since regional seismic monitoring began (1975), seismicity has been dominated by distal volcano tectonic earthquake (dVT) swarms. Precursory seismicity follows the Four Stage Conceptual Model of White and McCausland (2019), with Stage 1 (deep seismicity) and Stage 2 (dVTs) bot progressing since 1975, and Stage 3 (vent clearing seismicity) first prevalent in 2000. At least 56 dVT swarms occurred since 1975, indicating persistent small intrusions into the magma reservoir, helping to maintain the very high temperature summit fumaroles for decades. The recurrence rates of dVT swarms decreased from 261 days between 1975-83 and 1993-99, to 135 days between 2000-08, to 36 days in 2013, and to 23 days in the 17 months prior to the eruption. The latter decrease occurred immediately after nearby M6.1 and M5.3 tectonic events whose source models indicate both dextral and normal motion, with a combined net extensional change in the stress field at Momotombo (Higgins, 2021). We hypothesize that the resulting change facilitated increased magma supply. Seismicity reached Stage 3 (tremor and tornillos) twice prior to the eruption: in mid-2000 and in June 2006 when a phreatic emission occurred. After June 2006, seismicity was again dominated by Stages 1 and 2 with occasional Stage 3 tornillos. About 6 months before the eruption onset, two production wells at the geothermal plant (3 km south of the summit) recorded coincident, small, persistent, and unusual increases in water pressure. Four months prior to the eruption, unusually large numbers and magnitudes of Stage 1 deep earthquakes began to occur, reaching M3.8. No short-term forecast was provided nor was a final transition to Stages 3 and 4 observed, as the closest seismic station was inoperable. However, the Instituto Nicaragüense de Estudios Territoriales issued intermediate-term warnings based on the dramatic increases in Stage 1 and 2 seismicity during the 4 months prior to the eruption.

Additional Authors

Session: Multi-scale Models for Seismic Hazard Analysis

Type: Oral

Room: 209B

Date: 4/20/2023

Presentation Time: 02:45 PM Pacific

Presenting Author: Nico Schliwa

Description

We present large-scale modeling of near-field broadband (>5 Hz) ground motion parameters and deformation in dynamic rupture simulations of the 2019 Mw 7.1 Ridgecrest earthquake. We include an observationally constrained shallow low-velocity fault zone and band-limited self-similar fractal fault roughness. Fault roughness induces geometric and small-scale stress perturbations, which are effective in generating deterministic high-frequency radiation. The model also incorporates a strong velocity weakening rate-and-state friction law, complex non-vertical fault geometry, 3D velocity structure, high-resolution topography, 3D initial stress distribution, off-fault plasticity, and viscoelastic attenuation (Taufiqurrahman et al., ResearchSquare, 2022).

We examine different ground motion parameters (PSA, PGA, PGV, PGD) and compare them with observed seismograms and GMPEs. We are especially interested in the near-field, closer than 10 km to the rupture, where real observations of strong ground motions are sparse and GMPEs are poorly constrained. The modeled ground motion parameters exceed the empirical predictions close to the rupture zone (< 2 km), which is only partly explained by effects of the low-velocity fault zone. We discuss which (if any) surface observations can constrain the dynamics of earthquakes, in particular for rapidly decaying near-field source effects. For each examined parameter, we show how the values inferred from the surface compare to their equivalents inferred within the volume. We also analyze the distributions of plastic vs. elastic off-fault deformation and their partitioning into volumetric and shear components.

Additional Authors

Session: Crustal Imaging of High Seismic Hazard Regions

Type: Oral

Room: 209B

Date: 4/20/2023

Presentation Time: 10:15 AM Pacific

Presenting Author: Simona Gabrielli

Description

Between August and October 2016, the Central Apennines (Italy) was struck by a cascading seismic sequence known as the Amatrice (Mw 6.0) - Visso (Mw 5.9) - Norcia (Mw 6.5), which has been associated with fluid circulation in the normal fault network. Seismic attenuation parameters (e.g., scattering and absorption) are powerful tools for detecting fluid presence and fracturing. The aim of this work is to investigate with 3D images, in space and time, the scattering contribution to the total attenuation of two datasets, one before the mainshock of Amatrice (March 2013-August 2016) and a second during the sequence (August 2016-January 2017). To measure scattering, we used peak delay mapping.
A comparison between the 3D mapping of the pre-sequence and sequence shows a dissimilarity in scattering, with a general increase of it over time after the Amatrice main event. Main structural elements (e.g., Monti Sibillini and Acquasanta thrusts) and lithologies are strongly controlling the scattering losses, with changes from low to high scattering around 4 and 6 km depth, west of the Norcia basin and of the Monti Sibillini thrust, suggesting an increment of fracturing due to the intense seismic activity. Monti Sibillini and Acquasanta thrusts act also as rheological barriers, creating contact between different geological domains. This causes a scattering contrast due to different responses in fracturing and compacting before and during the seismic sequence. Low scattering anomalies are located at the footwall of the thrusts during a pre-sequence phase, replaced by high scattering after the mainshocks. Low scattering can be associated with increased pore pressure due to fluid circulation in Triassic deposit layers; the movement and release of the pressure starting from this formation possibly provoked the movement of the thrusts, triggering the main events of the 2016 sequence.

Additional Authors

Session: February 2023 Mw 7.8 Earthquake Sequence in Turkey

Type: Oral

Room: 201 A/B

Date: 4/19/2023

Presentation Time: 11:00 AM Pacific

Presenting Author: Chris W. D. Milliner

Description

The variation of stress on faults is important for our understanding of fault friction and the dynamics of earthquake ruptures. However, we still have little observational constraints on absolute stress magnitudes, or their variations in space and in time over the seismic cycle. Here, we use 3D surface deformation measurements in the near-field of the 2023 Mw 7.8 and Mw 7.6 Kahramanmaras earthquakes to estimate the distribution of 3D slip vectors along both ruptures and invert them for the stress state and frictional properties, using the approach of Milliner et al. (2022). To estimate the 3D coseismic surface deformation we invert azimuthal and range pixel offsets estimated from ascending and descending Sentinel-1 and descending ALOS-2 radar data. Radar were processed using ISCE, and pixel offsets from Sentinel-2 optical imagery estimated using the newly developed COSI-Corr+ software. The coseismic slip magnitude shows a marked decrease of ~3 m along a restraining bend of the Mw 7.8 mainshock rupture. We assume this reflects the quasi-static effect of a decrease in the initial shear stress due to the change of the fault geometry with respect to the ambient stress field. We use this to invert for the static and dynamic friction of the ruptured faults and the absolute stress magnitude and its orientation.

Additional Authors

Session: Crustal Imaging of High Seismic Hazard Regions

Type: Oral

Room: 209B

Date: 4/20/2023

Presentation Time: 10:45 AM Pacific

Presenting Author: Chiara Nardoni

Description

In high seismic hazard regions, modelling seismic attenuation and combining it with velocity models is a powerful tool for imaging crustal features, such as complex fault systems, sedimentary layers and fluids. A frequency-dependent attenuation model is also critical for improving wavefield predictions in ground motion studies.

In the northern Los Angeles area, sedimentary basins may act as a waveguide producing large amplifications from an earthquake rupture on the southern San Andreas Fault. While the Southern California Earthquake Center community models provide P and S wave velocities for this region, a 3D attenuation model has to be incorporated into Earth models for more accurate wavefield simulations. This work aims to develop the first high-resolution, physics-based attenuation model across the Chino and San Bernardino basins using local earthquakes recorded by eight nodal arrays (410 stations), deployed during the Basin Amplification Seismic Investigation (BASIN) experiment. We perform high-frequency attenuation analysis using the peak-delay and coda attenuation methods to measure the scattering and absorption contributions, respectively. We present seismic scattering and intrinsic attenuation tomography at 12, 18, and 24 Hz (down to 10 km depth with a horizontal and depth resolution of 3 km and 1 km, respectively). The results show that the main structural discontinuities, such as the Fontana fault in the Chino basin, control the scattering anomalies allowing for the discrimination between relatively compact and fractured media. High absorption anomalies across the basins mark sedimentary layers and suggest the presence of fluids, i.e., groundwater, in agreement with recent ambient noise tomography. We thus present imaging results of both elastic and anelastic properties that aim at constraining the effects of sediments, fluids, and fractures on seismic wave propagation beneath the Los Angeles metropolitan area.

Additional Authors

Session: Multi-scale Models for Seismic Hazard Analysis

Type: Oral

Room: 209B

Date: 4/20/2023

Presentation Time: 03:00 PM Pacific

Presenting Author: Evan Hirakawa

Description

We simulate 3D seismic wave propagation from the 2022 Mw 6.4 Ferndale, California earthquake and compare synthetic ground motions (<2 Hz) with recordings. The simulations use a 3D seismic velocity structure obtained by embedding a model of the Eel River Basin into the USGS San Francisco Bay region 3D seismic velocity model – regional domain (SFCVM-R). The basin model consists of a 2D bedrock-depth surface which was originally interpolated by Graves (1994) from cross sections drawn by Ogle (1953). We insert this as a new surface into the regional 3D geologic model and fill the basin with properties appropriate for Quaternary sediments. We model three moderate earthquakes (Mw<4.5) with hypocenters close to the 2022 earthquake as point sources and compare synthetics with recordings in order to test the velocity model.

We model the 2022 Ferndale earthquake source as a left-lateral rupture propagating unilaterally to the east, with our preliminary models using an assumed randomized slip distribution generated using the Graves and Pitarka (2016) approach. Synthetic ground motions are significantly amplified in the low velocity Eel River Basin, where >1g acceleration was recorded and MMI VIII shaking was experienced. We interpret the amplification of ground motions to result from a combination of rupture directivity and basin response effects.

Additional Authors

Session: Geophysical Data Analysis in Cloud Computing Environments [Poster]

Type: Poster

Room: Ballroom

Date: 4/18/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Carl Tape

Description

Observational seismology favors distributed memory workflows and embarrassing parallelization, for which cloud computing has an optimal architecture. Cloud computing offers horizontal scalability for computing workflows suited for single-station analysis. The orchestration of software, data, and resources on the cloud requires some thinking about the design. We present the initial experimentation of the SCOPED project. The SCOPED software ecosystem uses a base container and software-specific containers built on top, forming a registry of containers deployable on cloud instances. The SCOPED data uses S3-like objects, using mSEED files for raw seismic data and tileDB (the new Earthscope Consortium Cloud store) for data products. We propose the first cloudstore for Distributed Acoustic Sensing data. We illustrate the workflows using canonical examples in observational seismology: ambient-noise seismology and machine-learning earthquake workflows.

Additional Authors

Session: Constraining Seismic Hazard in the Cascadia Subduction Zone

Type: Oral

Room: 208A

Date: 4/20/2023

Presentation Time: 02:00 PM Pacific

Presenting Author: Tina Dura

Description

At the Cascadia subduction zone, quantitative foraminiferal-based coseismic subsidence estimates are most often used to constrain earthquake and tsunami source models. However, ecological factors affecting foraminifera such as dissolution in low pH environments, mixing across subsidence contacts, and limited supratidal and subtidal range can limit their application and accuracy. Diatoms have the potential to produce more accurate quantitative estimates of coseismic subsidence because they are composed of dissolution-resistant siliceous valves, do not burrow through sediment, have high species turnover in the intertidal zone, and are found in freshwater environments. Here, we develop a diatom-based Bayesian transfer function (BTF), which uses the empirical relationship between modern diatom assemblages and elevation within the tidal frame to estimate relative sea level (i.e., land level) changes from fossil diatom assemblages in cores and outcrops. We use a new modern dataset of >150 diatom samples from tidal wetlands along the Oregon and northern California coasts. We apply our diatom-based BTF to the 1700 CE earthquake contact at 13 Oregon and California coastal sites that also have existing foraminiferal-based BTF coseismic subsidence estimates. We find that at most sites diatom subsidence estimates are at the upper range (e.g., Salmon River) or exceed (e.g., Nestucca) foraminifera-based estimates. At Netarts Bay, the new larger diatom-based subsidence estimates are more similar to subsidence estimates at adjacent sites and help resolve ambiguities in rupture models for the 1700 CE earthquake. Future work will focus on understanding why foraminifera might underestimate subsidence and how diatom ecological factors can improve subsidence estimates, as well as determining the sensitivity of earthquake and tsunami source models to decimeter-scale increases in coseismic subsidence.

Additional Authors

Session: It’s All About Relocation, Relocation, Relocation [Poster]

Type: Poster

Room: Ballroom

Date: 4/20/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Barrett N. Johnson

Description

High-precision earthquake locations can illuminate seismogenic structures at depth, elucidate the spatial and temporal history of those structures, and can aid in source determination. This is especially important near volcanic centers considering their geological complexity and the significant volcanic hazards. In this study, we compare three approaches to perform high-precision earthquake relocation using our enhanced Mount Hood earthquake catalog to analyze their efficacy in determining locations for low magnitude events and events with few observations, and discuss the generality and scalability of each method. The enhanced earthquake catalog includes the 166 earthquakes located by the Pacific Northwest Seismic Network to routine operations and 1572 detected earthquakes derived from waveform template matching. Since monitoring began in 1979, the June 2021 sequence is both the best recorded and the most seismically productive sequence at Mount Hood. We relocate events using three different methods: absolute (NonLinLoc SSST + Coherence), hierarchical clustering (GrowClust), and double-difference (HypoDD). Across the methods, it is clear there are at least two clusters of earthquakes, a large cluster 1-2 km from the summit to the southeast and a smaller cluster at the summit. Based on our preliminary analysis NonLinLoc SSST + Coherence locations produced precise earthquake locations on par with the relative relocation methods. The fact that this method preserves the absolute location of earthquakes in geographical space is an asset.

Additional Authors

Session: Advances in Probabilistic Seismic Hazard Analysis and Applications

Type: Oral

Room: 208A

Date: 4/19/2023

Presentation Time: 08:30 AM Pacific

Presenting Author: Meredith L. Kraner

Description

Earthquakes larger than M_W7.0 have impacted the country of Mexico over the past century approximately once every 5-6 years, with six events having occurred since 2017. The Cocos and Rivera plates subduct eastward beneath the North America and Caribbean plates, creating a concentration of high hazard along the west coast seen on all existing seismic hazard maps (SHM). Despite frequent significant earthquakes, few peer-reviewed fully probabilistic seismic hazard analyses (PSHA) for the entire country exist. Risk assessment companies (e.g., Verisk Analytics) attempt to develop the most up to date view of risk using the most scientifically advanced, comprehensive, and accurate PSHA. Here, we present a comprehensive seismicity rate model that includes all significant seismic sources in Mexico including subduction zones, mapped active crustal faults, deep intra-slab earthquakes, and background seismicity to account for unknown faults. Our model incorporates a complete historical earthquake catalog homogenized to moment magnitude, an active fault database compiled from all regional and global data sources and publications, coupling information, and strain- and moment-rates from kinematic models of GPS data. We evaluate various ground motion models, including those from the NGA-Sub project, and select a suite of models appropriate for the various types of earthquakes in Mexico. Lastly, we integrate soil effects including a site amplification model for Mexico City based on published work. We present hazard results in terms of peak and spectral acceleration values (PGA & SA), and discuss the similarities and differences between our results and both peer- and non-peer-reviewed studies (e.g., GEM, CENAPRED, and national SHMs).

Additional Authors

Session: Opportunities and Challenges for Machine Learning Applications in Seismology [Poster]

Type: Poster

Room: Ballroom

Date: 4/19/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Renate Hartog

Description

The Pacific Northwest (PNW) region of the United States is a dynamic tectonic plate boundary between the North American continental plate and the Juan de Fuca oceanic plate. The active margin between the two plates is a subduction zone that hosts a wide range of earthquake, near surface, and volcanic behaviors.

The curation of seismic data and metadata sets is the cornerstone of seismological research and the starting point of machine-learning applications in seismology. We present a 21-year-long diverse data set of seismic events, their metadata, and their waveforms, as curated by the Pacific Northwest Seismic Network (PNSN) and ourselves. We describe the earthquake catalog, the temporal evolution of the data attributes (e.g., event magnitude type, channel type, waveform polarity and signal-to-noise ratio, phase picks) as the network cataloging system evolved through time. We also pick additional waveforms for the recent 20 December 2022 Northern California earthquake sequence, which was the largest event recorded recently in proximity to the PNSN authoritative boundaries. We propose this AI-ready data set as a new open-source benchmark data set.

Additional Authors

Session: Opportunities and Challenges for Machine Learning Applications in Seismology [Poster]

Type: Poster

Room: Ballroom

Date: 4/19/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Albert L. Aguilar

Description

We present an extensive quality-controlled dataset of waveforms of earthquakes recorded at regional distances. These waveforms are 5 minutes long and contain arrivals for the P, Pg, Pn, S, Sn and Sg phases, as well as event and station metadata. Each one of the examples in the dataset is required to have at least one of {P, Pg, Pn} arrivals and at least one of {S, Sg, Sn} arrivals. Arrivals in the dataset are recorded at a source-receiver distance between 1 and 20 degrees in three component instruments. After initially collecting over 3 million waveforms, we quality controlled the data using an ensemble of Machine Learning Models. First, we trained a Recurrent Neural Network that distinguishes between earthquake signals and synthetic noise. This model allows us to flag examples in the dataset for which there are labeled arrivals, but the waveforms do not show any distinguishable earthquake signal. On the other hand, given that 5 minutes is a long window, and many earthquakes can be recorded in such time, we used a fine-tuned version of our RNN to flag those examples for which there are multiple earthquakes, because only one of them is labeled. We show preliminary ML models trained on the dataset for seismic phase picking.

Additional Authors

Session: New Methods and Models for More Informative Earthquake Forecasting

Type: Oral

Room: 203

Date: 4/19/2023

Presentation Time: 08:30 AM Pacific

Presenting Author: Jose A. Bayona

Description

Since 2007, the Collaboratory for the Study of Earthquake Predictability (CSEP) has been prospectively evaluating 24-hour seismicity forecasting models for California to address seismological questions with important implications for time-dependent earthquake hazards. Among others, the pool of 24 models includes various flavors of ETAS, STEP, non-parametric models, and ensemble models. Using statistical methods developed by CSEP, we assess the consistency of these models against observed M3.95+ earthquakes, and compare their long-term performance with that of an early standard ETAS model. Our prospective dataset contains nearly 600 target events, including the 2010 M7.2 el Mayor-Cucapah, 2014 M6.9 Cape Mendocino, and 2014 M6 South Napa earthquakes. Here, we present preliminary test results that may be helpful in improving our ability to forecast earthquake clustering, and advancing Operational Earthquake Forecasting in California.

Additional Authors

Session: The 2020-2021 Southwest Puerto Rico Seismic Sequence: Current State of Knowledge and Implications

Type: Oral

Room: 209A

Date: 4/18/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Clara E. Yoon

Description

The still-ongoing 2020-2021 southwestern Puerto Rico seismic sequence is remarkable for its multiple-fault rupture complexity and unusually high aftershock productivity. We applied an automatic workflow to continuous data from 42 seismic stations to build an enhanced earthquake catalog with ~140,000 events for the 2+ year sequence, from 2019-12-28 to 2022-01-01. This workflow included the EQTransformer deep learning model for event detection and phase picking, the EikoNet-HypoSVI probabilistic earthquake location approach with a neural network trained to solve the eikonal wave equation, and relocation with event-pair waveform cross-correlation. EQTransformer increased the number of catalog events in the sequence by ~7 times; EikoNet-HypoSVI event location improved depth estimates. The enhanced catalog revealed how the sequence evolved on a complex system of many small normal and strike-slip faults. This sequence started on 2019-12-28 with a M4.7 strike-slip earthquake, followed by shallow strike-slip M5+ foreshocks in a compact region. The oblique normal-fault M6.4 mainshock then happened on 2020-01-07. Early aftershocks in January 2020, including several M5+ earthquakes, quickly expanded into two intersecting wide fault zones with diffuse seismicity: one extending ~35-km on a northward-dipping normal fault, the other ~60-km-long and oriented WNW-ESE on strike-slip faults. Later aftershocks moved westward, deeper, and to outer reaches of the active fault zones; abrupt rapid seismicity migration followed larger M4.7+ aftershocks in May, July, and December 2020. The observed seismicity space-time evolution indicates cascading failure of stress transfer on multiple critically stressed faults. The high aftershock productivity results from the complex multiple-fault network hosting the sequence, which is characteristic of an immature fault system in the diffuse deformation zone around Puerto Rico at the North American-Caribbean plate boundary region.

Additional Authors

Session: USGS National Seismic Hazard Models: 2023 and Beyond

Type: Oral

Room: 208A

Date: 4/18/2023

Presentation Time: 08:45 AM Pacific

Presenting Author: Yuehua Zeng

Description

We apply a fault-based crustal deformation model with deep driven dislocation sources to estimate long-term on-fault geodetic slip-rates and off-fault moment rate distribution in the Western United States (WUS) for the 2023 update to the National Seismic Hazard Model (NSHM). This model uses the method of Zeng and Shen (2017) to invert for slip-rate and strain-rate parameters based on inputs from global positioning system (GPS) velocities and geologic slip-rate constraints. The model connects adjacent major fault segments in California and the Cascadia subduction zone to form blocks that extend to the boundaries of the study area. Faults within the blocks are obtained from the NSHM geologic fault model. The geodetic slip rates are determined using a least-squares inversion with a normalized chi-square of 6.6. We also apply a time-dependent correction called “ghost transient” effect to the data to account for the viscoelastic responses from large historic earthquakes along the San Andreas Fault and Cascadia subduction zone. Major discrepancies between geodetic slip rates and geologic slip rates along the San Andreas Fault, for example, from the Cholame to the Mojave and San Bernardino segments of the San Andreas, are well-resolved after we apply the ghost transient correction to GPS velocities. Off-fault moment rate distribution is consistent with regional tectonics and seismicity patterns with a total rate of 1.6x10¹⁹ N˖m/year for the WUS.

Additional Authors

Session: USGS National Seismic Hazard Models: 2023 and Beyond

Type: Oral

Room: 208A

Date: 4/18/2023

Presentation Time: 02:30 PM Pacific

Presenting Author: Balakumar Anbazhagan

Description

Probabilistic seismic hazard analysis (PSHA) is the basis for the development of the National Seismic Hazard Maps (NSHM) produced by the USGS. These maps are adopted by building codes for the design of structures. This research intends to provide a framework by which the NSHM can be postulated in terms of a new paradigm: partially non-ergodic PSHA. Within this approach, PSHA analyses are conducted using the so-called single-station sigma, which is the aleatory variability that results after accounting for repeatable site effects at every station. In recent years, seismic hazard analyses that have been conducted for critical facilities, such as Nuclear Power Plants, have been performed within the framework of partially non-ergodic PSHA; the proposed research aims to introduce these advancements into the NSHM and national building codes. The proposed framework for implementing non-ergodic PSHA in building codes builds on the current framework by introducing a correction factor called ‘ergodicity factor’ that accounts for the inherent ergodic assumptions in PSHA. These factors are postulated as a function of the level of knowledge on site properties reflecting the fact that uncertainty in site effects is considered as epistemic in partially non-ergodic PSHA. The ergodicity factors are computed for NEHRP site classes for varying levels of site characterization. It was observed that the factors increase with increase in uncertainty on the estimates of shear wave velocity at a site, indicating increase in design motions with increase in epistemic uncertainty. Given that lower uncertainty leads to lower design values, the proposed framework incentivizes practitioners for investing more in site characterization in expectation of lower design seismic loads. The resulting reduction in hazard, however, is relatively minor.

Additional Authors

Session: Numerical Modeling in Seismology: Developments and Applications

Type: Oral

Room: 209C

Date: 4/20/2023

Presentation Time: 02:15 PM Pacific

Presenting Author: Peter Moczo

Description

Capability of a numerical modeling of seismic wave propagation and earthquake ground motion to faithfully represent material interfaces in realistic models of the Earth’s interior is of key importance. A sharp sediment-bedrock interface can be a key factor causing an anomalous earthquake ground motion and strong site effects. A sharp interface, e.g., a seabed, is an important structural feature in seismic exploration.

We address fundamental implications of presence of the material interface and spatial discretization for the finite-difference modeling by analyzing the equations of motion and constitutive relations in the wavenumber domain.

We present heterogeneous formulations of the equations of motion and constitutive relations for several basic configurations of a wavefield in an elastic isotropic medium. We Fourier-transform the entire equations to the wavenumber domain. Subsequently, we apply the band-limited inverse Fourier transform back to the space domain.

The heterogeneity of the medium, spatial discretization and the Nyquist-wavenumber band limitation of the entire equations have important implications for a FD modeling. The grid representation of the heterogeneous medium must be limited by the Nyquist wavenumber. The wavenumber band limitation replaces spatial derivatives both in the homogeneous medium and across a material interface by continuous spatial convolutions. The latter means that the wavenumber band limitation removes discontinuities of the spatial derivatives of the particle velocity and stress at the material interface. This allows to apply proper FD operators across material interfaces. A wavenumber band-limited heterogeneous formulation of the equations of motion and constitutive relations is the general condition for a heterogeneous FD scheme.

Additional Authors

Session: Earth’s Structure From the Crust to the Core

Type: Oral

Room: 209B

Date: 4/18/2023

Presentation Time: 08:45 AM Pacific

Presenting Author: Debajeet Barman

Description

Understanding the Alleghenian orogeny and its influence on the tectonic evolution of Laurentia depends on understanding the structural relationship between the Suwannee and its adjacent terranes. Our study focuses on creating high-resolution phase velocity estimates of the entire southeastern US region with less non-uniqueness than that of current models and with more accurate locations and orientations of terranes to constrain passive margin formation.

Vertical component seismograms are used to model Empirical Green’s functions (EGF) to measure Rayleigh waves via cross-coherence computation for all possible station pairs from approximately 300 broadband seismic stations deployed during the EarthScope project. Because the noise source distribution is not uniform from all directions, the EGF is corrected for all noise sources along the Great Circle Path (GCP) between the virtual source and receiver pair. This has been done with the help of Double Optimized Multiple Signal Comparison (DOPMUSIC), where the significant number of noise sources for all possible frequency bands along the GCP is used to estimate the appropriate EGF.

DOPMUSIC produces Rayleigh waves with significantly higher SNR for all frequency bands than the classical Double Beamforming (DBF) approach. The estimated EGF is used to calculate phase velocity dispersion curves with the help of spatially-variable Hankel transform functions. Classical phase velocity estimation approaches lack uniqueness because of cycle-skipping while matching the phase. The classical Frequency-Bessel transform often contaminates the phase velocity estimates by spatial aliasing and bi-directional velocity scans. And in effect, several crossed dispersion curves were observed in the spectrogram. We use the Hankel function instead of the Bessel function on the causal and acausal part of the EGF separately, which addresses this problem. Phase velocity maps are then used to estimate 1-D shear wave velocity profiles for the entire region.

[PJ1]Need to insert a comment here about what is missing from, or wrong about, current models or ideas about the passive margin formation.

Additional Authors

Session: Seismology's Role in Assessing Volcanic Hazard at Multiple Time Scales

Type: Oral

Room: 208B

Date: 4/18/2023

Presentation Time: 10:00 AM Pacific

Presenting Author: Barrett N. Johnson

Description

The region right around Mount Hood volcano, located 80 km east from Portland, is the most seismically area in Oregon, producing ~3k earthquakes since 1979. Seismicity near and below volcanoes is typically generated by fluids or gas migration, magmatic recharge and intrusions, through changes in regional stress regimes, or a combination of each. Careful examination of high-precision earthquake catalogs can elucidate seismogenic structures and uncover spatial and temporal patterns and evolutions of those structures at various timescales. The goal of this study is to characterize the seismicity of Mount Hood, the variability in sequence/swarm duration and location, measure the orientations of the faulting structures at depth, and the effects of improving the station density through time. To accurately characterize the recorded seismicity at Mount Hood, we constructed a high-resolution earthquake catalog. This was done by mining all continuous seismic data for stations within 50 km of Mount Hood, hunting hidden earthquakes using waveform template matching. Our detection efforts yielded a catalog of 24k new earthquakes. Locations were performed using NonLinLoc SSST + Coherence and of the 27k earthquakes in the enhanced catalog, 7,667 earthquakes were located. Our preliminary results suggest a complex faulting network in space with earthquake clusters located at and just below the summit, and on the flanks to the west, southwest, and southeast. Additionally, our preliminary temporal analysis results show a range of durations for different clusters. Further analysis is needed to identify the sources of these short, intermediate and long duration seismicity patterns.

Additional Authors

Session: Coseismic Ground Failure: Advances in Modeling, Impacts and Communication [Poster]

Type: Poster

Room: Ballroom

Date: 4/20/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Adel Asadi

Description

A series of earthquakes, with 7.3 Mw highest intensity, hit Kumamoto, Japan, over a period of two days in April 2016. The earthquakes caused numerous landslides and surface ruptures in the steep volcanic geological environment. In this study, pre- and post-event sets of high-resolution aerial (Geospatial Information Authority of Japan) and satellite (DigitalGlobe) imagery, paired with the USGS preferred geospatial model for landslide probability and individual geospatial inputs including elevation, surficial geology, slope, precipitation and landcover layer provided by National Mapping Organization of Japan, were used to develop a machine learning (ML) approach for landslide mapping. The ML approach was trained and validated using an inventory of human-drawn landslide occurrence polygons of the area as ground-truth labels provided by Kyoto University’s Disaster Prevention Research Institute and the Japanese National Research Institute for Earth Science and Disaster Resilience. The goal of this work is to improve automated image-based landslide mapping by adding data of physical parameters as well as temporal difference indices calculated from pre- and post-event imagery using ML algorithms. The selection of geospatial effective parameters was done via the supervised Bhattacharyya feature ranking method. The pixel-based ensemble classification algorithm used in this study not only learns from the color channels of the imagery, but also analyzes additional RGB-derived parameters, data of selected geospatial variables, and change information (vegetation and grayscale difference) derived by comparing pre-event and post-event imagery. Different combinations of input parameters were tested, and the results showed that adding data of selected geospatial parameters plus change indices to the imagery lead to the highest classification accuracy. Landslides in these formations are common throughout Japan and have been triggered by heavy precipitation as well as seismic events. Therefore, such data-driven, efficient and fast mapping methods can be effective for landslide prevention, mitigation and reconstruction operations.

Additional Authors

Session: Structure and Properties of Subducting Slabs and Deep Earthquakes [Poster]

Type: Poster

Room: Ballroom

Date: 4/19/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Yanbin Wang

Description

In the transformational faulting hypothesis for deep-focus earthquakes, metastable olivine in the subducted lithosphere transforms to wadsleyite and/or ringwoodite below ~300 km depth, triggering mechanical instability. Detailed physical mechanisms of this hypothesis are still under debate. We conducted controlled deformation experiments with acoustic emission (AE) monitoring to investigate this process. We examined three olivines: Mg₂GeO₄, Mn₂GeO₄, and (Mg_1-xFe_x)₂SiO₄ (with x=0.25 and 0.5). Under our experimental conditions, Mg₂GeO₄ olivine transforms directly to spinel (ringwoodite) structure, Mn₂GeO₄ to wadsleyite, whereas (Mg_1-xFe_x)₂SiO₄ into a mixture of olivine and ringwoodite.

Mechanical behavior of all three metastable olivines can be divided into three regimes, depending primarily on temperature. At low temperatures, metastable olivines are strong but ductile, with no AEs produced up to ~30% strain. At high temperatures, they are weak and ductile, with no AEs observed. Only within a narrow range of intermediate temperatures do these olivines behave in a brittle manner, emitting numerous AEs. Recovered brittle samples contain macroscopic faults. High-pressure products are primarily in the form of sub-parallel, long and narrow bands (widths on the order of 100 nm) cutting through individual olivine grains, sometimes shearing the olivine grains into several parts. These nano-shear bands (NSBs) are oriented about 30 – 60 degrees from the maximum compressive principal stress. Detailed microstructure analyses show that at intermediate temperature, olivines deformed by kink band formation and NSBs are primarily located within kink bands boundaries (KBBs). Within KBBs, olivine crystal lattice is severely distorted due to high density of dislocations, thereby promoting nucleation of high-pressure phases. The self-organization of NSB network produces macroscopic faulting, as evidenced by subsets of NSBs, which are either in the fault zone or very close to the faults.

Additional Authors

Session: High-frequency Ground Motion Measurements, Assessments and Predictions

Type: Oral

Room: 208C

Date: 4/18/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Adrian Rodriguez-Marek

Description

The incorporation of site effects into seismic hazard assessments requires the characterization of stiffness and attenuation (i.e., damping). This study presents the derivation of a model for small-strain damping for the seismic hazard assessment of the Groningen gas field in the Netherlands. The model is constrained by ground motion measurements over a large network of 200 m deep vertical arrays. Measurements of the seismic quality factor Q at the vertical arrays were conducted by first applying seismic interferometry by deconvolution to estimate the local transfer function. Two approaches were then used: the first by measuring the amplitude decay of the retrieved upgoing wave, and the second by measuring the amplitude difference between the upgoing and downgoing waves at a single geophone. The second approach has the advantage that corrections for elastic propagation are not needed; consequently, the use of this approach resulted in more robust measurements of Q. Moreover, the second approach is less sensitive to small differences in effective response functions of geophones at different depths because it relies on measurements at a single geophone. The measurements of Q at the vertical arrays were then used to build a model for the entire study area, which covers a region of approximately 40 by 50 km. The model was based on scaling small-strain damping models from the geotechnical engineering literature, which are based on geotechnical index properties, to match the measured Q values at the downhole arrays. This was possible because of the existence of a geostatistical model (the GeoTOP model) that provides estimates of geotechnical index properties for the entire study region. The scaling factor was determined by equating estimates of site kappa obtained using: a) the measured Q values and b) damping profiles obtained from the geotechnical damping models. We observed that the scaling factor is strongly correlated to the average shear wave velocity over the upper 30 m (V_S30). Q values at depths lower than the vertical arrays were derived from the attenuation of the microseism measured over the study region.

Additional Authors

Session: Above the Seismogenic Zone: Fault Damage and Healing in the Shallow Crust [Poster]

Type: Poster

Room: Ballroom

Date: 4/19/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Natasha Toghramadjian

Description

The Newport-Inglewood fault (NIF) is an active, complex strike-slip system that cuts over 60 km through metropolitan Los Angeles and poses one of the greatest deterministic seismic hazards in the US. A portion of the NIF sourced the 1933 Long Beach (M6.4) rupture, southern California’s deadliest earthquake. The event is thought to have arrested at Signal Hill, a large restraining bend formed by a left step in the NIF in the Long Beach oil field. The NIF is considered capable of generating much larger (M≈7) and more destructive earthquakes—yet key questions persist about its geometry, segmentation, and slip that are critical in assessing these hazards.

Integrating a diverse range of data, including ~4900 horizon picks and fault penetrations from 350 oil wells, 2D seismic reflection surveys, industry field maps and USGS QFaults surface traces, we present a new 3D model of the NIF at Long Beach, including its connections to the adjacent Seal Beach and Dominguez fault segments.

Current 3D representations of the NIF in the SCEC Community Fault Model describe it as a set of simple, disconnected, vertically-dipping strike-slip faults. Our analysis shows that the fault system is complex, with a series of dip-slip reverse faults of ~60° dip that orthogonally intersect and connect the en echelon, through-going strike-slip fault segments. The strike-slip faults are non-vertical and non-planar, and shallow their dip and merge with one another at depth, extending down through the seismogenic crust. These diverse hard fault linkages present numerous rupture pathways and arrest points that NIF earthquakes may take as they propagate through Long Beach.

We apply map-based restoration to quantify how much total slip has passed through each of the fault segments at Long Beach. Ultimately, we aim to compare this pattern with dynamic rupture simulations on our fault model, to see if they illuminate preferred rupture pathways and recreate the slip partitioning observed in the geologic record, which represents hundreds of earthquake cycles.

Additional Authors

Session: Collective Impact in Earthquake Science [Poster]

Type: Poster

Room: Ballroom

Date: 4/18/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Laura A. Navas Sánchez

Description

To quantify the human and economic losses caused by these events, seismic risk studies must deal with different types of uncertainties whose assessment methods are still under development, such as those associated with selecting fragility curves (FCs). The appropriate use of FC allows a better approximation of the level of performance of a structural system in the face of seismic hazards. In this respect, an inadequate selection of the FC can mean a notably unreliable estimation of damages and losses. This research proposes a methodology approach for assessing and selecting FCs for seismic risk studies from a catalogue of existing proposals available in the literature. To this end, a bibliographic search of different vulnerability and seismic risk studies for the Central American region has been conducted. A database was constructed with the parameters of the capacity and FCs available for diverse construction typologies. A multidimensional index based on several variables divided into key dimensions is proposed to facilitate the creation of a rational ranking of fragility curves in terms of adequacy. Additionally, the specific criteria considered to build this index were validated from a survey of experts and the implementation of the fuzzy Analytic Hierarchy Process (AHP).

The FCs used in different vulnerability projects in Costa Rica are explored, applying the methodology developed in this case. As a result, it is concluded that in the diverse vulnerability projects, FC with very different probabilities of damage is being used for buildings with the same attributes. Moreover, the proposed methodology permits knowing the reliability level of the FC used depending on the class the curve was classified into based on their score. Therefore, it allows establishing the adequacy and uncertainty related to the selected FC, a relevant subject that the researchers should consider in seismic vulnerability and risk studies.

Additional Authors

Session: Opportunities and Challenges in Source Modeling for Seismic Hazard Analysis

Type: Oral

Room: 208A

Date: 4/20/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: James S. Neely

Description

Current earthquake recurrence models have two major limitations. First, they predict that the
probability of a large earthquake stays constant or even decreases after it is “overdue” (past the
expected average recurrence interval), so additional accumulated strain does not make an
earthquake more likely. Second, the models assume that large earthquakes release all
accumulated strain, despite evidence of partial strain release in earthquake histories showing
clusters and gaps. These limitations arise because the models are purely statistical, assuming that
future earthquakes will satisfy a probability distribution that describes the times between past
large earthquakes. These models describe average earthquake behavior, but not deviations from
it, because they do not incorporate fundamental aspects of the strain accumulation and release
processes that cause earthquakes.

Here we calculate earthquake probabilities using our Generalized Long-Term Fault Memory
(GLTFM) model, which better reflects the strain accumulation and release processes. Our model
assumes that earthquake probability always increases with time between earthquakes and allows
the possibility of partial strain release. We apply this model to different faults including the
southern San Andreas fault. By allowing partial strain release, GLTFM incorporates the specific
timing of past earthquakes, which commonly used probability models cannot do, so it better
forecasts the short inter-event time preceding the 1857 Ft. Tejon earthquake. Looking to the
future, current models estimate the earthquake probability on the southern SAF will be
essentially unchanged in the next 80 years, but GLTFM predicts that the probability will
continue to grow, resulting in a 30-year earthquake probability approximately 60% higher than
the other models. These forecast differences could have major implications for hazard
assessment.

Additional Authors

Session: Numerical Modeling in Seismology: Developments and Applications [Poster]

Type: Poster

Room: Ballroom

Date: 4/20/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Ruth Harris

Description

We introduce 'non-linear radiation damping', a new mechanism that dissipates energy in elastic dynamic rupture simulations and results in more realistic ground motions and slip velocities, while not requiring the large number of assumed variables in viscoplastic methods. Computational simulations of dynamic earthquake rupture help us learn about earthquake source behavior and the resulting ground motions. A challenge with these types of simulations is that they require many assumptions, including the fault geometry, the fault friction constitutive laws, the rock properties, and the initial stress conditions. Although we have some information about these parameters, much of the information is unknown. This requires modelers to make many assumptions, with one of the most common assumptions being that rocks respond elastically to sudden changes in stress and strain. An elastic rock response is helpful in that it allows for a simplification of the initial stress conditions. However, an elastic rock response can also lead to simulated on-fault slip rates and off-fault ground motions that are faster than those inferred from earthquake observations. Our work provides a solution to this problem. We propose a new method, which we call ‘non-linear radiation damping’. The implementation is simple, and just requires adding the non-linear radiation damping term to the friction formulation. The result is slower on-fault peak slip rates and ground motions, similar to what might be achieved by incorporating a complex viscoplastic framework, but without the need to assume values for a variety of unknown features such as the initial stress conditions and rock yielding parameters at all locations in an earthquake region.

Reference:
Barall, M., and R.A. Harris, Non-linear radiation damping: A new method for dissipating energy in dynamic earthquake rupture simulations, manuscript submitted to The Seismic Record, January 2023.

Additional Authors

Session: From Earthquakes to Plate Boundaries: Insights Into Fault Behavior Spanning Seconds to Millennia [Poster]

Type: Poster

Room: Ballroom

Date: 4/20/2023

Presentation Time: 08:00 AM Pacific

Presenting Author: Katherine Scharer

Description

We explore how alluvial terraces form in the sheltered-corner (SC) of streams offset along strike slip faults as a function of the accumulation displacement (D) and of temporal variations of the stream erosive power (E). We combine new mapping and ¹⁰Be dating of the Littlerock Creek (LRC) terrace sequence at its intersection with the San Andreas Fault (SAF), and analysis of published flume experiments of strike-slip displaced streams. River incision at LRC is driven by rapid uplift and transpression. Elevyn terraces abandoned between 36 and 12ka are preserved in the SC on the downstream side of the SAF, only two of which are preserved on the upstream side. The fluvial architecture shows two sets of narrow terraces that measured parallel to fault strike are elevated along a similar gradient of ~6º. These sets are separated by two terraces that are 3-times wider with much lower along-strike gradient, and correlate temporally with the upstream terraces.

Applying our analysis of the flume experiments to the LRC record, we propose that the steeper terrace gradient provides a measure of the ratio of the uplift to strike-slip component, and was acquired during steady-state intervals with equilibrium between E and D. The intervening two-terraces interval is interpreted as a transient period, during which the geomorphic system evolved from lower E and higher deflection angle (the angle the stream intersects the fault) to higher E and lower deflection angle, leading to the abandonment of the upstream terraces. This sequence of events is supported by independent paleoclimate records of water discharge along the LRC. These results suggest that the stream response to D is governed by the following feedbacks: (1) E determines the extent of upstream-facing scarp erosion, which in turn (2) controls the deflection angle, and hence (3) the geometry of the SC. Quantitative analysis of the SC terrace stratigraphy along strike-slip displaced entrenched drainages might thus be used to derive cumulative displacements even in the absence of upstream piercing-points.

Additional Authors