USGS Science⌕ Search
Home/Archive/USGS

USGS · 70278166

Suitability of ground-motion models for seismic hazard assessment in Puerto Rico and the U.S. Virgin Islands

Brad T. Aagaard·Morgan P. Moschetti·Kyle B. Withers
View original source ↗Export metadata ↓

Abstract

We perform linear mixed-effects analyses with a ground-motion dataset to evaluate how well ground-motion models (GMMs) fit active crustal, subduction interface, and subduction intraslab earthquakes in Puerto Rico and the US Virgin Islands. Most of the GMMs reproduce the spatial variation in peak ground motions with earthquake magnitude and rupture distance but predict ground motions 0.3–1.0 natural log units (35%–270%) greater than observations. Two GMMs developed for Puerto Rico that are based on ground-motion records from mostly magnitude 4–5 earthquakes do not perform as well. We attribute the period-dependent overprediction in observed ground motions to differences between the observed site response and the linear site response in the GMMs. Consequently, we developed region-specific GMMs by adjusting the period-dependent linear site response coefficients and period-dependent constant coefficients to remove most of the bias between predicted ground motions and observations. For the analysis, we compile ground-motion records and process waveforms to build a dataset with 10,127 records at 72 stations from 849 magnitude 4.0–6.4 earthquakes between 1 January 2006 and 31 March 2024. The earthquakes include active crustal, subduction interface, and subduction intraslab events. We evaluate the GMMs using the time-averaged shear wave speed in the top 30 m ( ), which we compile from site surveys and proxy values computed from horizontal to vertical spectral ratios. Site terms exhibit strong consistency across GMMs and crustal and subduction earthquakes, indicating that the linear mixed-effects analysis successfully isolates the effects of local site response. The event terms show little spatial correlation and more substantial variability than in other regions, which we attribute to uncertainties in the earthquake magnitudes. This analysis guides the selection of the GMMs for the 2025 update of the National Seismic Hazard Model for Puerto Rico and the US Virgin Islands.

Explore related subjects

Earthquake SpectraArticle

Keep this discovery

BibTeXRIS

Brad T. Aagaard, Morgan P. Moschetti, Kyle B. Withers. 2026-04-09. Suitability of ground-motion models for seismic hazard assessment in Puerto Rico and the U.S. Virgin Islands. https://doi.org/10.1002/esp4.70047

Cite the original work for its findings. Save a collection to share your selection of sources.

Discover connections

Article subjectEarthquake Spectra subjectPuerto Rico, U.S. Virgin Islands placeUnited STates place

Connections use source metadata and explicit phrase matches, not verified experimental comparisons.

KEEP EXPLORING

Related discoveries

USGS2026

Evaluating the U.S. Geological Survey’s earthquake shaking hazard forecasts and their implications on seismic risk

We analyze the last six update cycles of the long-term probabilistic earthquake shaking hazard forecast from the U.S. Geological Survey (USGS) and discuss the changes in hazard estimates from the 1996 to the latest 2023 update of the National Seismic Hazard Model (NSHM) for the conterminous U.S. We summarize how our understanding of earthquake hazards has evolved over the last two decades and quantify the implications of changing hazards on people, buildings, and lifeline infrastructure. The net effect of changes in hazard estimates, with reference to the mean hazard from the 2023 NSHM update that has 2% probability of exceedance in 50 years, suggests an overall increase in total geographic areas facing very strong shaking levels (modified Mercalli Intensity (MMI) of VII or more) when compared with any of the previous cycles. This increase puts ~159.2 million people (an increase from 111.8 million when using the 2018 NSHM), ~47.3 million residential buildings (an increase from 32.5 million with the 2018 NSHM), and ~$25.6 trillion of economic exposure of all buildings (an increase from $17.4 trillion when using the 2018 NSHM) at risk from earthquakes. We demonstrate that small changes in hazard estimates do not necessarily imply small changes in risk estimates. Therefore, changes in risk estimates can be used to highlight key USGS NSHM updates for improving risk mitigation efforts.

Earthquake Spectra Journal↗
USGS2026

Opportunities for the U.S. Geological Survey’s National Seismic Hazard Model to improve seismic risk assessment of critical infrastructure.

As fragility and risk modeling techniques and computational capabilities evolve, complemented by moving toward more routine and systematic seismic risk assessment of all buildings and critical infrastructure, the authors pose a few critical questions to investigate how the U.S. Geological Survey (USGS) National Seismic Hazard Models (NSHMs) can be used and enhanced further to serve such issues. In this paper, we use three examples from multiple sectors to (1) identify the role of USGS NSHMs in evaluating seismic risks to critical infrastructure, (2) quantify potential impacts from NSHM enhancements (i.e., [i] hazard curves for the vertical component of ground motion, [ii] stochastic event sets, and [iii] maps of probabilistic ground failure hazards), and (3) clarify the feasibility of relevant NSHM improvements. We illuminate that NSHMs are commonly used in location-specific performance assessments, whereas earthquake effects on critical infrastructure can be widespread across large geospatial regions. Further, we found that without the NSHM extensions considered here, risk can be severely underestimated, e.g., neglecting ground failure hazards can underestimate regional loss by a factor of two or more. Although many challenges remain, we developed example prototypes to clarify the feasibility of the NSHM extensions, which can facilitate improved management of risks to critical infrastructure.

Earthquake Spectra Journal↗
USGS2025

Risk implications of Poisson assumptions and declustering inferred from a fully time-dependent earthquake forecast

We use the Third Uniform California Earthquake Rupture Forecast Epidemic Type Aftershock Sequence model, which is fully time-dependent in terms of including spatiotemporal clustering, to evaluate the effects of the Poisson assumption and declustering algorithms on statewide loss exceedance curves. The model is simulation based, meaning it produces synthetic catalogs that exhibit realistic behavior with respect to aftershocks and multi-fault earthquakes. A Poisson version of the model was constructed by randomizing event times, and the influence of two declustering algorithms was examined as well. We demonstrate that the probability of one-or-more loss exceedances (occurrence exceedance probability) is greater for the Poisson model because it has fewer seismically quiet time windows. The discrepancy between dollar loss estimates with a given exceedance probability is up to a factor of 32% but varies depending on the loss threshold (the x-axis value) and the forecast duration (we examined a range between 24 h and 50 years, with the discrepancy for the latter being negligible). We discuss how the one-or-more loss exceedance metric is questionable because it ignores all but the maximum loss experienced in each timeframe. An alternative metric based on total aggregate loss in each time window (aggregate exceedance probability) was therefore also examined, for which the Poisson model again implies higher risk at intermediate losses but lower risk at higher losses (because large, triggered events now contribute to total aggregate losses for the fully time-dependent model). We also argue that declustering is not a scientifically justifiable way to deal with full time dependence, in agreement with a chorus from other recent studies. It is difficult to draw generally applicable conclusions from our study, in part because application specific details will likely be important, but our results highlight how full time dependence can be reckoned with once authoritative forecast models are made available.

California↗