Jessica R. Murray

Source–station distance is a central input to ground-motion models (GMMs) for predicting seismic shaking. GMM development uses distance metrics including the Joyner–Boore distance, which is the shortest distance from an observation point to the surface projection of the earthquake rupture, and Rrup the shortest distance to the rupture in three dimensions. Thompson and Baltay (2018) proposed the generalized mean rupture distance Rp to address observed near-fault ground-motion saturation. Rp accounts for the contribution to the shaking of all parts of the rupture and provides a simple method for incorporating spatially variable slip. They used Rp to develop a GMM for shallow crustal earthquakes, assuming uniform slip. Here, we investigate the improvement offered by an Rp-based GMM for large subduction interface earthquakes by recalibrating the path term for a published GMM (Parker et al., 2022) using Rp derived from distributed slip models (DSMs). Inspection of within-event and total residuals indicates that the recalibrated model fits the data at least as well as an alternative in which the path term was recalibrated using Rrup and the same dataset. Incorporating slip information in its entirety or trimming the DSM geometry is preferable to assuming uniform moment release on a prescribed model fault that extends beyond the actual ruptured area. Rp tuned to minimize model uncertainty is closer to the maximum distance to rupture for peak ground velocity (PGV) than acceleration, possibly reflecting the contribution to high-frequency shaking of slip on local asperities. However, when the moment is concentrated far from stations, PGV is fit adequately with Rp closer to Rrup. Our results suggest incorporating slip-derived moment release through Rp could improve GMMs for slab interface events, especially if the implementation of Rp-based models is refined using a larger dataset of earthquakes with greater geographic diversity to account for regional ground-motion variations.

The ShakeAlert® earthquake early warning (EEW) system is designed to warn users of imminent strong ground motion with sufficient time to take protective actions. ShakeAlert currently uses three algorithms to characterize the earthquake source. One estimates the location and magnitude using the first few seconds of the P wave and, while fast, tends to underestimate magnitude for Mw 7.0+ earthquakes. A second estimates the location, orientation, length, and corresponding magnitude of a line source using observed peak ground acceleration and contributes primarily to Mw 5.5+ earthquakes. The third infers earthquake magnitude from peak ground displacement measured using Global Navigation Satellite System (GNSS) data and offers nonsaturating magnitudes for Mw 7.0+ earthquakes. Other EEW algorithms exist that infer temporally evolving spatially variable slip on a 3D fault surface from real-time GNSS data, information that might enable more accurate and timely alerts in the event of large-magnitude subduction interface earthquakes. Here, we evaluate the potential contribution of one such algorithm, BEFORES (Minson et al., 2014), to improve ShakeAlert performance through a simulated real-time implementation of Bayesian evidence-based fault orientation and real-time earthquake slip (BEFORES) and other ShakeAlert algorithms using data for eight Mw 7.6+ earthquakes. The test results demonstrate that BEFORES can produce well-constrained and accurate magnitude estimates as soon as or sooner than other EEW algorithms, in turn enabling it to increase the amount of warning time users receive in many cases. However, with a modified Mercalli intensity (MMI) threshold of 3.5, which is commonly used for issuing alerts, BEFORES would tend to alert large geographic regions that did not feel strong shaking (MMI 6+). This effect can be mitigated using a higher alert threshold of MMI 4.5 without negative impact on the amount of warning time obtainable with BEFORES.

The 5 December 2024 Mw 7.0 Offshore Cape Mendocino earthquake was a challenging test of the U.S. West Coast ShakeAlert earthquake early warning system due to its offshore epicenter and limited near-source station coverage. We analyzed real-time performance of all components of the ShakeAlert system, including the seismic algorithms (earthquake point-source integrated code [EPIC] and Finite-fault rupture Detector [FinDer]), the geodetic algorithm (Geodetic First Approximation of Size and Time–peak ground displacement [GFAST-PGD]), and network telemetry during the event. EPIC created the first solution for this earthquake 15 s after origin time with an initial magnitude estimate of M 5.6 and location error of 10 km from the Advanced National Seismic System epicenter. An early spurious trigger from station CE.89101 fortuitously maintained location accuracy and, correspondingly, magnitude accuracy. FinDer contributed its first solution at 18 s with a location estimate closer to the seismic network and produced two distinct rupture geometries, leading to minor fluctuations in estimated intensity contours. GFAST-PGD did not meet alerting thresholds but otherwise performed as expected. Network latencies were <2 s for most stations, supporting the rapid detection of this earthquake by the system. Roughly five million alerts were delivered to cell phone devices in California and Oregon during this event. This was also the first instance of a school district-wide ShakeAlert-powered system being activated. Comparisons to recorded seismograms demonstrate that the maximum warning times before potentially damaging shaking (intensity 6+) were in the range of 5–55 s. Although the ShakeAlert system provided accurate solutions and useful alert delivery, this earthquake raised awareness of potential issues within the system, including the need for improved offshore location estimates, a combination of solutions from ShakeAlert servers, and handling of spurious triggers.

The ShakeAlert earthquake early warning (EEW) system partners along with U.S. Geological Survey (USGS) licensed operators deliver EEW alerts to the public and trigger automated systems when a significant earthquake is expected to impact California, Oregon, or Washington. ShakeAlert’s primary goal is to provide usable warning times before the arrival of damaging shaking. EEW is most likely to achieve this goal in large-magnitude earthquakes. In recent years, ShakeAlert has gone through a series of upgrades to its underlying scientific algorithms aimed at improved performance during large earthquakes. Version 3 of this software recently went live in the production system and includes improvements to all algorithms. The main seismic algorithms that detect an earthquake and characterize its location, magnitude, and fault rupture orientation are faster than older versions. Other key changes include: using real-time geodetic data to characterize the magnitude growth in large earthquakes; the introduction of an alert pause procedure to compromise between speed near the epicenter and improved accuracy at larger distances; and the inclusion of a nonergodic site-response model in the ground-motion predictions. ShakeAlert has achieved its primary goal of usable warning times before strong shaking at some locations in real-time operations in recent M 6 earthquakes. Using offline tests, we demonstrate usable warning times are possible for many sites with peak shaking values of modified Mercalli intensity (MMI) 7–8 in M 7+ earthquakes and also for many MMI 8–9 sites in M 8+ earthquakes. ShakeAlert partners use a variety of MMI and magnitude thresholds in deciding when to alert their users within bounds set by the USGS. Our study shows that there is room to raise the magnitude thresholds up to about M 5.5 without adversely affecting performance in large earthquakes. The ground-motion criteria are more complex owing to a significant drop-off in warning times between the MMI 4 and 5 levels of predicted shaking. However, widely used ShakeAlert products, such as the MMI 3 and 4 contour products, can provide sufficiently long warning times before strong shaking in moderate-to-great earthquakes to enable a range of protective actions.

The conference report accompanying the fiscal year (FY) 2022 Consolidated Appropriations Act (Public Law 117–103) for the U.S. Department of the Interior and related agencies directed the U.S. Geological Survey (USGS) to “work with the State of Alaska to develop an implementation plan to be completed within two years in order to put ShakeAlert/Earthquake Early Warning in Alaska” (p. 29). Congress included $1 million in the FY 2022 appropriation to conduct this effort.The USGS Earthquake Hazards Program, along with partner organizations, has developed the ShakeAlert earthquake early warning (EEW) system for the West Coast, which currently operates in California, Oregon, and Washington. The purpose of the system and its alert delivery partners is to reduce the impact of earthquakes and save lives and property by delivering ShakeAlert-powered alerts that are transmitted to the public via mass notification technologies, and by providing more detailed data streams to institutional users and commercial service providers to trigger automated, user-specific, protective actions.ShakeAlert was designed in such a way that it could be expanded to other U.S. regions with high earthquake risk, after the build-out of seismic and geodetic networks to support ShakeAlert in a specified region is completed and the necessary funding is secured for long-term operation and maintenance.When an earthquake occurs, seismic waves radiate from the rupturing fault like waves on a pond. It is these waves that people feel as earthquake shaking and that can cause damage to structures. Using networks of ground-motion sensors and sophisticated computer algorithms, ShakeAlert can detect an earthquake seconds after it begins, calculate its location and magnitude, and estimate the resulting intensity of shaking. Early warnings of impending shaking are then sent to people and systems that may experience damaging shaking, allowing them to take appropriate protective actions. Depending on the user’s distance from the earthquake, alerts may be delivered before, during, or after the arrival of strong shaking. There will almost always be a region near the earthquake epicenter where alerts arrive after damaging shaking has begun. The ShakeAlert system updates its ground-motion estimates as an earthquake grows larger.In response to the FY 2022 congressional direction, the USGS worked with the State of Alaska to devise this implementation plan for ShakeAlert expansion to Alaska. The USGS engaged with the Alaska Division of Homeland Security and Emergency Management (DHS&EM) and the Alaska Division of Geological and Geophysical Surveys (DGGS). A cooperative agreement was awarded to the Alaska Earthquake Center (AEC) at the University of Alaska Fairbanks (UAF) for their contributions to the plan and their work coordinating with other networks in Alaska. The USGS engaged with the Alaska Seismic Hazards Safety Commission (ASHSC) throughout the process. The USGS also held a series of Alaska stakeholder engagements. The process of developing the implementation plan was facilitated by contracted staff from Corner Alliance, which is a government consulting firm.This implementation plan describes the details and estimates the costs for a Phase 1 expansion of the ShakeAlert system to Alaska. A geographically limited Phase 1 goal was chosen that covers the highest risk and most populated areas of Alaska. The areas proposed encompass the State’s main population centers and 90 percent of the State’s population. This Phase 1 design is considered very challenging and ambitious from the viewpoint of network operators. The lessons learned if this plan is implemented could be used to consider subsequent phases to expand EEW beyond Phase 1 in Alaska in the future.ShakeAlert is built on the foundation of the sensor networks and data processing infrastructure of the USGS-led Advanced National Seismic System (ANSS). This implementation plan calls for a total of 450 high-quality, real-time EEW-capable ANSS seismic stations in Alaska: 270 new stations, 160 upgraded stations, and 20 existing stations. These seismic station numbers are based on a station spacing of 10 kilometers (km) in urban areas, 20 km in seismic source areas that endanger population centers, and 40 km in other areas. The associated costs also include support for some EEW-capable global navigation satellite system (GNSS) stations, with a focus on improving warnings for large subduction zone earthquakes. For effective EEW, ShakeAlert requires low-latency, high-availability, robust telemetry links to deliver continuous, real-time data from field stations to the data centers.The Alaska data processing hardware infrastructure would follow the general design for fail-safe operation that is used for the ShakeAlert system on the West Coast. The ShakeAlert architecture uses two independent layers: the production layer for earthquake processing and the alert layer to make alerting decisions and serve alerts to users. This implementation plan includes two geographically separated data centers in Alaska, each with two fully independent production and alert layers using the same system design developed for the West Coast. As of March 2024, the ShakeAlert system is at version 3.0.1, with more advanced versions in the development and testing pipeline. ShakeAlert originally used two algorithms to determine the location and magnitude of earthquakes using seismic data. A third algorithm that can calculate very large magnitudes of very large earthquakes with geodetic data was added in March 2024.ShakeAlert publishes several data and alert products to meet the needs of different users. All messages include the location of the earthquake, either as a point or a line, and its magnitude. Ground-shaking estimates are published in two forms, as ground-motion contours and a map grid. Providing adequate warning time for strong shaking (the “target threshold”)...

The fore-arc of the southern Cascadia subduction zone (CSZ), north of the Mendocino triple junction (MTJ), is home to a network of Quaternary-active crustal faults that accumulate strain due to the interaction of the North American, Juan de Fuca (Gorda), and Pacific plates. These faults, including the Little Salmon and Mad River fault (LSF and MRF) zones, are located near the most populated parts of California’s north coast and show paleoseismic evidence for three slip events of several-meter scale in the past 1700 yr. However, the geodetic slip rates of these faults are poorly constrained. In this work, we analyze a new compilation of interseismic geodetic velocities from Global Navigation Satellite Systems, leveling, and tide gauge data near the MTJ to constrain present-day slip deficit rates on upper-plate faults and coupling on the megathrust. We construct Green’s functions for interseismic slip deficit for discrete faults embedded in an elastic plate overlying a viscoelastic mantle. We then use a constrained least-squares inversion to determine best-fitting slip rates on the major faults and investigate slip rate trade-offs between faults. Results indicate that the LSF and MRF systems together accumulate 4–5 mm/yr of reverse-slip deficit, although their separate slip rates cannot be determined independently. Modeling of the horizontal and vertical velocities suggests that the southernmost CSZ is coupled interseismically to deeper than 25 km depth. We also find that 6–17 mm/yr of right-lateral slip deficit extends north of the MTJ and into the southern Cascadia fore-arc. These results reinforce the notion that both the southernmost Cascadia megathrust and the smaller fore-arc faults above it contribute to regional seismic hazard.

Widespread surface creep is observed across a number of active faults included in the United States (US) National Seismic Hazard Model (NSHM). In northern California, creep occurs on the central section of the San Andreas fault, along the Hayward and Calaveras faults through the San Francisco Bay Area, and to the north coast region along the Maacama and Bartlett Springs faults. In southern California, creep is observed across the Coachella segment of the San Andreas fault, through the Brawley Seismic Zone, and along the Imperial and Superstition Hills faults. Seismic hazard assessments for California have accounted for creep using various data and methods, including the most recent Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3) in 2013. The purpose of this study is to expand and update the UCERF3 creep rate data set for the 2023 release of the US NSHM and to invert geodetic data and the surface creep rate data for the spatial distribution of interseismic fault creep on California faults using an elastic model with physical creep constraints. The updated surface creep rate compilation consists of a variety of data types including alignment arrays, offset cultural markers, creepmeters, Interferometric Synthetic Aperture Radar, and Global Positioning System data. We compile a total of 497 surface creep rate measurements, 400 of which are new and 97 of which appear in the UCERF3 compilation. We compute creep rate distributions for each of the five 2023 NSHM geodetic-based and geologic-based deformation models. Computed creep rates are used to reduce the total fault moment rate available for earthquake sequences in the NSHM model. We find that, despite relatively large variability in model long-term slip rates across all five deformation models, the variability in depth-averaged creep rate across all models is relatively small, typically 5–10 mm/yr along the creeping San Andreas fault section and only 2–4 mm/yr along the Maacama and Rodgers Creek-Hayward faults.

This report describes geodetic and geologic information used to constrain deformation models of the 2023 update to the National Seismic Hazard Model (NSHM), a set of deformation models to interpret these data, and their implications for earthquake rates in the western United States. Recent updates provide a much larger data set of Global Positioning System crustal velocities than used in the 2014 NSHM, as well as hundreds of new faults considered as active sources for the 2023 NSHM. These data are interpreted by four geodetic models of deformation that estimate fault slip rates and their uncertainties together with off-fault moment release rates. Key innovations in the 2023 NSHM relative to past practice include (1) the addition of two new (in addition to two existing) deformation models, (2) the revision and expansion of the geologic slip rate database, (3) accounting for fault creep through development of a creep-rate model that is employed by the four deformation models, and (4) accounting for time-dependent earthquake-cycle effects through development of viscoelastic models of the earthquake cycle along the San Andreas fault and the Cascadia subduction zone. The effort includes development of a geologic deformation model that complements the four geodetic models. The current deformation models provide a new assessment of outstanding discrepancies between geologic and geodetic slip rates, at the same time highlighting the need for both geologic and geodetic slip rates to robustly inform the earthquake rate model.