Uniform California Earthquake Rupture Forecast Research Articles

A Comprehensive Fault-System Inversion Approach: Methods and Application to NSHM23

ABSTRACT We present updated inversion-based fault-system solutions for the 2023 update to the National Seismic Hazard Model (NSHM23), standardizing earthquake rate model calculations on crustal faults across the western United States. We build upon the inversion methodology used in the Third Uniform California Earthquake Rupture Forecast (UCERF3) to solve for time-independent rates of earthquakes in an interconnected fault system. The updated model explicitly maps out a wide range of fault recurrence and segmentation behavior (epistemic uncertainty), more completely exploring the solution space of viable models beyond those of UCERF3. We also improve the simulated annealing implementation, greatly increasing computational efficiency (and thus inversion convergence), and introduce an adaptive constraint weight calculation algorithm that helps to mediate between competing constraints. Hazard calculations show that ingredient changes (especially fault and deformation models) are the primary driver of hazard changes between NSHM23 and UCERF3. Updates to the inversion methodology are also consequential near faults in which the slip rate in UCERF3 was poorly fit or was satisfied primarily using large multifault ruptures that are now restricted by explicit b-value and segmentation constraints.

The New Zealand Paleoseismic Site Database, Version 1.0

AbstractSite-specific paleoseismic data are key datasets underpinning the parameterization of large (Mw∼7+) earthquakes on faults in seismic hazard models. For the New Zealand National Seismic Hazard Model 2022 Revision Project (NZ NSHM 2022), a new database, the NZ Paleoseismic Site Database (NZ PSDB), was developed that contains paleoseismic (slip rate, earthquake timings, recurrence interval [RI], and single-event displacement [SED]) data for crustal (upper plate) active faults. The first edition version 1.0 (v.1.0) was modeled on datasets underpinning the Third Uniform California Earthquake Rupture Forecast (UCERF3) and builds on and significantly improves previous NZ compilations. Key improvements include documentation of the data underlying these parameters, such as offset measurements and ages for slip rate, and assignment of quality rankings for key attributes. The NZ PSDB v.1.0 contains 2811 records, subdivided into slip rate (871), earthquake timings (953), and SED (987); the earthquake timings dataset also includes 320 last event and 98 RI records. Because of time constraints, the most effort went into compiling the slip-rate records, in priority from highest to lowest slip rates, and the least amount of time was spent on the SED dataset, prioritizing records from historical earthquakes. The compilation also aimed to obtain as much geographical spread as possible, but the records only lie on a relatively small (maximum, 24%) number of faults and fault sections in the NZ Community Fault Model v.1.0. The most significant spatial gaps are offshore and in the southern and western South Island. The NZ PSDB v.1.0 contains both published and unpublished data, and many sites were relocated using high-resolution basemaps. However, radiocarbon ages were not recalibrated using a consistent calibration curve. Developing a process to efficiently recalibrate radiocarbon ages, as well as filling key data gaps, are current focuses of work toward v.2.0.

Validation of earthquake ground-motion models in southern California, USA, using precariously balanced rocks

Accurate estimates of earthquake ground shaking rely on uncertain ground-motion models derived from limited instrumental recordings of historical earthquakes. A critical issue is that there is currently no method to empirically validate the resultant ground-motion estimates of these models at the timescale of rare, large earthquakes; this lack of validation causes great uncertainty in ground-motion estimates. Here, we address this issue and validate ground-motion estimates for southern California utilizing the unexceeded ground motions recorded by 20 precariously balanced rocks. We used cosmogenic 10Be exposure dating to model the age of the precariously balanced rocks, which ranged from ca. 1 ka to ca. 50 ka, and calculated their probability of toppling at different ground-motion levels. With this rock data, we then validated the earthquake ground motions estimated by the Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3) seismic-source characterization and the Next Generation Attenuation (NGA)-West2 ground-motion models. We found that no ground-motion model estimated levels of earthquake ground shaking consistent with the observed continued existence of all 20 precariously balanced rocks. The ground-motion model I14 estimated ground-motion levels that were inconsistent with the most rocks; therefore, I14 was invalidated and removed. At a 2475 year mean return period, the removal of this invalid ground-motion model resulted in a 2−7% reduction in the mean and a 10−36% reduction in the 5th−95th fractile uncertainty of the ground-motion estimates. Our findings demonstrate the value of empirical data from precariously balanced rocks as a validation tool for removing invalid ground-motion models and, in turn, reducing the uncertainty in earthquake ground-motion estimates.

Lower Seismogenic Depth Model for Western U.S. Earthquakes

Abstract We present a model of the lower seismogenic depth of earthquakes in the western United States (WUS) estimated using the hypocentral depths of events M > 1, a crustal temperature model, and historical earthquake rupture depth models. Locations of earthquakes are from the Advanced National Seismic System Comprehensive Earthquake Catalog from 1980 to 2021 supplemented with seismicity in southern California for event hypocenters that were relocated by Hauksson et al. (2012) to obtain higher precision and better resolution in the model. We calculated the average depth of the deepest 10% of the merged catalog using an adaptive radius of 50 km or more. Along the San Andreas fault, the deepest seismogenic depths are located at 23 km around the Cholame segment, whereas the shallowest depths are located at about 10 km along the Rodgers Creek and Maacama faults. For the WUS outside California, the depth generally varies between 10 and 25 km with an average around 14 km but could extend to 35 km along Cascadia subduction zone. We find good agreement between the small-magnitude depths and rupture depths derived from coseismic slip of large earthquakes across the region. Our estimates are generally deeper than the previous seismogenic depths determined for the Uniform California Earthquake Rupture Forecast, Version 3 model based on work by Petersen et al. (1996) who used seismicity cross sections along major fault zones in California. Our new seismogenic depth distribution correlates closely with crustal temperature derived from WUS heat flow (Blackwell et al., 2011). This correlation allowed us to develop a map of the brittle–ductile transition that we use to replace seismogenic depths in the model east of the Intermountain West Seismic Belt where the seismicity rate is low. This updated depth model is useful for recalibrating the lower geologic fault rupture depths, and constraining deformation and seismicity source models in updates of the U.S. Geological Survey National Seismic Hazard Model.

Creep Rate Models for the 2023 US National Seismic Hazard Model: Physically Constrained Inversions for the Distribution of Creep on California Faults

Abstract 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.

Mechanical Models of Fault-Slip Rates in the Transverse and Peninsular Ranges, California

AbstractThe Transverse and Peninsular Ranges of Southern California host a geometrically complex network of seismically active faults with a range of slip senses. Here, we present 3D mechanical models of this region that are driven by the total Pacific-North American plate motion and slip on the San Andreas, San Jacinto, and Garlock faults. Based on these boundary conditions, we solve for the full 3D distribution of slip rates on 83 faults and compare model slip-rate predictions to long-term slip-rate data from Uniform California Earthquake Rupture Forecast version 3 (UCERF3) as well as individual geologic point-based estimates of long-term slip. About 46/68 (68%) model-predicted average fault-slip rates are within the UCERF3 slip-rate ranges with a root mean squared error (RMSE) of 1.03 mm/yr to the nearest (upper or lower) UCERF3 limit. The largest slip-rate discrepancies occur on the San Gabriel, Pine Mountain, and Big Pine faults, which may be presently inactive. We find that removing these three faults (i.e., assuming they are inactive) results in an improvement in all goodness-of-fit metrics with 47/63 (75%) UCERF3 slip rates fit with an RMSE of 0.40 mm/yr. We also compare this same model to existing point-based long-term slip-rate estimates and fit 39/69 (60%) with an RMSE of 0.79 mm/yr. The ability for the model to fit independent slip-rate data implies that strike slip along the “Big Bend” of the San Andreas fault (in conjunction with slip on the Garlock and San Jacinto faults) along with total plate motion is sufficient to reproduce both reverse-slip rates throughout the Transverse ranges and strike-slip rates in the Peninsular Ranges regions without additional driving forces needed. Overall, the models provide slip rates and distributions for 83 faults, including faults that currently do not have geologic slip-rate estimates and/or may not produce interseismic deformation.

GPS Velocity Field of the Western United States for the 2023 National Seismic Hazard Model Update

Abstract Global Positioning System (GPS) velocity solutions of the western United States (WUS) are compiled from several sources of field networks and data processing centers for the 2023 U.S. Geological Survey National Seismic Hazard Model (NSHM). These solutions include both survey and continuous-mode GPS velocity measurements. I follow the data processing procedure of Parsons et al. (2013) for the Uniform California Earthquake Rupture Forecast, version 3 and McCaffrey, Bird, et al. (2013) and Zeng and Shen (2013) for their WUS deformation models in support of the 2014 NSHM update. All GPS velocity vectors are first rotated to a common North American reference frame. I edit the velocities to remove outliers and data with significant influence from volcanism. The solutions are then combined into a final GPS velocity field consisting of 4979 horizontal velocity vectors. I compute strain rates based on these GPS velocities using the method of Shen et al. (2015). These strain rates correlate closely with seismicity rates in the WUS. The results are used for WUS geodetic and geologic deformation modeling in support of the 2023 NSHM update.

Enumerating Plausible Multifault Ruptures in Complex Fault Systems with Physical Constraints

ABSTRACT We propose a new model for determining the set of plausible multifault ruptures in an interconnected fault system. We improve upon the rules used in the Third Uniform California Earthquake Rupture Forecast (UCERF3) to increase connectivity and the physical consistency of ruptures. We replace UCERF3’s simple azimuth change rules with new Coulomb favorability metrics and increase the maximum jump distance to 15 km. Although the UCERF3 rules were appropriate for faults with similar rakes, the Coulomb calculations used here inherently encode preferred orientations between faults with different rakes. Our new rules are designed to be insensitive to discretization details and are generally more permissive than their UCERF3 counterparts; they allow more than twice the connectivity compared with UCERF3, yet heavily penalize long ruptures that take multiple improbable jumps. The set of all possible multifault ruptures in the California fault system is nearly infinite, but our model produces a tractable set of 326,707 ruptures (a modest 29% increase over UCERF3, despite the greatly increased connectivity). Inclusion in the rupture set does not dictate that a rupture receives a significant rate in the final model; rupture rates are subsequently determined by data constraints used in an inversion. We describe the rupture building algorithm and its components in detail and provide comparisons with ruptures generated by a physics-based multicycle earthquake simulator. We find that greater than twice as many ruptures generated by the simulator violate the UCERF3 rules than violate our proposed model.

Diaphragm abutment Californian bridges subjected to UCERF2 rupture scenarios: Complete damage state evolution with improvements to seismic codes

The seismic design of bridges in California has evolved largely around 1970 to 1990 in terms of transverse reinforcement spacing and ductile detailing. This study focuses on diaphragm abutment bridges with design criteria ranging from pre-70's design to post 90's design in California and subjecting them to future possible earthquakes on California faults, according to the Uniform California Earthquake Rupture Forecast (UCERF). Using the synthetic ground motion from Southern California (Elysian Park, San Jacinto, and Southern San Andreas) and Northern California (Hayward, and Northern San Andreas) at DesignSafe data repository as base excitation, fragility analysis of diaphragm abutment bridges within Los Angeles, San Bernardino, and San Francisco regions is presented. The scenario with the highest threat to bridges in each of the 3 regions is highlighted. Furthermore, the result reveals the importance of the orientation of bridges on the median fragility for the complete damage limit state and compares them with HAZard US (HAZUS) fragility parameters.

The Seismic Hazard Implications of Declustering and Poisson Assumptions Inferred from a Fully Time-Dependent Model

ABSTRACT We use the Third Uniform California Earthquake Rupture Forecast (UCERF3) epidemic-type aftershock sequence (ETAS) model (UCERF3-ETAS) to evaluate the effects of declustering and Poisson assumptions on seismic hazard estimates. Although declustering is necessary to infer the long-term spatial distribution of earthquake rates, the question is whether it is also necessary to honor the Poisson assumption in classic probabilistic seismic hazard assessment. We use 500,000 yr, M ≥ 2.5 synthetic catalogs to address this question, for which UCERF3-ETAS exhibits realistic spatiotemporal clustering effects (e.g., aftershocks). We find that Gardner and Knopoff (1974) declustering, used in the U.S. Geological Survey seismic hazard models, lowers 2% in 50 yr and risk-targeted ground-motion hazard metrics by about 4% on average (compared with the full time-dependent [TD] model), with the reduction being 5% at 40% in 50 yr ground motions. Keeping all earthquakes and treating them as a Poisson process increases these same hazard metrics by about 3%–12%, on average, due to the removal of relatively quiet time periods in the full TD model. In the interest of model simplification, bias minimization, and consideration of the probabilities of multiple exceedances, we agree with others (Marzocchi and Taroni, 2014) that we are better off keeping aftershocks and treating them as a Poisson process rather than removing them from hazard consideration via declustering. Honoring the true time dependence, however, will likely be important for other hazard and risk metrics, and this study further exemplifies how this can now be evaluated more extensively.

Impacts on catastrophe risk assessments from multi-segment and multi-fault ruptures in the UCERF3 model

The Uniform California Earthquake Rupture Forecast Version 3 (UCERF3) relaxes fault segmentation, allowing multi-segment and multi-fault ruptures through fault-to-fault “jumps,” with lengths up to ∼1200 km along the San Andreas Fault. Local faults are also highly interconnected, including ruptures on the order of hundreds of kilometers. These prescribed long ruptures did not exist in older models. Longer ruptures produce larger aggregate loss estimates for geographically dispersed assets (portfolios) due to the wider areas that are affected by strong ground shaking. In this study, we model probabilistic earthquake losses of a hypothetical state-wide building portfolio in California. We develop risk deaggregation methods to identify multi-segment and multi-fault ruptures that contribute significantly to high portfolio-wide risks. Three risk measures that are commonly used in risk management decisions are examined: Average Annual Loss (AAL), Return Period Loss (RPLα), and Tail Conditional Expectation (TCEα), for an annual exceedance probability “α,” or corresponding return period “1/α.” Our results show that while the super long ruptures (>500 km) contribute modestly (∼7%) to the portfolio AAL estimate, they contribute more significantly to portfolio catastrophe risk estimates. Specifically, at a 250 year return period, these long ruptures contribute about 26% and 32% to RPL250 and TCE250 estimates, respectively. At a 500-year return period, the corresponding contributions reach about 35% and 39%. Ruptures that connect complex fault systems are also found to be highly influential to estimated portfolio risks. At a 500-year return period, a mere six rupture groups contribute nearly 70% to the RPL500 estimate. Due to the importance of the UCERF3 model to many risk management and public policy decisions, a critical examination of the limit and uncertainty of fault connectivity and rupture lengths of future earthquakes, as well as their impacts on catastrophe risk assessments, is warranted in future model updates.

Improvements to the Third Uniform California Earthquake Rupture Forecast ETAS Model (UCERF3-ETAS)

Abstract We describe recent improvements to the Third Uniform California Earthquake Rupture Forecast ETAS Model (UCERF3-ETAS), which continues to represent our most advanced and complete earthquake forecast in terms of relaxing segmentation assumptions and representing multifault ruptures, elastic-rebound effects, and spatiotemporal clustering (the latter to represent aftershocks and otherwise triggered events). The two main improvements include adding aleatory variability in aftershock productivity and the option to represent off-fault events with finite-rupture surfaces. We also summarize the studies that led to these modifications, and reflect on how past and future uses of the model can improve our understanding of earthquake processes and the hazards and risks they pose.

Estimating Earthquake Probabilities by Jaynes’s Method of Maximum Entropy

ABSTRACT An important midcentury statistical method, already applied in several other scientific disciplines, enables epistemic ignorance to be incorporated more fully in multidecadal earthquake forecasts. Codified by the physicist E. T. Jaynes in the 1950s, the method of maximum entropy (MaxEnt) is suited for settings of great epistemic uncertainty. Shortly after its initial formulation it was being applied by industrial engineers to rates of failure and expected lifetimes—engineering problems analogous to fault ruptures and their recurrence intervals. Here, I show how MaxEnt can improve upon previous estimates of time-dependent conditional probabilities in Cascadia, California, and New Zealand. The method formalizes probabilistic inferences from long geology-based earthquake histories as truncated-Gaussian curves of time-dependent hazard that match the observed mean and mean square of the historical recurrence data. For each example, rupture forecasts by Brownian passage time, lognormal, Weibull, or Poisson methods have been published previously. In the Cascadia example, the MaxEnt estimate for a 50 yr exposure, 8.3%, compares with previous estimates of up to 14% for passage time or lognormal models. For other datasets drawn from the third Uniform California Earthquake Rupture Forecast study of California earthquakes and from the southern Alpine fault of New Zealand, MaxEnt recurrence probabilities similarly differ by up to a factor of nearly 2.0 from estimates published earlier. The lognormal approach is a variant of the MaxEnt model in which the estimate’s constraints concern the logarithms of the observed recurrence intervals; there is a continuum of methods in between. The discrepancies between MaxEnt and these other probability claims are not trivial; they arise from assumptions built into other methods that do not correspond to the actual information content of the data at hand. Discrepancies like these are directly relevant to the communication of these aspects of earthquake hazards to stakeholders concerned with the consequent risks.

Rupture Passing Probabilities at Fault Bends and Steps, with Application to Rupture Length Probabilities for Earthquake Early Warning

ABSTRACTEarthquake early warning (EEW) systems can quickly identify the beginning of a significant earthquake rupture, but the first seconds of seismic data have not been found to predict the final rupture length. We present two approaches for estimating probabilities of rupture length given the rupture initiation from an EEW system. In the first approach, bends and steps on the fault are interpreted as physical mechanisms for rupture arrest. Arrest probability relations are developed from empirical observations and depend on bend angle and step size. Probability of arrest compounds serially with increasing rupture length as bends or steps are encountered. In the second approach, time-independent rates among ruptures from the Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3), are interpreted to apply to the time-dependent condition in which rupture grows from a known starting point. Length probabilities from a Gutenberg–Richter magnitude–frequency relation provide a reference of comparison. We illustrate the new approach using the discretized fault model for California developed for UCERF3. For the case of rupture initiating on the southeast end of the San Andreas fault we find the geometric complexity of the Mill Creek section impedes most ruptures, and only ∼5% are predicted to reach to San Bernardino on the eastern edge of the greater Los Angeles region. Conditional probabilities of length can be precompiled in this manner for any initiation point on the fault system and thus are of potential value in seismic hazard and EEW applications.

Generalizing the Inversion-Based PSHA Source Model for an Interconnected Fault System

ABSTRACTThis article represents a step toward generalizing and simplifying the procedure for constructing an inversion-based seismic hazard source model for an interconnected fault system, including the specification of adjustable segmentation constraints. A very simple example is used to maximize understandability and to counter the notion that an inversion approach is only applicable when an abundance of data is available. Also exemplified is how to construct a range of models to adequately represent epistemic uncertainties (which should be a high priority in any hazard assessment). Opportunity is also taken to address common concerns and misunderstandings associated with the third Uniform California Earthquake Rupture Forecast, including the seemingly disproportionate number of large-magnitude events, and how well hazard is resolved given the overall problem is very underdetermined. However, the main aim of this article is to provide a general protocol for constructing such models.

More Fault Connectivity Is Needed in Seismic Hazard Analysis

ABSTRACTDid the third Uniform California Earthquake Rupture Forecast (UCERF3) go overboard with multifault ruptures? Schwartz (2018) argues that there are too many long ruptures in the model. Here, I address his concern and show that the UCERF3 rupture-length distribution matches empirical data. I also present evidence that, if anything, the UCERF3 model could be improved by adding more connectivity to the fault system. Adding more connectivity would improve model misfits with data, particularly with paleoseismic data on the southern San Andreas fault; make the model less characteristic on the faults; potentially improve aftershock forecasts; and reduce model sensitivity to inadequacies and unknowns in the modeled fault system.

Pseudoprospective Evaluation of UCERF3-ETAS Forecasts during the 2019 Ridgecrest Sequence

ABSTRACTThe 2019 Ridgecrest sequence provides the first opportunity to evaluate Uniform California Earthquake Rupture Forecast v.3 with epidemic-type aftershock sequences (UCERF3-ETAS) in a pseudoprospective sense. For comparison, we include a version of the model without explicit faults more closely mimicking traditional ETAS models (UCERF3-NoFaults). We evaluate the forecasts with new metrics developed within the Collaboratory for the Study of Earthquake Predictability (CSEP). The metrics consider synthetic catalogs simulated by the models rather than synoptic probability maps, thereby relaxing the Poisson assumption of previous CSEP tests. Our approach compares statistics from the synthetic catalogs directly against observations, providing a flexible approach that can account for dependencies and uncertainties encoded in the models. We find that, to the first order, both UCERF3-ETAS and UCERF3-NoFaults approximately capture the spatiotemporal evolution of the Ridgecrest sequence, adding to the growing body of evidence that ETAS models can be informative forecasting tools. However, we also find that both models mildly overpredict the seismicity rate, on average, aggregated over the evaluation period. More severe testing indicates the overpredictions occur too often for observations to be statistically indistinguishable from the model. Magnitude tests indicate that the models do not include enough variability in forecasted magnitude-number distributions to match the data. Spatial tests highlight discrepancies between the forecasts and observations, but the greatest differences between the two models appear when aftershocks occur on modeled UCERF3-ETAS faults. Therefore, any predictability associated with embedding earthquake triggering on the (modeled) fault network may only crystalize during the presumably rare sequences with aftershocks on these faults. Accounting for uncertainty in the model parameters could improve test results during future experiments.

Operational Earthquake Forecasting during the 2019 Ridgecrest, California, Earthquake Sequence with the UCERF3-ETAS Model

AbstractThe first Uniform California Earthquake Rupture Forecast, Version 3–epidemic-type aftershock sequence (UCERF3-ETAS) aftershock simulations were running on a high-performance computing cluster within 33 min of the 4 July 2019 M 6.4 Searles Valley earthquake. UCERF3-ETAS, an extension of the third Uniform California Earthquake Rupture Forecast (UCERF3), is the first comprehensive, fault-based, epidemic-type aftershock sequence (ETAS) model. It produces ensembles of synthetic aftershock sequences both on and off explicitly modeled UCERF3 faults to answer a key question repeatedly asked during the Ridgecrest sequence: What are the chances that the earthquake that just occurred will turn out to be the foreshock of an even bigger event?As the sequence unfolded—including one such larger event, the 5 July 2019 M 7.1 Ridgecrest earthquake almost 34 hr later—we updated the model with observed aftershocks, finite-rupture estimates, sequence-specific parameters, and alternative UCERF3-ETAS variants. Although configuring and running UCERF3-ETAS at the time of the earthquake was not fully automated, considerable effort had been focused in 2018 on improving model documentation and ease of use with a public GitHub repository, command line tools, and flexible configuration files. These efforts allowed us to quickly respond and efficiently configure new simulations as the sequence evolved. Here, we discuss lessons learned during the Ridgecrest sequence, including sensitivities of fault triggering probabilities to poorly constrained finite-rupture estimates and model assumptions, as well as implications for UCERF3-ETAS operationalization.

Relaxing Segmentation on the Wasatch Fault Zone: Impact on Seismic Hazard

ABSTRACTThe multisegment Wasatch fault zone is a well-studied normal fault in the western United States that has paleoseismic evidence of recurrent Holocene surface-faulting earthquakes. Along the 270 km long central part of the fault, four primary structural complexities provide possible along-strike limits to these ruptures and form the basis for models of fault segmentation. Here, we assess the impact that the Wasatch fault segmentation model has on seismic hazard by evaluating the time-independent long-term rate of ruptures on the fault that satisfy fault-slip rates and paleoseismic event rates, adapting standard inverse theory used in the Uniform California Earthquake Rupture Forecast, Version 3, and implementing a segmentation constraint in which ruptures across primary structural complexities are penalized. We define three models with varying degrees of rupture penalization: (1) segmented (ruptures confined to individual segments), (2) penalized (multisegment ruptures allowed, but penalized), and (3) unsegmented (all ruptures allowed). Seismic-hazard results show that, on average, hazard is highest for the segmented model, in which seismic moment is accommodated by frequent moderate (moment magnitude Mw 6.2–6.8) earthquakes. The unsegmented model yields the lowest average seismic hazard because part of the seismic moment is accommodated by large (Mw 6.9–7.9) but infrequent ruptures. We compare these results to model differences derived from other inputs such as slip rate and magnitude scaling relations and conclude that segmentation exerts a primary control on seismic hazard. This study demonstrates the need for additional geologic constraints on rupture extent and methods by which these observations can be included in hazard-modeling efforts.

Characteristic Earthquake Magnitude Frequency Distributions on Faults Calculated From Consensus Data in California

AbstractAn estimate of the expected earthquake rate at all possible magnitudes is needed for seismic hazard forecasts. Regional earthquake magnitude frequency distributions obey a negative exponential law (Gutenberg‐Richter), but it is unclear if individual faults do. We add three new methods to calculate long‐term California earthquake rupture rates to the existing Uniform California Earthquake Rupture Forecast version 3 efforts to assess method and parameter dependence on magnitude frequency results for individual faults. All solutions show strongly characteristic magnitude‐frequency distributions on the San Andreas and other faults, with higher rates of large earthquakes than would be expected from a Gutenberg‐Richter distribution. This is a necessary outcome that results from fitting high fault slip rates under the overall statewide earthquake rate budget. We find that input data choices can affect the nucleation magnitude‐frequency distribution shape for the San Andreas Fault; solutions are closer to a Gutenberg‐Richter distribution if the maximum magnitude allowed for earthquakes that occur away from mapped faults (background events) is raised above the consensus threshold of M = 7.6, if the moment rate for background events is reduced, or if the overall maximum magnitude is reduced from M = 8.5. We also find that participation magnitude‐frequency distribution shapes can be strongly affected by slip rate discontinuities along faults that may be artifacts related to segment boundaries.