Geology topics

Thomas E. Parsons

Publications and source records attributed to Thomas E. Parsons.

At least 19 recordsLinked to original sources

USGS2026

Dimensionality reduction techniques for analyzing tsunami simulations in hazard assessment and forecasting applications

A wide selection of linear and non-linear dimensionality reduction techniques is evaluated on synthetic tsunami data that is the basis for tsunami hazard assessment and computational forecasting. The data were computed from earthquake rupture forecasts (ERFs) supplying initial generation conditions and a linear long-wave Green's function approach to efficiently calculate tsunami propagation. The dimensionality reduction techniques include various forms of principal component analysis, manifold learning, and neural networks. Both external (e.g., mutual information) and internal (silhouette) metrics are used to evaluate clustering in the reduced dimension, focusing on peak-nearshore tsunami amplitudes (PNTA) along a nearshore isobath and time-series (marigrams) at specific nearshore locations. This evaluation is intentionally scoped to cluster-based downsampling to produce probabilistic inundation maps. For our test case along the Nankai subduction zone, manifold learning methods produced the highest clustering metric scores for PNTA profiles among the examined dimensionality reduction techniques. The marigram results are less consistent, owing primarily to uncertainty in defining appropriate clusters in the original high-dimensional (ambient) space to establish a “ground truth.” A novel application of using the integrative ERF-tsunami approach is that features extracted in the reduced domain, particularly from manifold learning, can be mapped back to rupture zones along the fault. This yields targeted fault information associated with unique PNTA profiles and marigram characteristics, such as late-arriving waves. Further testing would be needed for application-specific workflows and more complex fault systems, but results from this single subduction zone case study support the consideration of non-linear techniques (e.g., manifold learning) as alternatives to linear methods.

Nankai megathrust↗

USGS2026

Temporal and spatial changes in seismic attenuation associated with inferred fluid migration in the 2016 central Apennines earthquake sequence

Prior work suggests that high‐frequency seismic attenuation acts as a highly sensitive proxy for crustal permeability and fluid mobility in fractured media. We test the hypothesis that the fault system responsible for the 2016–2017 Amatrice–Visso–Norcia–Capitignano sequence acted as an impermeable seal, compartmentalizing pressurized fluids until dynamic rupture triggered widespread fluid diffusion. By tracking across the sequence the spatiotemporal evolution of the S ‐wave anelastic attenuation parameter, we identify large, positive low‐frequency attenuation anomalies emerging within the hanging wall following the Amatrice mainshock and strictly preceding subsequent large ruptures. Conversely, we observe weaker, negative anomalies in the footwall, anticorrelated in time with those of the hanging wall, revealing a massive asymmetry in fluid redistribution and permeability evolution across the fault system. Furthermore, aftershock migration rates reveal distinct linear alignments in a distance‐reduced time space, allowing us to explicitly track and quantify episodes of lateral and upward fluid migration. These physically consistent patterns suggest that stress‐driven fluid diffusion directly weakens adjacent fault patches, dictating the spatiotemporal migration of seismicity. We conclude that near‐real‐time monitoring of seismic attenuation may help detect fluid redistribution in active fault systems and may provide useful information for time‐dependent seismic hazard assessment.

central Apennines↗

USGS2026

Frameworks for assessing tsunami hazard and risk

Tsunamis are multiscale phenomena resulting from a water column displacement that may be induced by multiple sources, and range from local scale inundation processes to ocean-wide scale wave propagation. Different strategies may be required to model tsunami evolution at different scales and to characterize various intensity measures. Research in tsunami hazard and risk has focused mostly on the tsunami effects such as the wave heights or flow depths. This chapter reviews the evolution of tsunami hazard and risk assessment methodologies, with particular emphasis on the development of probabilistic approaches. Building on advances in numerical modeling and uncertainty analysis, two main frameworks for Probabilistic Tsunami Hazard and Risk Analysis (PTHA/PTRA) are described. Framework 1 (FW1) focuses on quantitative methods, including fully simulation-based assessments (FW1A), integration of hazard with vulnerability and loss models (FW1B), consideration of dynamic processes such as tidal and sea-level variations (FW1C), and approaches using limited scenario sets (FW1D). Framework 2 (FW2) complements this by incorporating indicator-based vulnerability assessments, both physical (FW2A) and social, multi-dimensional (FW2B).

Book chapter↗

USGS2025

An exploration of the relative influence of physical models for Omori’s law

Omori’s law states that the rate of aftershocks decays as a function of inverse time. There are multiple physical explanations that we reduce into a nonlinear mixed effects relation of three terms: (1) a Rate/State expression that can account for static/dynamic and viscoelastic triggering caused directly by the mainshock, (2) a fluid diffusion triggering term, and (3) a randomized secondary triggering (cascade) term. We fit free physical-model parameters to an observed aftershock sequence through two nonlinear regression methods to find the relative contributions of physics-based models in an observed aftershock sequence. Results from both methods show that Rate/State models overpredict aftershock rates by ∼0–30%. Secondary aftershocks cause a net negative contribution (seismicity rate reduction that corrects overprediction by other terms) ranging between ∼0 and 30%. All regression solutions yield negative secondary triggering contributions without being guided to do so. A physical explanation for this is that aftershock occurrence relieves stress from the crust, ultimately causing the sequence to extinguish itself. Fluid diffusion triggering contributions range from ∼0 to 20%. Diffusion processes are observed to be shorter in time than the full duration of an aftershock sequence and they are also spatially limited, diminishing their influence. Our results apply to an aftershock decay curve from the 2016 Central Apennines earthquake sequence, meaning that our specific results may not be general. Our primary conclusion is that any one physical model cannot alone fit the observed sequence as well as the combination of three we investigated.

Frontiers in Earth Science↗

USGS2025

A scientific vision and roadmap for earthquake rupture forecast developments, a USGS perspective

We articulate a scientific vision and roadmap for the development of improved Earthquake Rupture Forecast models, which are one of the two main modeling components used in modern seismic hazard and risk analysis. One primary future objective is to provide fully time-dependent models that include both elastic rebound and spatiotemporal clustering nationwide, which is particularly important for shorter-term hazard and risk considerations (e.g., earthquake insurance products). We also discuss the importance and perennial challenges associated with quantifying epistemic uncertainties, including those associated with deformation-model slip rates, un-quantified sampling errors with respect to off-fault seismicity, and any spatial covariances. The need for more physics-based approaches is also emphasized, as is the benefit of adding model valuation (quantifying usefulness) to our verification and validation protocols. Given the multidisciplinary and system-level nature of this activity, modular design is critical. Future updates will also draw from best-available science by both the United States Geological Survey and the external community. The primary goal of this paper is to highlight plans that guide research and facilitate community engagement with model development, especially with respect to lowering the entry barrier for early career scientists and engineers. The paper is written so readers can focus on the sections that interest them most (see table of contents), with the Introduction and Discussion providing a stand-alone overview and summary.

Bulletin of the Seismological Society of America↗

USGS2025

Do Graviquakes exist?

The “Graviquake” model, proposed in 2015 as an alternative to the elastic dislocation model, posits that normal faults are passive features dominated by coseismic gravitational collapse into a dilated crustal wedge, and that normal faulting is fundamentally distinct from strike‐slip and reverse faulting. Developed using finite‐element modeling before the 2016 central Apennines earthquake sequence, the model was revamped based on interpreted Differential Interferometric Synthetic Aperture Radar data from these events and used as evidence for a gravitational collapse episode. However, this interpretation relies on miscalculated elevation changes and is not corroborated by independent geophysical and seismological observations. Our analysis exposes fundamental flaws in the Graviquake model. By assuming that faults are passive players, it underrepresents the dynamic role of strain accumulation and release in rocks adjacent to faults. The hypothesized rapid expulsion of overpressurized fluids appears inconsistent with observed diffusion rates and lacks supporting seismological evidence. Part of the uplifted–subsided volume imbalance is likely an artifact arising from data processing, and in part is a transient effect due to the delayed response of the lower crust. Moment tensor analyses detect no isotropic components indicative of gravitational collapse, and observed ground motion and stress‐drop levels remain fully consistent with elastic dislocation theory. In addition, finite‐element modeling of normal faulting replicates observed surface deformation without invoking a collapsing wedge. The Graviquake model proposes a representation of normal‐faulting mechanics that differs significantly from established models and observations. Gravity does play a role in normal faulting, but the elastic dislocation theory remains the definitive framework of fault mechanics. Reinterpreting the 2016 earthquakes as a cascade of gravitational episodes, based on incorrect data processing and modeling, fails to substantiate the Graviquake hypothesis. Persistence in advocating this model could mislead seismic hazard assessment and undermine our understanding of normal faulting.

Bulletin of the Seismological Society of America↗

USGS2025

Identification of representative earthquakes for probabilistic tsunami hazard analysis (PTHA) using earthquake rupture forecasts and machine learning

As probabilistic tsunami hazard analysis (PTHA) focuses more on assessments for localized, populous regions, techniques are needed to identify a subsample of representative earthquake ruptures to make the computational requirements for producing high-resolution hazard maps tractable. Moreover, the greatest epistemic uncertainty in seismic PTHA is related to source characterization, which is often poorly defined and subjective. We address these two salient issues by applying streamlined earthquake rupture forecasts (ERFs), based on combinatorial optimization methods, to an unsupervised machine learning workflow for identifying representative ruptures. ERFs determine the optimal distribution of a millennia-scale sample of earthquakes by inverting the observed slip rate on major faults. We use two previously developed combinatorial optimization ERFs, integer programming and greedy sequential, to produce the optimal location of ruptures with seismic moments sampled from a regional Gutenberg–Richter magnitude–frequency distribution. These ruptures in turn are used to calculate peak nearshore tsunami amplitude, using computationally efficient tsunami Green's functions. An unsupervised machine learning workflow is then used to identify a small subsample of the earthquakes input to ERFs for onshore PTHA analysis. We eliminate epistemic uncertainty related to source distribution under traditional PTHA analysis; in its place, a quantifiable, less subjective and generally smaller uncertainty related to the input to ERFs is included. The Nankai subduction zone is used as a test case, where previous ERFs have been conducted. Results indicate that the locations of representative earthquakes are sensitive to choice of magnitude–area relation and to whether a minimum cumulative stress objective is imposed on the fault. In general, incorporating ERFs into PTHA provide a physically self-consistent method to incorporate fault slip information in determining representative earthquakes for onshore PTHA, eliminating a major source of epistemic uncertainty.

Nankai subduction zone↗

USGS2024

Seismic attenuation and stress on the San Andreas Fault at Parkfield: Are we critical yet?

The Parkfield transitional segment of the San Andreas Fault (SAF) is characterized by the production of frequent quasi-periodical M6 events that break the very same asperity. The last Parkfield mainshock occurred on 28 September 2004, 38 years after the 1966 earthquake, and after the segment showed a ∼22 years average recurrence time. The main reason for the much longer interevent period between the last two earthquakes is thought to be the reduction of the Coulomb stress from the M6.5 Coalinga earthquake of 2 May 1983, and the M6 Nuñez events of June 11th and 22 July 1983. Plausibly, the transitional segment of the SAF at Parkfield is now in the late part of its seismic cycle and current observations may all be relative to a state of stress close to criticality. However, the behavior of the attenuation parameter in the last few years seems substantially different from the one that characterized the years prior to the 2004 mainshock. A few questions arise: (i) Does a detectable preparation phase for the Parkfield mainshocks exist, and is it the same for all events? (ii) How dynamically/kinematically similar are the quasi-periodic occurrences of the Parkfield mainshocks? (iii) Are some dynamic/kinematic characteristics of the next mainshock predictable from the analysis of current data? (e.g., do we expect the epicenter of the next failure to be co-located to that of 2004?) (iv) Should we expect the duration of the current interseismic period to be close to the 22-year “undisturbed” average value? We respond to the questions listed above by analyzing the non-geometric attenuation of direct S-waves along the transitional segment of the SAF at Parkfield, in the close vicinity of the fault plane, between January 2001 and November 2023. Of particular interest is the preparatory behavior of the attenuation parameter as the 2004 mainshock approached, on both sides of the SAF. We also show that the non-volcanic tremor activity modulates the seismic attenuation in the area, and possibly the seismicity along the Parkfield fault segment, including the occurrence of the mainshocks.

California↗

USGS2023

Evaluating a prospective fault-based stress-transfer forecast for the M7.9 Wenchuan earthquake region 15 years later

Four days after the 12 May 2008 M 7.9 Wenchuan earthquake struck the Sichuan region of China, we submitted a prospective earthquake forecast based on transfer of stress from the mainshock onto significant faults crossing through populated areas. We identified where the largest aftershocks were likely to occur that could cause loss of life. We returned the revised article to the journal on 5 June 2008, marking the last day of our observation period. The primary testable features are locations and focal mechanisms of larger ( M ≥ 4.5) earthquakes; did these events happen on or very near the faults we said they would? Did they have the same strikes, dips, and rakes as the faults we modeled? In retrospect, is the stress transfer method consistent with all M ≥ 4.5 earthquakes that occurred? We find all but one M ≥ 4.5 aftershock with known focal mechanisms located on stress‐increased faults, and their focal mechanism parameters overlap with geological characteristics we used in making calculations. Six of the seven lethal M > 4.5 earthquakes that occurred in the region since 5 June 2008 were located on stress‐increased faults, with the lone exception triggered by hydraulic fracturing.

The Seismic Record↗

USGS2023

The weight of New York City: Possible contributions to subsidence from anthropogenic sources

New York City faces accelerating inundation risk from sea level rise, subsidence, and increasing storm intensity from natural and anthropogenic causes. Here we calculate a previously unquantified contribution to subsidence from the cumulative mass and downward pressure exerted by the built environment of the city. We enforce that load distribution in a multiphysics finite element model to calculate expected subsidence. Complex surface geology requires multiple rheological soil models to be applied; clay rich soils and artificial fill are calculated to have the highest post-construction subsidence as compared with more elastic soils. Minimum and maximum calculated building subsidence ranges from 0 to 600 mm depending on soil/rock physical parameters and foundation modes. We compare modeled subsidence and surface geology to observed subsidence rates from satellite data (Interferometric Synthetic Aperture Radar and Global Positioning System). The comparison is complicated because the urban load has accumulated across a much longer period than measured subsidence rates, and there are multiple causes of subsidence. Geodetic measurements show a mean subsidence rate of 1–2 mm/year across the city that is consistent with regional post-glacial deformation, though we find some areas of significantly greater subsidence rates. Some of this deformation is consistent with internal consolidation of artificial fill and other soft sediment that may be exacerbated by recent building loads, though there are many possible causes. New York is emblematic of growing coastal cities all over the world that are observed to be subsiding (Wu et al., 2022, https://doi.org/10.1029/2022GL098477 ), meaning there is a shared global challenge of mitigation against a growing inundation hazard.

New York↗

USGS2023

Combinatorial optimization of earthquake spatial distributions under minimum cumulative stress constraints

We determine optimal on‐fault earthquake spatial distributions using a combinatorial method that minimizes the long‐term cumulative stress resolved on the fault. An integer‐programming framework was previously developed to determine the optimal arrangement of a millennia‐scale earthquake sample that minimizes the misfit to a target slip rate determined from geodetic data. The resulting cumulative stress from just slip‐rate optimization, however, can greatly exceed fault strength estimates. Therefore, we add an objective function that minimizes cumulative stress and broad stress constraints to limit the solution space. We find that there is a trade‐off in the two objectives: minimizing the cumulative stress on a fault within fault strength limits concentrates earthquakes in specific areas of the fault and results in excursions from the target slip rate. Both slip‐rate and stress objectives can be combined in either a weighted or lexicographic (hierarchical) method. Using a combination of objectives, we demonstrate that a Gutenberg–Richter sample of earthquakes can be arranged on a constant slip‐rate finite fault with minimal stress and slip‐rate residuals. We apply this method to determine the optimal arrangement of earthquakes on the variable slip‐rate Nankai megathrust over 5000 yr. The sharp decrease in slip rate at the Tokai section of the fault results in surplus cumulative stress under all scenarios. Using stress optimization alone restricts this stress surplus to the northeast end of the fault at the expense of decreasing the slip rate away from the target slip rate at the southwest end of the fault. A combination of both slip‐rate and stress objectives provides an adequate fit to the data, although alternate model formulations for the fault are needed at the Tokai section to explain persistent excess cumulative stress. In general, incorporating stress objectives and constraints into the integer‐programming framework adds an important aspect of fault physics to the resulting earthquake rupture forecasts.

Bulletin of the Seismological Society of America↗

USGS2023

“Aftershock Faults” and what they could mean for seismic hazard assessment

We study stress‐loading mechanisms for the California faults used in rupture forecasts. Stress accumulation drives earthquakes, and that accumulation mechanism governs recurrence. Most moment release in California occurs because of relative motion between the Pacific plate and the Sierra Nevada block; we calculate relative motion directions at fault centers and compare with fault displacement directions. Dot products between these vectors reveal that some displacement directions are poorly aligned with plate motions. We displace a 3D finite‐element model according to relative motions and resolve stress tensors onto defined fault surfaces, which reveal that poorly aligned faults receive no tectonic loading. Because these faults are known to be active, we search for other loading mechanisms. We find that nearly all faults with no tectonic loading show increase in stress caused by slip on the San Andreas fault, according to an elastic dislocation model. Globally, faults that receive a sudden stress change respond with triggered earthquakes that obey an Omori law rate decay with time. We therefore term this class of faults as “aftershock faults.” These faults release ∼4% of the moment release in California, have ∼0.1%–5% probability of M 6.7 earthquakes in 30 yr, and have a 0.001%–1% 30 yr M 7.7 probability range.

California↗

USGS2022

On the use of high-resolution and deep-learning seismic catalogs for short-term earthquake forecasts: Potential benefits and current limitations

Enhanced earthquake catalogs provide detailed images of evolving seismic sequences. Currently, these data sets take some time to be released but will soon become available in real time. Here, we explore whether and how enhanced seismic catalogs feeding into established short-term earthquake forecasting protocols may result in higher predictive skill. We consider three enhanced catalogs for the 2016–2017 Central Italy sequence, featuring a bulk completeness lower by at least two magnitude units compared to the real-time catalog and an improved hypocentral resolution. We use them to inform a set of physical Coulomb Rate-and-State (CRS) and statistical Epidemic-Type Aftershock Sequence (ETAS) models to forecast the space-time occurrence of M3+ events during the first 6 months of the sequence. We track model performance using standard likelihood-based metrics and compare their skill against the best-performing CRS and ETAS models among those developed with the real-time catalog. We find that while the incorporation of the triggering contributions from new small magnitude detections of the enhanced catalogs is beneficial for both types of forecasts, these models do not significantly outperform their respective near real-time benchmarks. To explore the reasons behind this result, we perform targeted sensitivity tests that show how (a) the typical spatial discretizations of forecast experiments ( 2 km) hamper the ability of models to capture highly localized secondary triggering patterns and (b) differences in earthquake parameters (i.e., magnitude and hypocenters) reported in different catalogs can affect forecast evaluation. These findings will contribute toward improving forecast model design and evaluation strategies for next-generation seismic catalogs.

Journal of Geophysical Research--Solid Earth↗

USGS2022

Crustal permeability changes observed from seismic attenuation: Impacts on multi-mainshock sequences

We use amplitude ratios from narrowband-filtered earthquake seismograms to measure variations of seismic attenuation over time, providing unique insights into the dynamic state of stress in the Earth’s crust at depth. Our dataset from earthquakes of the 2016-2017 Central Apennines sequence allows us to obtain high-resolution time histories of seismic attenuation (frequency band: 0.5-30 Hz) characterized by strong earthquake dilatation-induced fluctuations at seismogenic depths, caused by the cumulative elastic stress drop after the sequence, as well as damage-induced ones at shallow depths caused by energetic surface waves. Cumulative stress drop causes negative dilatation, reduced permeability, and seismic attenuation, whereas strong-motion surface waves produce an increase in crack density, and so in permeability and seismic attenuation. In the aftermath of the main shocks of the sequence, we show that the M ≥ 3.5 earthquake occurrence vs. time and distance is consistent with fluid diffusion: diffusion signatures are associated with changes in seismic attenuation during the first days of the Amatrice, Visso-Norcia, and Capitignano sub-sequences. We hypothesize that coseismic permeability changes create fluid diffusion pathways that are at least partly responsible for triggering multi-mainshock seismic sequences. Here we show that anelastic seismic attenuation fluctuates coherently with our hypothesis.

Frontiers in Earth Science↗

USGS2021

The weight of cities: Urbanization effects on Earth’s subsurface

Across the world, people increasingly choose to live in cities. By 2050, 70% of Earth's population will live in large urban areas. Upon considering a large city, questions arise such as, how much does that weigh? What are its effects on the landscape? Does it cause measurable subsidence? Here I calculate the weight of San Francisco Bay region urbanization, where 7.75 million people live at, or near the coast. It is difficult to account for everything that is in a city. I assume that most of the weight is buildings and their contents, which allows the use of base outline and height data to approximate their mass, which is cumulatively 1.6·10 12 kg. I build a series of finite element models to study effects of pressure exerted by the weight distribution. Within the elastic realm, I look at compression, flexure, isostatic compensation, stress change, dilatation, and fluid flow changes. Within the nonlinear realm I show example calculations of primary and secondary settlement of soils under load. The combined modeled subsidence from building loads is at least 5–80 mm, with the largest contributions coming from nonlinear settlement and creep in soils. A general result is closing of pore space and redirection of pore fluids. While the calculated subsidence of the Bay Area is relatively small compared with other sources of elevation change such as pumping and recharge of aquifers, all sources of subsidence are concerning given an expected 200–300 mm sea level rise at San Francisco by the year 2050.

AGU Advances↗

USGS2020

Seismic attenuation monitoring of a critically stressed San Andreas fault

We show that seismic attenuation ( ) along the San Andreas fault (SAF) at Parkfield correlates with the occurrence of moderate‐to‐large earthquakes at local and regional distances. Earthquake‐related anomalies are likely caused by changes in permeability from dilatant static stress changes, damage by strong shaking from local sources, and pore unclogging/clogging from mobilization of colloids by dynamic strains. We find that, prior to the 2004 M 6 Parkfield earthquake, prefailure conditions for some local events of moderate magnitude correspond to positive anomalies of on the Pacific side, with local and regional earthquakes producing sharp attenuation reversals. After the 2004 Parkfield earthquake, we see higher anomalies along the SAF, but low sensitivity to local and regional earthquakes, probably because the mainshock significantly altered the permeability state of the rocks adjacent to the SAF, and its sensitivity to earthquake‐induced stress perturbations.

California↗

USGS2020

On the use of receiver operating character tests for evaluating spatial earthquake forecasts

Spatial forecasts of triggered earthquake distributions have been ranked using receiver operating characteristic (ROC) tests. The test is a binary comparison between regions of positive and negative forecast against positive and negative presence of earthquakes. Forecasts predicting only positive changes score higher than Coulomb methods, which predict positive and negative changes. I hypothesize that removing the possibility of failures in negative forecast realms yields better ROC scores. I create a ‘perfect’ Coulomb forecast where all earthquakes only fall into positive stress change areas and compare with an informationless all-positive forecast. The ‘perfect’ Coulomb forecast barely beats the informationless forecast, and adding as few as 4 earthquakes occurring in the negative stress regions causes the Coulomb forecast to be no better than an informationless forecast under a ROC test. ROC tests also suffer from data imbalance when applied to earthquake forecasts because there are many more negative cases than positive.

Geophysical Research Letters↗

USGS2020

Distribution of earthquakes on a branching fault system using integer programming and greedy sequential methods

A new global optimization method is used to determine the distribution of earthquakes on a complex, connected fault system. The method, integer programming, has been advanced in the field of operations research, but has not been widely applied to geophysical problems until recently. In this application, we determine the optimal distribution of earthquakes on mapped faults to minimize the global misfit in slip rates for multi-fault ruptures. Integer programming solves for a decision vector composed of every possible location that a sample of earthquakes can occur on every fault, subject to slip-rate uncertainty constraints. Step over connections are straightforward to include, whereas branching fault connections are not. To include branching ruptures, we distinguish between individual multi-fault rupture paths, as opposed to formulating the integer-programming problem based on individual faults as in previous studies. The new method is applied to the complex fault system in the San Francisco Bay Area as a case study. Results from the integer-programming method are compared to those from a local optimization method, termed the greedy-sequential method. Several experiments using these two methods indicate that shape of the on-fault magnitude distributions and which branching faults are involved in multi-fault ruptures depend on how much emphasis is placed on fitting the target slip rate. In cases where the underlying data are not strong enough to warrant chasing the target slip rate, it is better to focus on the distribution of feasible results that better represents the uncertainty in the solutions imposed by the data.

Geochemistry, Geophysics, Geosystems↗