Austin Elliott

Accurately characterizing 3D fault geometry is vital for improving our understanding of earthquake behavior and informing the development of seismic hazard models. Despite their importance, subsurface fault structures tend to be poorly constrained because of limitations in observational data. Improvements to the seismic networks and earthquake detection algorithms have increased the precision and volume of earthquake catalogs, which help illuminate detailed subsurface fault structure and provide the most direct information available about fault geometries at depth. We present a Python package to automate generating 3D fault geometries directly from hypocentral seismicity patterns. This method begins with clustering events based on their spatial density, identifying coherent patterns. Nearby clusters are then merged based on the similarity of their orientations. We fit nonplanar surfaces using support vector regression to balance surface accuracy with minimal deviations from planarity. The fault models are output as quadrilateral meshes at user-defined resolution. In the process of generating the 3D fault surfaces, we compute the spatial density of seismicity around the surface and the planarity as quantitative metrics of the model outputs. As a proof of concept, we apply this approach to the San Andreas–Calaveras fault junction region and the 2019 Ridgecrest earthquake sequence, both in California, which contain complex subparallel faults well defined at the Earth’s surface and abundant microseismicity. These case studies demonstrate the method’s ability to model complex fault structures, including long continuous fault surfaces, crossing faults, variably dipping segments, and subparallel faults. We test the method on both standard network catalogs and double-difference relocated catalogs. We find that our seismicity-based fault model results align with published 3D models that incorporate additional constraints and interpretations (Plesch et al., 2020; Aagaard and Hirakawa, 2021). This workflow provides a low-user-input solution for estimating fault geometries at depth from earthquake catalogs.

The Concord fault constitutes a major branch of the Pacific–North America transform plate boundary in Northern California, bridging the strike-slip Bartlett Springs - Green Valley Fault system to the north with the Greenville and Calaveras Faults to the south. Like many faults in the San Francisco Bay Area its long-term slip is partially accommodated by aseismic slip (creep). Although creep has been recognized and monitored on the northern half of the fault for decades, the precise location of the southern half of the fault and its slip rate—whether accommodated seismically or aseismically—has remained enigmatic. How slip transfers between the Concord and Greenville or Calaveras faults to the south remains an outstanding question. New field observations presented here indicate that the active trace of the fault south of downtown Concord is not where previously interpreted and is indeed actively creeping. We report observations of shallow creep continuing >7 km farther south along the Concord fault than previously reported, along a fault strand not previously recognized for most of its length. This is evident as right-laterally deflected concrete curbs and sidewalk slabs on both sides of every street that crosses the fault at a high angle in southeast Concord and northeast Walnut Creek. We document the magnitude and location of these deflections to estimate accumulated right-lateral aseismic slip expressed in engineered structures. Offsets of these piercing lines range from 8 to 18 cm, over widths varying from narrow breaks along centimeter-scale concrete joints to 10-m-wide zones of deflection. Significantly, this active trace is ∼400 m west of where the Quaternary active trace has previously been inferred, placing it within—rather than bounding—the built area of suburban Concord. Slip along the fault has already caused infrastructure damage. These results revise our understanding of the southern Concord fault and help constrain its seismic potential.

This book presents overviews of modern, twenty-first century methods that are used to investigate the physical characteristics of past earthquakes. Where did earthquakes occur? How large were they? Which tectonic faults caused them? How strong were the ground motions and where?

The study of surface ruptures is key to understanding the earthquake occurrence of faults especially in the absence of historical events. We present a detailed analysis of geomorphic displacements along the Dzhungarian Fault, which straddles the border of China and Kazakhstan. We use digital elevation models derived from structure‐from‐motion analysis of Pléiades satellite imagery and drone imagery from specific field sites to measure surface offsets. We provide direct age constraints from alluvial terraces displaced by faulting and indirect dating from morphological analysis of the scarps. We find that the southern 250 km of the fault likely ruptured in a single event in the last 4,000 years, with displacements of 10–15 m, and potentially up to 20 m at one site. We infer that this Dzhungarian rupture is likely linked with a previously identified paleo‐earthquake rupture on the Lepsy Fault through a system of splays in the intervening highlands. Though there are remaining uncertainties regarding consistency in age constraints between the two fault ruptures, most of the sites along the two faults are consistent with a most recent event 2,000–4,000 years ago. Rupture on the Dzhungarian Fault alone is likely to have exceeded M_w 8, and the combined Lepsy‐Dzhungarian rupture scenario may have been up to M_w 8.4. Despite being at the upper end of known or inferred continental earthquake magnitudes, our proposed scenario combining the 375 km of the Dzhungarian and Lepsy ruptures yields a slip‐to‐length ratio consistent with global averages and so do other historical intracontinental earthquakes in Central Asia.

Reported magnitudes ranging between 7.8 and 8.7 highlight a confusion about the true size of the 1920 Haiyuan earthquake, the largest earthquake recorded in the northeast Tibetan Plateau. We compiled a global data set of previously unlooked‐at historical seismograms and performed modern computational analyses on the digitized seismic records. We found the epicenter to be near Haiyuan town and obtained a moment magnitude of M_W=7.9±0.2. Following traditional approaches, we obtained m_B=7.9±0.3 with data from 21 stations and M_S(20)=8.1±0.2 with data from three stations. Geomorphologically, we mapped the surface rupture and horizontal offsets on high‐resolution Pleiades satellite and drone imagery that covered the entire western and middle sections of the 1920 Haiyuan earthquake rupture and compiled offsets reported on the eastern section from field measurements in the 1980s. Careful discrimination between single‐event and cumulative offsets suggests average horizontal slips of 3.0±1.0 m on the western section, 4.5±1.5 m on the middle section, and 3.5±0.5 m on the eastern section, indicating a total moment magnitude of M_W=7.8±0.1. Thus, the seismological and geomorphological results agree within the uncertainties, a weighted average giving a moment magnitude of M_W=7.9±0.2 for the 1920 Haiyuan earthquake. It is likely that earthquake magnitudes based on the historical M were systematically overestimated.

SUMMARY: On 7 December 2015, a shallow Mw 7.2 strike-slip earthquake struck the Murghab River Valley in the Central Pamirs of Tajikistan. Seismologically this event was similar to a large seismic event in 1911 whose causative fault has never been identified. We measure the displacement field of the 2015 event from satellite observations using Sentinel-1 radar interferometry, Landsat-8 optical pixel-tracking, and surface rupture mapping from high resolution SPOT-6/7 imagery to characterize the role this earthquake rupture plays in the accommodation of strain on its causative structure, the Sarez-Karakul fault. We present geomorphic mapping and interpretations of other Quaternary-active reaches of this fault system, which highlight variable rupture history of the different sections. These sections appear to be separated by inherited bedrock structural boundaries. Significantly, the reaches of the fault northeast and southwest of the 2015 rupture exhibit the freshest morphology prior to 2015, indicative of a more recent rupture than elsewhere. Using new high resolution imagery we map fresh scarps at the northern and southern ends of the Sarez-Karakul fault which may represent this 1911 rupture. To test which of these reaches could have been the source of the elusive 1911 event, we compare synthetic seismograms from three plausible fault sources determined from geomorphology, with observed seismic traces from 1911 at early recording stations throughout Europe. We find that the best fitting fault source is in fact southwest of the 2015 rupture, meaning that we have a record of three distinct recent events on the Sarez-Karakul fault system—two of them instrumentally recorded. Our mapping of these separate events reveals a correlation between their boundaries and the active and inherited thrust and suture systems that intersect the northeast striking left-lateral fault, suggesting structural control over the extents of individual ruptures on the active strike-slip fault.

Large pre-historical earthquakes leave traces in the geological and geomorphological record, such as primary and secondary surface ruptures and mass movements, which are the only means to estimate their magnitudes. These environmental earthquake effects (EEEs) can be calibrated using recent seismic events and the Environmental Seismic Intensity Scale (ESI2007). We apply the ESI2007 scale to the 1992 M_S7.3 Suusamyr Earthquake in the Kyrgyz Tien Shan, because similar studies are sparse in that area and geological setting, and because this earthquake was very peculiar in its primary surface rupture pattern. We analyze literature data on primary and secondary earthquake effects and add our own observations from fieldwork. We show that the ESI2007 distribution differs somewhat from traditional intensity assessments (MSK (Medvedev-Sponheuer-Karnik) and MM (Modified Mercalli)), because of the sparse population in the epicentral area and the spatial distribution of primary and secondary EEEs. However, the ESI2007 scale captures a similar overall pattern of the intensity distribution. We then explore how uncertainties in the identification of primary surface ruptures influence the results of the ESI2007 assignment. Our results highlight the applicability of the ESI2007 scale, even in earthquakes with complex and unusual primary surface rupture patterns.

We study rupture behavior of the Aksay bend along the Altyn Tagh fault in northwest China over multiple earthquake cycles. A finite element method is used to numerically simulate spontaneous rupture during a coseismic process, and a viscoelastic model is used to analytically compute the fault stresses during an interseismic process. We find that the Aksay bend is an effective barrier to halt dynamically propagating ruptures from either side of the bend within a range of model parameters, with statistically only about 10% of ruptures jumping across the bend and propagating through almost the entire local fault system. Secondary complexities in fault geometry within the bend, in particular those portions that align relatively well with the regional strike of the fault system, play a critical role in these occasionally jumping ruptures. Well‐developed fault patches with shear stress close to shear strength allow dynamically propagating ruptures to penetrate into the bend and are more susceptible to the dynamic triggering that enables rupture to jump across the bend onto the other strand. We identify additionally nine large rupture scenarios with different occurrences, and most of them rupture one strand outside the bend with triggered slip on some portions of the same or the other strand within the bend. Slip rate distributions from the models show significantly reduced fault slip within the bend and a permanently locked portion on the south strand near the peak of the Altun Mountains. These findings have important implications for seismic hazard assessments of complex fault systems worldwide.

Along-strike variation in scarp morphology reflects differences in a fault's geomorphic and structural development and can thus indicate fault rupture history and mechanical segmentation. Parameters that define scarp morphology (height, width, slope) are typically measured or calculated manually. The time-consuming manual approach reduces the density and objectivity of measurements and can lead to oversight of small-scale morphological variations that occur at a resolution impractical to capture. Furthermore, inconsistencies in the manual approach may also lead to unknown discrepancies and uncertainties between, and also within, individual fault scarp studies. Here, we aim to improve the efficiency, transparency and uniformity of calculating scarp morphological parameters by developing a semi-automated Scarp PARameTer Algorithm (SPARTA). We compare our findings against a traditional, manual analysis and assess the performance of the algorithm using a range of digital elevation model (DEM) resolutions. We then apply our new algorithm to a 12 m resolution TanDEM-X DEM for four southern Malawi fault scarps, located at the southern end of the East African Rift system: the Bilila–Mtakataka fault (BMF) and three previously unreported scarps – Thyolo, Muona and Malombe. All but Muona exhibit first-order structural segmentation at their surface. By using a 5 m resolution DEM derived from high-resolution (50 cm pixel⁻¹) Pleiades stereo-satellite imagery for the Bilila–Mtakataka fault scarp, we quantify secondary structural segmentation. Our scarp height calculations from all four fault scarps suggest that if each scarp was formed by a single, complete rupture, the slip–length ratio for each earthquake exceeds the maximum typical value observed in historical normal faulting earthquakes around the world. The high slip–length ratios therefore imply that the Malawi fault scarps likely formed in multiple earthquakes. The scarp height distribution implies the structural segments of both the BMF and Thyolo fault have merged via rupture of discrete faults (hard links) through several earthquake cycles, and the segments of the Malombe fault have connected via distributed deformation zones (soft links). For all faults studied here, the length of earthquake ruptures may therefore exceed the length of each segment. Thus, our findings shed new light on the seismic hazard in southern Malawi, indicating evidence for a number of large (M_w 7–8) prehistoric earthquakes, as well as providing a new semi-automated methodology (SPARTA) for calculating scarp morphological parameters, which can be used on other fault scarps to infer structural development.