Search

Seismic source locations are fundamental to many fields of Earth and planetary sciences, such as seismology, volcanology and tectonics. However, seismic source detection and location are challenging when events cluster closely in space and time with signals tangling together at observing stations, such as they often do in major aftershock sequences. Though emerging algorithms and artificial intelligence (AI) models have made processing high volumes of seismic data easier, their performance is still limited, especially for complex aftershock sequences. In this study, we propose a novel approach that utilizes three‐dimensional image segmentation—a computer vision technique—to detect and locate seismic sources, and develop this into a complete workflow, Source Untangler Guided by Artificial intelligence image Recognition (SUGAR). In our synthetic and real data tests, SUGAR can handle complex, energetic earthquake sequences in near real time better than skillful analysts and other AI and non‐AI based algorithms. We apply SUGAR to the 2016 Kaikōura, New Zealand sequence and obtain five times more events than the analyst‐based GeoNet catalog. The improved aftershock distribution illuminates a continuous fault system with extensive fracture zones beneath the segmented, discontinuous surface ruptures. Our method has broader applicability to non‐earthquake sources and other time series image data sets. Plain Language Summary: Detecting and locating earthquakes is fundamental to seismology, volcanology, and tectonics. A number of emerging algorithms, including some based upon artificial intelligence (AI), have made processing large volumes of seismic data much easier. However, their performance is still limited, especially in clustered aftershock sequences whose signals overlap at observing seismographs. We propose a new, AI computer vision‐based approach to this problem, and develop it into a complete earthquake detection and location workflow, named SUGAR. Tests on synthetic and real earthquake data sets show that SUGAR characterizes complex earthquake sequences better than other AI and non‐AI algorithms or professional analysts. We apply SUGAR to the complex aftershock sequence of the 2016 Mw 7.8 Kaikōura, New Zealand earthquake, detecting five times more events than the analyst‐based GeoNet catalog. Whereas surface breaks of the Kaikōura earthquake are highly discontinuous, our improved aftershock distribution supports a continuous fault system surrounded by extensive fracture zones at depth. Our method has broader potential for other types of seismic sources and image series. Key Points: We propose a new seismic source detection and location approach based on the source‐scanning algorithm and 3D image segmentationThis approach outperforms human analysts and popular artificial intelligence (AI) and non‐AI based methods in characterizing intense aftershock sequencesThe resulting catalog of the 2016 Kaikōura earthquake sequence suggests a continuous fault system surrounded by extensive fracturing [ABSTRACT FROM AUTHOR]

The violent eruption of the Hunga (Tonga) submarine volcano on 15 January 2022 caused a 58 km‐heigh ash plume, catastrophic tsunami, and significant global seismic and infrasound waves. However, the physical mechanism underpinning its multiple‐explosive events remains unclear, and its resolvability relies on the seismic waveform source inversion. The studies of two different point‐source models, the seismic moment tensor (MT) and the single force (SF), have been performed separately for this eruption, which, interestingly, can explain the seismic data adequately. Here, we use a joint inversion of MT and SF to unravel a composite source of an explosion‐like MT and a significant upward force for the first major explosive event. Regarding the direction and magnitude, we propose that the upward force is likely a rebound force in response to the pressure drop on the seafloor because the water body above the volcano was abruptly uplifted by the shallow underwater explosion. Plain Language Summary: The physical process of the violent eruption of the Hunga (Tonga) submarine volcano on 15 January 2022 remains unclear. To date, the common source model for volcano eruptions—a single force (SF) and the common source model for earthquakes and explosions—a moment tensor (MT), have been inferred individually for this eruption. Interestingly, both can explain the recorded seismic signals reasonably well. A question arises whether a combination of sources is a better physical model. Therefore, we combine the MT and SF to represent the eruption process in this study. The source analysis for the first major event of this eruption reveals a possible composite process of a shallow underwater explosion and a significant upward force. The upward force is opposite the common downward‐reaction force to the material jetting. It is likely caused by the abrupt displacement of the water above the volcano resulting from the shallow underwater explosion. When the downward water pressure on the seafloor vanishes, the seafloor responds by an upward‐rebound force. Key Points: We perform probabilistic inversion of seismic waveform data to study the force equivalent system of the 2022 Hunga (Tonga) eruptionThe first major explosive event of the shallow eruption sequences may consist of an explosion‐like moment tensor and a large upward forceA possible mechanism of the accompanying upward force is the rebound force responding to the sudden pressure drop of uplifted water body [ABSTRACT FROM AUTHOR]

A minequake of magnitude ML 4.1 occurred on 18 May 2020 early in the morning at the LKAB underground iron ore mine in Kiruna, Sweden. This is the largest mining‐induced earthquake in Scandinavia. It generated acoustic signals observed at three infrasound arrays at 9.3 (KRIS, Sweden), 155 (IS37, Norway), and 286 km (ARCI, Norway) distance. We perform full‐waveform focal mechanism inversion based on regional seismic data and local infrasound data. These independently highlight that this event was dominated by a shallow‐depth collapse in agreement with in‐mine seismic station data. However, regional infrasound data cannot inform the inversion process without an accurate model of atmospheric winds and temperatures. Yet, our numerical simulations demonstrate a potential of using local and regional infrasound data to constrain an event's focal mechanism and depth. Plain Language Summary: The largest mining‐induced earthquake in Scandinavia (ML 4.1) occurred on 18 May 2020 early in the morning at the LKAB underground iron ore mine in Kiruna, Sweden. The seismic waves coupled to the atmosphere and propagated large distances as sound waves which were observed at three infrasound arrays at 9.3 (KRIS, Sweden), 155 (IS37, Norway), and 286 km (ARCI, Norway) distance. Our seismic and acoustic modeling results highlight a strong collapse event within the northern section of the mine. The modeling of acoustic and seismic waves across the Earth‐atmosphere suggests that sound wave data can help when determining the location and properties of a seismic source. Key Points: Seismic and acoustic records indicate a strong collapse at shallow depth of 1 ± 0.5 km for the 2020 Kiruna minequakeFocal mechanisms and depths of shallow seismic sources can be retrieved from local infrasound recordsInversions using regional infrasound data is possible when accurate weather models are available [ABSTRACT FROM AUTHOR]

Genuine non‐double‐couple (non‐DC) components of a seismic source, defined here as the non‐DC components that are not due to summation of pure double‐couple (DC) components, provide important insight into special physical processes in non‐earthquake sources such as explosion, volcano eruption and collapse etc. Yet they remain challenging to be resolved. To address the issue and explore the physical mechanism of those special events, we develop a waveform‐polarity‐based moment tensor (WPMT) inversion method and employ it to study physical process in the 2014–2015 Bárðarbunga volcano event sequence. The WPMT method incorporates P‐wave polarity data and seismic waveforms in the source inversion, designs a source simplicity test to check possible complex rupture in the seismic source, and employs a simulated annealing algorithm to search the best source solution. The simplicity test checks consistency of the source processes in the initiation stage of the event and the major energy release process of the event, thus ensuring that the inferred non‐DC components are genuine to the seismic source. Real event and synthetic tests indicate that the WPMT method can identify and resolve genuine non‐DC components in a seismic source. The WPMT inversions of the Bárðarbunga sequence yield many genuine non‐DC source components and reveal that the eruptions are accompanied by seismic activities in depths of 1–5 km with magma migrations out of chambers, collapses of conduits, failures of normal faults, and a magma recharge at a depth of 9 km accompanied by a failure on a nearby normal fault. Plain Language Summary: Special seismic events such as explosion, collapse and volcanic eruption possess source components that are different from fault solutions used to represent a typical tectonic earthquake of simple shear. The identification and quantification of the source components of those special seismic sources thus play an important role in deciphering the physical process during those events. Yet, it remains challenging to resolve the source components of special seismic events due to the ambiguity of distinguishing them from an apparent one due to complex earthquake rupture. We develop a waveform‐polarity‐based moment‐tensor inversion to resolve the source components of a special seismic source and employ the method to study the physical process in the 2014–2015 Bárðarbunga volcano event sequence in Iceland. Real event and synthetic tests confirm method's capability of identifying and constraining the source components of special non‐earthquake events. The study of the Bárðarbunga sequence reveals that the eruptions are accompanied by seismic activities in depths of 1–5 km with magma migrations out of chambers, collapses of conduits, failures of normal faults, and a magma recharge at a depth of 9 km accompanied by a failure on a nearby normal fault. Key Points: We propose a waveform‐polarity‐based moment tensor inversion method to extract non‐double‐couple components of moderate seismic eventsReal event and synthetic tests show the method is effective in distinguishing and resolving genuine non‐double‐couple componentsMethod application reveals magma activities, conduit collapses and induced earthquakes during the 2014–2015 Bárðarbunga eruptions [ABSTRACT FROM AUTHOR]

We analyze seismic waves excited by small Strombolian explosions to understand the source process of volcanic explosions. We deployed five broadband seismometers at only 100–300 m away from the active craters of Stromboli volcano, Italy. Moment tensor inversion of the entire seismic signals in the 0.05–0.2 Hz band locates the source at a depth of 170 and 150–200 m west/southwest of the crater where acoustic waves are excited. Contrary, the sources of seismic waves in the 0.2–0.5 and 0.5–1.0 Hz bands are excited almost at the explosion onset and are located close to the crater. We show for the first time that explosions are preceded of about 10–20 s by a small amplitude seismic phase. Semblance analysis shows that this phase is radiated from a depth of 170 m beneath the western part of the crater area. Our analysis indicates that the source moves about 50 m toward the active crater 10–20 s before the explosion occurs at the surface. At the explosion onset, the source moves back to the same location of the small preceding phase. These lateral migrations of the seismic source are estimated by moment tensor inversion and semblance analysis. We suggest that migration reflects the bending of the shallow feeding system toward northeast. Seismic waves are thus reflecting the history pressure generated by the rising of a gas‐rich pocket in the very shallow portion of a magma mush and by the following restoring force occurring after the explosion. Plain Language Summary: We analyze seismic waves repeatedly excited before, during and after small explosions at the summit crater of Stromboli volcano, Italy, to understand the magma motions in the shallow conduit and reservoir system. To improve the spatio‐temporal resolution and detect slight movement of magma, we deployed broad‐band seismometers very close to the active craters with a distance range of 100–300 m. As a result, we succeeded in clarifying detailed horizontal spatio‐temporal change of seismic sources for about 1 min before and after each explosion. First seismic source, which is for the first time detected in seismogram as a very small amplitude phase, appears at a depth of about 170 m beneath the western part of the active craters. About 20 s later, the seismic source moves about 50 m toward the craters to generate an explosion. Just after the explosion, the seismic source moves back to the west where the initial small seismic source is located. Such detailed horizontal spatio‐temporal change reflects the migration and/or the resonance of gas‐rich magma batch in the very shallow reservoir. Key Points: Migration of seismic source in a set of explosion at Stromboli is estimated by moment tensor inversion and semblance analysesSource laterally migrates from the west to the crater just before an explosion, and moves back to the initial location during the explosionSmall seismic signal in the 0.05–0.2 Hz frequency band is detected 10–20 s before the explosive onset [ABSTRACT FROM AUTHOR]

The detection of deformation in Global Navigation Satellite System (GNSS) time series associated with (a)seismic events down to a low magnitude is still a challenging issue. The presence of a considerable amount of noise in the data makes it difficult to reveal patterns of small ground deformation. Traditional analyses and methodologies are able to effectively retrieve the deformation associated with medium to large magnitude events. However, the automatic detection and characterization of such events is still a complex task, because traditionally employed methods often separate the time series analysis from the source characterization. Here we propose a first end‐to‐end framework to characterize seismic sources using geodetic data by means of deep learning, which can be an efficient alternative to the traditional workflow, possibly overcoming its performance. We exploit three different geodetic data representations in order to leverage the intrinsic spatio‐temporal structure of the GNSS noise and the target signal associated with (slow) earthquake deformation. We employ time series, images, and image time series to account for the temporal, spatial, and spatio‐temporal domain, respectively. Thereafter, we design and develop a specific deep learning model for each dataset. We analyze the performance of the tested models both on synthetic and real data from North Japan, showing that image time series of geodetic deformation can be an effective data representation to embed the spatio‐temporal evolution, with the associated deep learning method outperforming the other two. Therefore, jointly accounting for the spatial and temporal evolution may be the key to effectively detect and characterize fast or slow earthquakes. Plain Language Summary: The continuous monitoring of ground displacement with Global Navigation Satellite System allowed, at the beginning of the 2000s, the discovery of slow earthquakes—a transient slow slippage of tectonic faults that releases stress without generating seismic waves. Nevertheless, the detection of small events is still a challenge, because they are hidden in the noise. Most of the methods which are traditionally employed are able to extract the deformation down to a certain signal‐to‐noise level. However, one can ask if deep learning can be a more efficient and powerful alternative. To this end, we address the problem by using deep learning, as it stands as a powerful way to automatize and possibly overcome traditional methods. We use and compare three data representations, that is time series, images, and image time series of deformation, which account for the temporal, spatial, and spatio‐temporal variability, respectively. We train our methods on synthetic data, since real datasets are still not enough to be effectively employed with deep learning, and we test on synthetic and real data as well, claiming that image time series and its associated deep learning model may be more effective toward the study of the slow deformation. Key Points: We develop deep learning approaches on synthetics mimicking the spatio‐temporal structure of static deformation and realistic Global Navigation Satellite System (GNSS) noiseWe design three deep learning models and we train and test them against three GNSS data representationsTransformers and image time series of deformation can effectively characterize small deformation patterns associated with the seismic source [ABSTRACT FROM AUTHOR]

A prototype electromagnetic vibrator, referred to here as E‐Vib, was upgraded and developed for broadband hardrock and mineral exploration seismic surveys. We selected the iron oxide mine in Blötberget, central Sweden, for a test site in 2019 for the newly developed E‐Vib because of the availability of earlier seismic datasets (from 2015 to 2016) for verification of its performance for hardrock imaging purposes. The two‐dimensional data acquisition consisted of a fixed geometry with 550 receiver locations spaced at every 5 m, employing both cabled and wireless seismic recorders, along an approximately 2.7 km long profile. The E‐Vib operated at every second receiver station (i.e. 10 m spacing) with a linear sweep of 2–180 Hz and with a peak force of 7 kN. The processing workflow took advantage of the broadband signal generated by the E‐Vib in this challenging hardrock environment with varying ground conditions. The processed seismic section shows a set of reflections associated with the known iron oxide mineralization and a major crosscutting reflection interpreted to be from a fault system likely to be crosscutting the mineralization. The broadband source data acquisition and subsequent processing helped to improve signal quality and resolution in comparison with the earlier workflows and data where a drophammer seismic source was used as the seismic source. These results suggest new possibilities for the E‐Vib source for improved targeting in hardrock geological settings. [ABSTRACT FROM AUTHOR]

We present probabilistic centroid‐moment tensor solutions inferred from the combination of Hamiltonian Monte Carlo sampling and a 3‐D full‐waveform inversion Earth model of the Japanese islands. While the former provides complete posterior probability densities, the latter allows us to exploit waveform data with periods as low as 15 s. For the computation of Green's functions, we employ spectral‐element simulations through the radially anisotropic and visco‐elastic model, leading to substantial improvements of data fit compared to layered models. Focusing on 13 Mw 4.8–Mw 5.3 offshore earthquakes with a significant non‐double‐couple (non‐DC) component, we simultaneously infer the centroid location, time and moment tensor without any a priori constraints on the faulting mechanism. Furthermore, we perform the inversions across several period bands, varying the minimum period between 15 and 50 s. Accounting for 3‐D Earth structure at shorter periods can increase the double‐couple (DC) component of an event, compared to the GCMT solution, by tens of percent. This suggests that non‐DC events in the GCMT catalog may result from unmodeled Earth structure and the related limitation to longer‐period data. We also observe that significant changes in source parameters, and the DC component in particular, may be related to only small waveform changes, thereby accentuating the importance of a reliable Earth model. Posterior probability density distributions become increasingly multimodal for shorter‐period data that provide tighter constraints on source parameters. This implies, in our specific case, that stochastic approaches to the source inversion problem are required for periods below ∼20 s to avoid trapping in local minima. Plain Language Summary: In the majority of global earthquake catalogs, the earthquake solution, that is, centroid location, time and a rupture mechanism, is typically inferred assuming a 1‐D Earth model. However, both earthquake source and Earth structure contribute to seismic recordings, meaning that unaccounted structure might map into and pollute the source solution. In this study we use a 3‐D Earth structure of the Japanese islands to model the waveforms and infer earthquake parameters of 13 small‐to‐moderate magnitude offshore events. We do not put any a priori constraints on the faulting mechanism and let it be determined by the data. We perform stochastic inversions, which provide us with a collection of all plausible models ranked by their respective probability. When a 3‐D Earth structure at shorter periods is taken into account, the earthquake mechanisms, investigated in this study, can be largely explained by a slip on the fault. We also observe that significant changes in source parameters may be related to tiny waveform changes, thereby accentuating the importance of a reliable Earth model. Key Points: Full‐waveform inversion models enable the use of shorter‐period data in earthquake source inversionUsing regional waveforms at 15 s period may reduce non‐double‐couple components by tens of percentProbability densities of source parameters below 20 s are non‐Gaussian, thus demanding stochastic inversion approaches [ABSTRACT FROM AUTHOR]

Both natural and anthropogenic seismic sources generate so‐called ambient seismic waves. One in turn can use ambient seismic waves to estimate these source distributions and study source characteristics, for instance the source mechanism. A commonly used estimation method is called matched field processing (MFP), and the MFP results are inherently smeared by the array geometry. Another approach to estimate ambient seismic sources is to apply full waveform inversion (FWI) to the crosscorrelations of ambient seismic wave recordings. Both methods have pros and cons, but the model resolution and uncertainty in these two methods are important for the interpretation. Unfortunately, this topic has attracted little attention in the past. We propose to estimate both the model resolution matrix and model covariance matrix of the inversion using singular value decomposition. We demonstrate our estimates using two examples, one of which is an actual field array geometry. We quantitatively compare the model resolution of the two methods and discuss the model null space. We demonstrate that FWI has superior resolution with enough independent data and should be used when computational resources permit. Plain Language Summary: Many natural phenomena (such as ocean waves due to a hurricane or glacier tremor due to subglacial water flow) can shake the ground and generate seismic waves. By studying the seismic waves, we can image these natural phenomena in space and monitor the temporal evolution of these phenomena. We focus on two common methods to image these passive sources: matched field processing (MFP) and full waveform inversion (FWI). The resolution and uncertainties of these two methods are important for interpreting imaging results. Thus, we derive and calculate the model resolution for each method. We also study the model trade‐off of full waveform inversion to quantify the uncertainty in the inversion results. We compare the resolution of the two methods using a realistic field seismic array form a past experiment. We show that FWI has better resolution than MFP when all seismic sensors in the array are used, indicating that FWI should be the preferred method for ambient noise source imaging when computational resources permit its use. Key Points: We present the model resolution and point spread function for two ambient seismic source estimation methods, FWI and MFPWe demonstrate that with enough independent data, FWI possesses higher resolution than MFPWe quantify the model covariance of FWI to provide the source strength trade‐offs [ABSTRACT FROM AUTHOR]

Earthquake source parameters are important for understanding earthquake physics and crustal fault properties. However, strong trade‐offs between parameters (e.g., depth and origin time) and a lack of accurate velocity models and near‐field seismic stations could cause large uncertainties of these parameters in seismic catalogs, particularly for shallow events. To further improve the resolution of earthquake source parameters, we use Interferometric Synthetic Aperture Radar (InSAR) images to derive source solutions of 33 moderate‐size (Mw 4.1–6.6) earthquakes that occurred at shallow depths (<20 km) from November 2014 to July 2020 in western China. After evaluating the uncertainties of the InSAR solutions, we systematically compare the location, centroid depth, focal mechanism and magnitude from InSAR models with that from seismic catalogs. We find that all seismic catalogs generally report deeper (4–10 km) hypocenters or centroid depths. The uncertainties of moment tensor solutions are partially related to the percentage of the non‐double‐couple components in the seismic catalogs. The InSAR solutions indicate that considerable seismic moments (i.e., ∼2×1018 ${2\times 10}^{18}$ Nm) were released in the uppermost crust (i.e., <5 km) in a period of ∼6 years, which is not resolvable in the seismic catalogs. The smooth seismic moment distribution along depth indicates a gradual change of the frictional properties from the surface to the middle crust. As most of the studied earthquakes are located on secondary and/or unmapped faults, these findings imply a considerable portion of velocity‐weakening friction in the uppermost crust along immature, secondary fault systems, which should be included in the seismic hazard evaluation. Plain Language Summary: Uncertainties of parameters that describe earthquake location, fault geometry, and fault slip are generally poorly known in the seismological catalogs. Due to these unknown uncertainties, the distribution of earthquakes in the shallow crust (e.g., <10 km) is not well understood, hindering our understanding of its seismic capacity. Here, we study the moderate‐size earthquakes that occurred in western China from November 2014 to July 2020. We use Interferometric Synthetic Aperture Radar (InSAR) data to obtain surface deformations caused by earthquakes in the shallow crust. We find 33 events (Mw 4.1–6.6) out of 310 earthquakes that have clear InSAR signals, which are used to constrain the source parameters of these earthquakes. We estimate the uncertainties of these InSAR solutions and systematically compare them with the solutions in the seismic catalogs. Our analysis shows that all seismic catalogs report deeper source depth and the horizontal mislocations of the Global Centroid Moment Tensor catalog show geographic correlation features. InSAR solutions also show fault geometry and rupture complexity more precisely than the seismic catalogs. Our results indicate considerable earthquake energy released in the uppermost crust that is not resolvable in the seismic catalogs, which raises additional concerns on the seismic hazard possessed by the shallow earthquakes. Key Points: Coseismic deformations and source solutions of 33 shallow moderate‐size earthquakes (Mw 4.1–6.6) derived from Interferometric Synthetic Aperture Radar (InSAR) dataSource parameters from InSAR solutions and seismic catalogs are systematically compared to estimate their uncertaintiesInSAR solutions show considerable moments released in the uppermost crust, revealing the seismogenic capacity of the shallow crust [ABSTRACT FROM AUTHOR]

Back azimuth (BAz) information can be determined from combined measurements of rotations and translations at a single site. Such six degree‐of‐freedom (6‐DoF) measurements are reasonably stable in delivering similar information compared to a small‐scale array of three‐component seismometers. Here we investigate whether a 6‐DoF approach is applicable to tracking seismic sources. While common approaches determining the timing and location of energy sources generating seismic waves rely on the information of P‐waves, here we use S‐waves. We track back azimuths of directly arriving SH‐waves in the two‐dimensional case, P‐converted SV‐waves, direct SV‐ and direct SH‐waves in the three‐dimensional case. For data analysis, we compare a cross‐correlation approach using a grid‐search optimization algorithm with a polarization analysis method. We successfully recover the rupture path and rupture velocity with only one station, under the assumption of an approximately known fault location. Using more than one station, rupture imaging in space and time is possible without a priori assumptions. We discuss the effects of rupture directivity, supershear rupture velocity, source‐receiver geometry, wavefield interference, and noise. We verify our approach with the analysis of moving traffic noise sources using 6‐DoF observations. The collocated classic seismometer and newly built ring laser gyroscope ROMY near Munich, Germany, allow us to record high‐fidelity, broadband 6‐DoF (particle velocity and rotational rate) ground motions. We successfully track vehicles and estimate their speed while traveling along a nearby highway using the estimated BAz as a function of time of a single station observation. Key Points: We demonstrate concepts for seismic source tracking from combined point measurements of translational and rotational motionsWe show that S‐waves can be used for seismic source tracking due to the wavefield separation in rotational componentsWe discuss the effects of directivity, rupture speed, source‐receiver geometry, and present a real data example of tracking traffic sources [ABSTRACT FROM AUTHOR]

We have developed a portable active seismic source (PASS) to monitor CO2 storage reservoirs at a depth of approximately 1 km. Despite its small size, stacking the signals generated by the PASS improves the signal‐to‐noise ratio of the seismometer data far from the source. The smaller size and lower cost of the PASS enables its permanent deployment in many locations to continuously monitor CO2 storage reservoirs. To achieve continuous monitoring, distributed acoustic sensing (DAS) is also a vital technology. Based on DAS, we can continuously record the signal from the PASS in an extensive area, including within boreholes and offshore fields. Here we report application of the PASS for the borehole DAS system. We confirmed the PASS signal propagation to a depth of ∼1 km when we used a PASS with 630N at 50 Hz close to the wellhead and recorded the signal by the borehole fiber optic cable. The ability of the system to propagate the PASS signal to a depth of ∼1 km enables continuous monitoring of most CO2 storage reservoirs with high temporal resolution. Furthermore, deploying multiple PASS systems could improve the spatial resolution of monitoring results. © 2023 Society of Chemical Industry and John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Most seismological analysis methods require knowledge of the geographic location of the stations comprising a seismic network. However, common machine learning tools used in seismology do not account for this spatial information, and so there is an underutilized potential for improving the performance of machine learning models. In this work, we propose a graph neural network (GNN) approach that explicitly incorporates and leverages spatial information for the task of seismic source characterization (specifically, location and magnitude estimation), based on multistation waveform recordings. Even using a modestly‐sized GNN, we achieve model prediction accuracy that outperforms methods that are agnostic to station locations. Moreover, the proposed method is flexible to the number of seismic stations included in the analysis and is invariant to the order in which the stations are arranged, which opens up new applications in the automation of seismological tasks and in earthquake early warning systems. Plain Language Summary: To determine the location and size of earthquakes, seismologists use the geographic locations of the seismic stations that record the ground shaking in their data analysis workflow. By taking the distance between stations and the relative timing of the onset of the shaking, the origin of the seismic waves can be accurately reconstructed. In recent years, machine learning (a subfield of artificial intelligence) has shown great potential to automate seismological tasks, such as earthquake source localization. Most machine learning methods do not take into consideration the geographic locations of the seismic stations, and so the usefulness of these methods could still be improved by providing the locations at which the data were recorded. In this work, we propose a method that accounts for geographic locations of the seismic stations, and we show that this improves the machine learning predictions. Key Points: We propose a deep learning approach for automated earthquake location and magnitude estimation based on graph neural network theoryThis new approach processes multistation waveforms and incorporates station locations explicitlyIncluding station locations improves the accuracy of epicenter estimation compared to models that are location agnostic [ABSTRACT FROM AUTHOR]

Locked segments are widely present in the slip surface of large rock slopes and seismogenic faults of rock underground engineering. Each cracking occasion of the locked segment results in a seismic event. Accurate determination of the seismic source parameters of a locked-segment cracking event is crucial for the reliable evaluation of rock-mass stability associated with slopes and underground openings. The theoretical framework for calculating seismic source parameters in previous studies is mostly based on the stick-slip model, which is not applicable to describing the locked segment's damage process, and research on seismic source parameter estimation of a locked-segment cracking event is insufficient. Hence, based on the principle of energy conversion and distribution during the locked segment's damage process, we proposed an equation for the radiated seismic energy of a locked-segment cracking event. Using this equation, we established a mechanical relationship between the earthquake magnitude and the stress drop or shear strain increment (or maximum coseismic displacement) of a locked-segment cracking event. Typical case studies of rock slope and rock underground engineering showed that the proposed calculation method of seismic source parameters was reliable. In addition, this paper discusses the controversy surrounding the relationship between earthquake magnitude and stress drop. Relevant results lay a firm physical foundation to accurately calculate the seismic source parameters of a locked-segment cracking event and obtain detailed insights into the generation mechanism of the locked-segment cracking event. [ABSTRACT FROM AUTHOR]

Abstract: The variability in obtaining estimates of tsunami inundation and runup on a near‐real‐time tsunami hazard assessment setting is evaluated. To this end, 19 different source models of the Maule Earthquake were considered as if they represented the best available knowledge an early tsunami warning system could consider. Results show that large variability can be observed in both coseismic deformation and tsunami variables such as inundated area and maximum runup. This suggests that using single source model solutions might not be appropriate unless categorical thresholds are used. Nevertheless, the tsunami forecast obtained from aggregating all source models is in good agreement with observed quantities, suggesting that the development of seismic source inversion techniques in a Bayesian framework or generating stochastic finite fault models from a reference inversion solution could be a viable way of dealing with epistemic uncertainties in the framework of nearly‐real‐time tsunami hazard mapping. [ABSTRACT FROM AUTHOR]

Explosives are the most common seismic source in seismic explorations, whose excitation effect is closely associated with their properties, weight, buried depth, and charging constitution. This study analyzes the influence of seismic source parameters on the features of the explosion wavelet. The explosion process of the seismic source in marine carbonate is numerically simulated with the finite element software LS-DYNA. We consider three parameters, including buried depth, charge weight, and decoupling coefficient. To verify the simulation outcomes, we also carry out field observations. The appropriate buried depth of the explosive should be determined according to stratum lithology. As the charge weight increases, low-frequency energy of the source wavelet increases rapidly. The optimum charge weight is 16–18 kg, and the ideal charging structure is in a concentrated, short-column shape. Compared with the buried depth and charge weight, the decoupling coefficient shows more noticeable influence on the excitation effect of the source wavelet, and the optimum coefficient lies between 2.5 and 3. The results of this study may provide reference data for designing explosion parameters in field seismic explorations. [ABSTRACT FROM AUTHOR]

The number of gigaton‐sized iceberg‐calving events occurring annually at Greenland glaciers is increasing, part of a larger trend of accelerating mass loss from the Greenland Ice Sheet. Though visual observation of large calving events is rare, ∼60 glacial earthquakes generated by these calving events are currently recorded each year by regional and global seismic stations. An empirical relationship between iceberg size and MCSF, a summary measure of glacial‐earthquake size, was recently demonstrated by Olsen and Nettles (2019), https://doi.org/10.1029/2019JF005054. However, MCSF is known to be sensitive to choices made in modeling the seismic source. We incorporate constraints on the seismic source from laboratory studies of calving and test multiple source time functions using synthetic and observed glacial‐earthquake waveforms. We find that a simple, fixed time function with a shape informed by laboratory results greatly improves estimates of earthquake size. The average ratio of estimated to true peak force values is 1.03 for experiments using our preferred source model, compared with an average of 0.3 for models used in previous studies. We find that maximum‐force values estimated from waveform modeling depend far less on model choices than does MCSF, and therefore prefer maximum force as a measure of glacial‐earthquake size. Using both synthetic and real data, we confirm a correlation between maximum force and iceberg mass. Our results support the possibility of developing useful scaling relationships between seismic observables and physical parameters controlling glacier calving. Plain Language Summary: The Greenland Ice Sheet is losing ice mass. About half of that ice is lost when large icebergs break off, or calve, from the fronts of glaciers into the ocean. Knowing the sizes of these icebergs would be valuable, but iceberg calving is rarely captured on camera. However, the largest icebergs produce seismic signals when they calve, referred to as glacial earthquakes. We investigate the relationship between the size of an iceberg and the magnitude of the glacial earthquake it produces, building new models to describe the forces that generate a glacial earthquake. Previously, most details of the force evolution during iceberg calving were unknown. We use observations from laboratory experiments conducted using a plastic block in a tank of water, built to mimic the glacier‐ocean setting. We find that incorporating information from these laboratory experiments into our seismic model greatly improves estimates of earthquake size. Using our new models, we confirm a correlation between glacial‐earthquake magnitude and iceberg size, and show that our improved estimates are likely to be more realistic. Our results suggest that using seismic information to estimate iceberg size and related quantities is a promising path forward. Key Points: A physics‐based source model for glacial‐earthquake modeling improves recovery of seismic‐magnitude valuesMaximum force is less sensitive to model choices than MCSF and is preferred for describing glacial‐earthquake sizeA rapid force reversal during iceberg calving is the most important feature to capture in a glacial‐earthquake source model [ABSTRACT FROM AUTHOR]

The sources of ambient seismic noise are extensively studied both to better understand their influence on ambient noise tomography and related techniques, and to infer constraints on their excitation mechanisms. Here we develop a gradient-based inversion method to infer the space-dependent and time-varying source power spectral density of the Earth's hum from cross correlations of continuous seismic data. The precomputation of wavefields using spectral elements allows us to account for both finite-frequency sensitivity and for three-dimensional Earth structure. Although similar methods have been proposed previously, they have not yet been applied to data to the best of our knowledge. We apply this method to image the seasonally varying sources of Earth's hum during North and South Hemisphere winter. The resulting models suggest that hum sources are localized, persistent features that occur at Pacific coasts or shelves and in the North Atlantic during North Hemisphere winter, as well as South Pacific coasts and several distinct locations in the Southern Ocean in South Hemisphere winter. The contribution of pelagic sources from the central North Pacific cannot be constrained. Besides improving the accuracy of noise source locations through the incorporation of finite-frequency effects and 3-D Earth structure, this method may be used in future cross-correlation waveform inversion studies to provide initial source models and source model updates. [ABSTRACT FROM AUTHOR]

Earthquake source properties such as seismic moment and stress drop are routinely estimated from far-field body wave amplitude spectra. Some quantitative but model-dependent relations have been established between seismic spectra and source parameters. However, large variability is seen in the parameter estimates, and it is uncertain how the variability is partitioned among real variability in the source parameters, observational error, and modeling error due to complexity of earthquake behaviors. Earthquake models with dynamic weakening have been found to exhibit two different modes of rupture: expanding-crack and self-healing pulse modes. Four representative models are generated to model the transition from crack-like to pulse-like. Pulse-like rupture leads to development of a second corner frequency, and the intermediate spectral slope is approximately 2 in most cases. The focal-sphere-averaged lower P and S wave corner frequencies are systematically higher for pulse-like models than crack models of comparable rupture velocity. The slip-weighted stress drop Δ σ E exceeds the moment-based stress drop Δ σ M for pulse-like ruptures, with the ratio ranging from about 1.3 to 1.65, while they are equal for the crack-like case. The variations in rupture mode introduce variability of the order of a factor of 2 in standard (i.e., crack model-based) spectral estimates of stress drop. The transition from arresting- to growing-pulse rupture is accompanied by a large (factor of ∼1.6) increase in the radiation ratio. Thus, variations in rupture mode may account for the portion of the scatter in observational spectral estimates of source parameters. [ABSTRACT FROM AUTHOR]

A site experiment is performed herein within a 100 m range using a high-frequency structure activity monitor to explore the impact of different factors on the microseismic source location and analyze the range of influence of the velocity model, number of stations, and array surface on the seismic source location. Moreover, the impact of wave velocity, velocity-free location algorithm, and position of the seismic source on the microseismic location error of mines is discussed by establishing the ideal theoretical model of the wave velocity location and with particle swarm optimization. The impact of the number of stations and tables on the location precision is also explored by using the microseismic signals produced by the artificial seismic source. The results show that, for the location model containing the velocity, the velocity error would greatly affect the location precision, and the velocity-free algorithm receives good location results. The location result is more satisfactory when the seismic source point falls in between array envelope lines. The seismic source location precision is in direct proportion to the number of stations. According to the experiment, within a 100 m range, when the number of stations is over 12, the effect does not significantly grow with the increase of stations; the number of tables affects the location precision; and the multitable location effect is significantly superior to the single-table effect. The research shows that the optimal station density is 0.0192%, and the appropriate sensor layout to form a multitable monitoring network may effectively enhance the microseismic source precision of mines through the selection of a velocity-free location model. On the contrary, the number of stations can be reduced on the premise of the allowable error of the seismic source location, which may effectively reduce the monitoring cost. [ABSTRACT FROM AUTHOR]

Up to now, the possibility to obtain images of seismic source from ionospheric Global Navigation Satellite Systems (GNSS) measurements (seismo‐ionospheric imagery) has only been demonstrated for giant earthquakes with moment magnitude Mw ≥ 9.0. In this work, we discuss difficulties and restrictions of this method, and we apply for the first time the seismo‐ionospheric imagery for smaller earthquakes. The latter is done on the example of the Mw7.4 Sanriku‐oki earthquake of 9 March 2011. Analysis of 1‐Hz data of total electron content (TEC) shows that the first coseismic ionospheric disturbances (CID) occur ~470–480 s after the earthquake as TEC enhancement on the east‐northeast from the epicenter. The location of these first CID arrivals corresponds to the location of the coseismic uplift that is known as the source of tsunamis. Our results confirm that despite several difficulties and limitations, high‐rate ionospheric GNSS data can be used for determining the seismic source parameters for both giant and smaller/moderate earthquakes. In addition to these seismo‐ionospheric applications, we raise several fundamental questions on CID nature and evolution, namely, one of the most challenging queries—can moderate earthquake generate shock‐acoustic waves? Plain Language Summary: Ionosphere is a layer of charged particles of the Earth's atmosphere located at altitudes ~60–800 km. However, despite being high above the Earth's surface, the ionosphere is sensible to numerous near‐ground geophysical events (earthquakes, tsunamis, volcano eruptions, etc). Acoustic and gravity waves emitted by these events propagate upward and generate atmospheric/ionospheric perturbations. Ionospheric disturbances generated by earthquakes are known as coseismic ionospheric disturbances (CID). Recently, it has been suggested that analysis of CID and their first arrivals can provide information on the position and on the structure of seismic fault ruptured in earthquake. This method is known as ionospheric imagery of seismic source. However, so far, this method has been only applied to giant earthquakes (Mw ≥ 9.0). In this work, for the first time, we apply the ionospheric imagery for smaller earthquakes on the example of the M7.4 Sanriku‐oki earthquake that occurred on 9 March 2011 in Japan. Our results show that this method is applicable to smaller earthquakes, and despite some difficulties, it can indicate the position of coseismic uplift ~8 min after the earthquake. The uplift generates CID but also tsunamis. Therefore, our method can be used as independent or complementary one for near–real‐time tsunami warning systems. Key Points: By applying the method of seismo‐ionospheric imagery, we show the location of the seismic source for the Mw7.4 2011 Sanriku‐oki earthquakeWe discuss possibilities, difficulties, and challenges of the method of ionospheric imagery of seismic sourceSimultaneous use of the ray‐tracing technique and GNSS observations can be useful to resolve some difficulties of the ionospheric imagery [ABSTRACT FROM AUTHOR]

ABSTRACT A compact representation is obtained for the separation of a scalar wavefield on a closed surface into parts due to internal and external sources. The formula assumes that the total field and its gradients are known on the surface, as is the exact Green function of the medium. The derivation involves four rather straightforward applications of Green's theorem or the representation theorem, though it is a remarkable result in that waves from either source that traverse the boundary many times are appropriately separated. The intermediate results at the four steps of the derivation also shed light on the possibility of acoustic shielding from unwanted sources without knowledge of the Green function for the medium. [ABSTRACT FROM AUTHOR]

This paper reports the field setup and preliminary results of experiments utilizing an airgun array in a reservoir in north China for a seismotectonic study. Commonly used in offshore petroleum resource exploration, the airgun source was found to be more useful than a traditional explosive source for large-scale and long offset land seismic surveys. The airgun array, formed by four 1,500 in3 airguns (a total of 6,000 in3 in volume) was placed at a depth of 6–9 m into the reservoir to generate the pressure impulse. No direct evidence was found that the airgun source adversely affected the fish in the reservoir. The peak ground acceleration recorded on the top of the reservoir dam 100 m away was 17.8 gal in the horizontal direction; this is much less than the designed earthquake-resistance threshold of 125 gal for this dam. The energy for one shot of this airgun array is about 6.68 MJ, equivalent to firing a 1.7 kg explosive. The seismic waves generated by the airgun source were recorded by receivers of the regional seismic networks and a temporary wide-angle reflection and refraction profile formed by 100 short-period seismometers with the maximum source-receiver offset of 206 km. The seismic wave signature at these long-offset stations is equivalent to that generated by a traditional blast source in a borehole with a 1,000–2,000 kg explosive. Preliminary results showed clear seismic phases from refractions from the multi-layer crustal structures in the north China region. Forward modelling using numerical simulation confirms that the seismic arrivals are indeed from lower crustal interfaces. The airgun source is efficient, economical, environmentally friendly and suitable for being used in urbanized areas. It has many advantages over an explosive source for seismotectonic studies such as the high repeatability that is supreme for stacking to improve signal qualities. The disadvantage is that the source is limited to existing lakes or reservoirs, which may restrict experimental geometry. [ABSTRACT FROM AUTHOR]

Multicomponent seismic while-drilling data were analysed in order to define better the different wavefield components and the energy radiation pattern of a tricone drill-bit. A unique cross-hole acquisition scheme from one of ENI-AGIP's special test sites was used. Direct measurements of high quality P- and S-waves allowed the determination of the radial horizontal component (HR) and of the corresponding transverse horizontal component (HT). Subsequently, analysis of the vertical transverse component (T), derived by combining Z and HR, led to the computation of a simplified radiation pattern for the SV component of the drill-bit signal. A large time splitting between horizontally and vertically polarized shear waves, influenced by the acquisition geometry and by the geological setting, was analysed. An unconventional SV velocity tomographic reconstruction from the drill-bit signal is also derived and discussed. [ABSTRACT FROM AUTHOR]

On New Year's Day 2024, a magnitude 7.6 event struck the Noto Peninsula in central Japan. Prior to this event, an intense earthquake swarm had persisted beneath the northeastern peninsula for more than five years. Geophysical evidence provides insight into the upwelling of deep fluids from the uppermost mantle that triggers the seismic swarm activity. The noble gases and their isotopes have been used as geochemical indicators to determine the origin of the fluids associated with the swarms and their upwelling. Gas samples collected from boreholes around the seismic source region are characterized by anomalously high 3He/4He ratios (∼3.9 RAcor), indicating infiltration of mantle fluids from the subcrustal lithosphere. Using a steady‐state advection model, we calculated mantle helium fluxes of 1.1–2.4 × 10−15 mol cm−2 a−1, similar to those estimated for other representative fault zones, such as the San Andreas and North Anatolian faults. Plain Language Summary: In the Noto Peninsula, central Japan, a long‐lasting, intense earthquake swarm has continued since May 2018. On 1 January 2024, the largest M 7.6 earthquake to date occurred. Geophysical evidence provides insight into the upwelling of deep fluids from the uppermost mantle that triggers the seismic swarm activity. The origin of fluids associated with earthquake swarms and their upward fluxes were determined using noble gases and their isotopes. We collected gas samples from hot spring wells around the seismic source region of earthquake swarms. The observed 3He/4He ratios several times higher than the atmospheric value. This indicates infiltration of mantle fluids derived from the subcrustal lithosphere. In addition, we calculated mantle helium fluxes of ∼2.4 × 10−15 mol cm−2 a−1, which is consistent with estimates for other major fault zones around the world. Key Points: Before the M 7.6 earthquake that occurred in the Noto Peninsula on 1 January, there was a persistent earthquake swarm for over five yearsElevated 3He/4He ratios in gas samples from hot spring wells suggest the emission of mantle‐derived helium from the seismic source regionThe mantle fluids infiltrate the seismogenic crust and then diffuse in the fault zone, triggering the earthquake swarm [ABSTRACT FROM AUTHOR]

In this paper, a new geometric inversion method is proposed for obtaining the elastic parameters of an anisotropic focal regions More precisely, the eigenvalues of a vertical transversely isotropic elasticity tensor of a focal region are inverted up to a constant for a given only one moment tensor with the accuracy depending on the strength of anisotropy. The reason for using only one moment tensor is that although there might occur several earthquakes in the same focal region, the orientation of the sources is similar to each other. Hence, one cannot obtain independent equations from each earthquake in order to use them in the inversion of elastic parameters of the focal region. Moreover, this method can be applied for real‐time inversion once the moment tensor of an earthquake is evaluated. The inversion method relies on the geometric fact that a moment tensor can be written as a linear combination of the eigenvectors of the anisotropic focal region's elasticity tensor. Then, in the inversion, we use the fact that each coefficient of this unique decomposition is proportional to the eigenvalues of the focal region's elasticity tensor. Two approximations are used in this inversion method, in particular for the potency and the source orientations. The strength of anisotropy of the focal region determines how accurate these approximations are, and, hence, it also determines the resolutions of the inverted eigenvalues. Because of the anisotropy of the focal region, the errors in the inversion do depend on orientations of the dip and rake angles but not on the strike angle since the focal region is vertically transversely isotropic. The accuracy of the inversion for the five parameters of vertical transversely isotropic is shown for every 10°‐gridded orientation of the dip and rake angles on the stereographic projection. The results are very promising along some orientations as shown in the figures. The last section of the paper deals with the inversion of eigenvalues, up to a constant, for a given set of moment tensors; not by using only one moment tensor. It turns out that the best fit corresponds to the average of inversions obtained for different orientations of the sources. [ABSTRACT FROM AUTHOR]

Body waves traversing the Earth's interior from a seismic source to receivers on the surface carry rich information about its internal structures. Their travel time measurements have been widely used in seismology to constrain Earth's interior at the global scale by mapping the time anomaly along their ray paths. However, picking the travel time of global seismic waves, suitable for studying Earth's fine‐scale structures, requires highly skilled personnel and is often fairly subjective. Here, we report the development of an automatic picker for PKIKP waves traveling through the inner core (IC), especially nearly along Earth's diameters, based on the latest advances in supervised deep learning. A convolutional neural network (CNN) we develop automatically determines the PKIKP onset on vertical seismograms near its theoretical prediction of cataloged earthquakes. As high‐quality manual onset picks of global seismic phases are limited, we employ a scheme to generate a synthetic supervised training data set containing 300,000 waveforms. The PKIKP onsets picked by our trained CNN automatic picker exhibit a mean absolute error of ∼0.5 s compared to 1,503 manual picks, comparable to the estimated human‐picking error. In an integration test, the automatic picks obtained from an extended waveform data set yield a cylindrically anisotropic IC model that agrees well with the models inferred from manual picks, which illustrates the success of this pilot model. This is a significant step closer to harvesting an unprecedented volume of travel time measurements for studying the IC or other regions of the Earth's deep interior. Plain Language Summary: Seismic body waves traversing the Earth's interior provide critical constraints on structures and dynamics of the Earth's deep interior, including the solid inner core (IC). The onset time of compressional waves from a seismic source passing through the IC, known as PKIKP waves, has been widely used to study the Earth's deepest shell. However, the collection of onset time picks meticulously analyzed by experienced analysts is scarce due to the labor‐intensive nature of the process. At the same time, large collections compiled by multiple data centers, such as the International Seismological Center's PKIKP onset data set, are inhomogeneous and deemed less reliable for inner‐core research than the data collected by individual researchers. Here, we develop an automatic PKIKP onset picker based on recent advancements in machine learning in computer vision, the convolutional neural network. We train the network with synthetic waveforms, mimicking the influence of the Earth's structure on the initial waveform shape. Our comprehensive tests benchmark the consistency between the automatically picked and the researcher‐examined data sets. The automatic picker allows for the further exploration of extensive seismic archives, enabling the study of the Earth's deep interior in greater detail with unprecedentedly large data sets. Key Points: We employed a convolutional neural network to automatically pick the onsets of inner‐core sensitive PKIKP wavesOur automatic picker approaches near human‐level precision and initially reproduces established inner‐core anisotropic modelsAutomatic phase onset measurements would open unprecedented avenues for studying the Earth's deep interior, including the inner core [ABSTRACT FROM AUTHOR]

When the seismic source is shallow or the wind turbine structure is moderately distant from the epicenter, it is necessary to consider the oblique incidence of planar body waves. To investigate the dynamic response law of wind turbine structures under oblique seismic wave incidence, this study establishes an integrated approach for analyzing the dynamic response of wind turbine structures considering oblique seismic wave incidence, incorporating one‐dimensional time‐domain site response analysis and viscoelastic artificial boundaries. By using the above seismic input method, the dynamic response law of a 5 MW monopile wind turbine under the obliquely incident P and SV waves are analyzed in detail when different characteristics, incident angles, propagation angles and vibration directions of waves are considered. Numerical studies show that the above factors have an important influence on the seismic response of the wind turbine structure. [ABSTRACT FROM AUTHOR]

Tremors are a type of slow earthquake with long‐duration signals compared to ordinary earthquakes. The long signals have been considered to solely reflect their long source process. However, here, we provide evidence suggesting that the source processes of tremors are not always long. We refer to these observations as short‐duration tremors. They were recorded by ocean‐bottom seismometers placed very close to the source. Although these tremors exhibit a short‐duration signal when recorded near the source, they exhibit a typical long‐duration signal elsewhere. Our numerical simulations demonstrate that the features can be captured by considering a strongly scattering medium around their source. One such structure could be small low‐velocity inclusions distributed around the seismic source. The inclusions may represent the seismic expression of geologically detected aquifers in tremor source regions. Furthermore, this medium could be embedded along the slow earthquake fault zone and play a critical role in their source process. Plain Language Summary: At least two types of earthquakes occur in shallow subduction zones: ordinary earthquakes and tremors. Tremors are known to exhibit long signal duration compared to ordinary earthquakes. To date, tremors' long‐duration signal has been solely interpreted by their source process. Here, we discovered tremors that exhibit short duration signals when recorded close from the source which we referred to as "short‐duration tremors". They suggest that tremors' source process is not always long and structural effects may partially form the typical long‐duration signals. We performed numerical simulations on elastic wave propagation and demonstrated that the observations can be qualitatively reproduced by assuming a strongly scattering material surrounding the seismic source. On the other hand, it has been reported based on ocean bottom drilling project that tremor source region may consist of patchily distributed aquifers. The inclusions in our model may be the seismic expressions of the geologically detected aquifers. Further, such a structure could be embedded along the slow‐earthquake fault zone and play a key role in their source process. Key Points: Short‐duration tremors in ocean‐bottom seismometer records suggest that the source process of tremors is not always longShort‐duration tremors can be interpreted by placing a strongly scattering medium around their sourceSuch a medium embedded along the slow earthquake fault zone could play a key role in the tremor source process [ABSTRACT FROM AUTHOR]

Deep‐focus earthquakes (DFEs) originating at the Mantle‐Transition‐Zone (MTZ) (400–700 km) have a Double Couple (DC) radiation pattern similar to crustal earthquakes; however, their mechanism is different and governed by high pressures (15–25 GPa) at nucleation depths. We present a model of nucleation and growth of regions of phase transformation, undergoing a sudden reduction in volume (5%–10%), "volume collapse." Successive symmetry‐breaking instabilities minimize the energy spent to move the boundary of phase discontinuity and a collapsing volume expands as a flattened pancake‐like self‐similarly expanding Eshelby ellipsoidal inclusion. At the vanishing of the M integral, expressing the balance of flows of energy across the inclusion boundary, at a critical value of the pressure, an arbitrarily small inclusion nucleates and grows at constant potential energy driven by the pressure acting on the change in volume. The inclusion develops shear eigenstrains that decompose into two DC, placing one on the basal plane to radiate without energy losses. The symmetric volume collapse radiates out as an anti‐symmetric DC, and the radiated energy is obtained as the "excess energy," of the ambient pressure acting on the "volume collapse," reduced by the energy consumed for the growth of the pancake surfaces, with a "pressure drop" (p0 − pcr) driving the expansion emitting a DC, even under full isotropy. The solution explains some features of the DFEs, (a) the DC radiation, (b) their large energies (the Mw 8.3 Okhotsk earthquake [2013]), (c) the absence of volumetric radiation, and (d) why they can originate in the MTZ, a long‐standing open problem. Plain Language Summary: Seismic activity is highest at the Earth's surface (30 km) and declines sharply at greater depths until it peaks again in the Mantle‐Transition‐Zone (MTZ, 400–700 km). Deep‐focus earthquakes (DFEs) that originate in the MTZ are registered at above 7.5 on the Richter scale and are similar in radiation pattern to surface ones. It has remained a mystery how the surface and the deep earthquakes both present a Double‐Couple radiation, although their mechanisms are expected to be different, due to vastly differing pressures and material properties. Seismological observations and analog experiments indicate that, under high pressure, the geological material of the MTZ undergoes a phase transition to a denser form (5%–10% "volume collapse"). A theoretical model is presented here proving that the immense pressure (15–25 GPa) acting on the volume collapse nucleates and drives the DFE as a Double‐Couple shear seismic source, minimizing the energy losses by breaking the symmetries, with the symmetric volume collapse radiating as an anti‐symmetric seismic source. The model explains some features of DFEs, such as (a) Double‐Couple radiation, (b) why DFEs originate at the pressures and geological material in the MTZ, and (c) their large energies, (e.g., the Mw 8.3 Okhotsk earthquake, 2013), thus unraveling the long‐standing puzzle. Key Points: In a dynamic Eshelby inclusion model a collapsing volume minimizes losses by breaking symmetries and radiating as a Double CoupleThe radiated energy is the "pressure drop" acting on the volume collapse, the energy input minus the one spent for nucleation and growthThe model can explain the nucleation of deep earthquakes in the Mantle‐Transition‐Zone (400–700 km) and their large energies (e.g., Mw 8.3 Okhotsk, 2013) [ABSTRACT FROM AUTHOR]

In geophysical inverse problems, the distribution of physical properties in an Earth model is inferred from a set of measured data. A necessary step is to select data that are best suited to the problem at hand. This step is performed ahead of solving the inverse problem, generally on the basis of expert knowledge. However, expert‐opinion can introduce bias based on pre‐conceptions. Here we apply a trans‐dimensional algorithm to automatically weigh data on the basis of how consistent they are with the fundamental hypotheses made to solve the inverse problem. We demonstrate this approach by inverting arrival times for the location of a seismic source in an elastic half‐space, assuming a point‐source and uniform weights in concentric shells. The key advantage is that the data do no longer need to be selected by an expert, but they are assigned varying weights during the inversion procedure. Plain Language Summary: In the Big data era, automated approaches to data evaluation are needed for two main reasons: to be able to process a large amount of data in a limited time, and to avoid bias introduced by data analysists. In this study we present a novel approach to data analysis, where the data themselves measure their consistency with our hypotheses. The approach is applied to earthquake location in mines, where millions of seismic events occur every year, and automatic processing of seismic data is mandatory. We demonstrate that our approach outperforms standard ones when almost nothing is known about the data and their measurement errors. Key Points: We develop a novel approach for automatic weighting of data in geophysical inverse problems, based on a trans‐dimensional algorithmWe apply the novel approach to seismic event location in mines, obtaining consistent results compared to a more standard methodOur approach outperforms standard seismic monitoring approaches, when limited information are available on local seismic structure [ABSTRACT FROM AUTHOR]

bstract [ABSTRACT FROM AUTHOR]