Search

Research developments during 1996 in earthquake seismology, tectonics and structural geology, volcanology, sequence stratigraphy, and studies on core and mantle to elucidate the inner structure of earth are reported. The studies on the slow earthquakes which are considered to precede bigger earthquakes are detailed. The discovery of the post-collisional extension collapse of the Taiwan mountain is also discussed. The studies indicate differential rotation rates between the inner core and the mantle.

Analysis of the unusually diverse right-lateral, left-lateral, reverse, and normal faulting aftershocks of the Loma Prieta earthquake reveals a systematic relation between fault-plane orientation and slip direction; right-lateral aftershock planes strike slightly more northward (10 degrees-15 degrees) than the main shock, left-lateral planes strike slightly more westward, reverse planes dip slightly less steeply, and normal planes dip slightly more steeply than the main shock. We demonstrate that these slip patterns are consistent with an approximately uniaxial stress field acting nearly perpendicular to the main-shock fault plane. These results imply an extremely weak fault zone, possibly resulting from near-lithostatic pore pressure.

The article presents a study that ambient noise Love wave attenuation tomography for the LASSIE array across the Los Angeles basin. It mentions that seismic attenuation tomography has the potential to provide important constraints on wave propagation in the basin and to provide supplementary information on structure in the form of the distribution of anelastic properties.

Earthquake location is one of the oldest problems in seismology, yet remains an active research topic. With dense seismic monitoring networks, it is possible to obtain reliable locations for microearthquakes; however, in many cases dense networks are lacking, limiting the location accuracy, or preventing location when there are too few observations. For small events in all settings, recording may be sparse and location may be difficult due to low signal-to-noise ratio. In this work, we introduce a new, distance-geometry-based method to locate seismicity clusters using only one or two seismic stations. A distance geometry problem consists in determining the location of sets of points based only on the distances between member pairs. Applied to seismology, our approach allows earthquake location using the interevent distance between earthquake pairs, which can be estimated using only one or two seismic stations. We first validate the method with synthetic data that resemble common cluster shapes, and then test the method with two seismic sequences in California: the August 2014    M w 6.0 Napa earthquake and the July 2019  M w 6.4 Ridgecrest earthquake sequence. We demonstrate that our approach provides robust and reliable results even for a single station. When using two seismic stations, the results capture the same structures recovered with high-resolution double-difference locations based on multiple stations. The proposed method is particularly useful for poorly monitored areas, where only a limited number of stations are available. [ABSTRACT FROM AUTHOR]

The important task of tracking seismic activity requires both sensitive detection and accurate earthquake location. Approximate earthquake locations can be estimated promptly and automatically; however, accurate locations depend on precise seismic phase picking, which is a laborious and time‐consuming task. We adapted a deep neural network (DNN) phase picker trained on local seismic data to mesoscale hydraulic fracturing experiments. We designed a novel workflow, transfer learning‐aided double‐difference tomography, to overcome the 3 orders of magnitude difference in both spatial and temporal scales between our data and data used to train the original DNN. Only 3,500 seismograms (0.45% of the original DNN data) were needed to retrain the original DNN model successfully. The phase picks obtained with transfer‐learned model are at least as accurate as the analyst's and lead to improved event locations. Moreover, the effort required for picking once the DNN is trained is a small fraction of the analyst's. Plain Language Summary: Seismic sensors are widely used to monitor many energy‐related systems. To monitor these systems effectively, we need to process a very large amount of data, which is very labor intensive. A few deep learning models have been developed to perform these tasks for earthquake‐generated signals. We adopted one of these deep learning models developed for kilometer scale and updated it for signals recorded from a meter‐scale project. This process not only allows us to overcome the significant spatial and temporal scale difference between our data and the data used by the original deep learning model but also significantly reduces the amount of required training data. Our results show that the updated model matches human performance but with a much faster speed. A workflow that combines the deep learning algorithm with existing imaging technologies enables improvements for both monitoring small earthquakes and studying subsurface structure. Key Points: We successfully applied transfer learning to a neural network for data with orders of difference in spatial and temporal characteristicsWe created a novel workflow by combining deep learning and double‐difference seismic imagingNew workflow provides better seismic catalog and larger amount of phase picks compared to human analysts [ABSTRACT FROM AUTHOR]

We calibrate the local magnitude (ML) scale in southern Kansas, a region of increased seismicity due to oil and gas activities, using both an empirical and a synthetic approach to derive the attenuation curve. In the empirical approach, we use the classic Richter (1935),logA0 attenuation model to calculate ML using amplitude data from the southern Kansas Network catalog and calibrate it using moment magnitudes (Mw) from St. Louis University moment tensor solutions and Trugman et al. (2017). In the synthetic approach, we utilize a crustal velocity model to generate synthetic seismograms from which we measure amplitudes to estimate the attenuation curve. This second approach presents a novel way to calibrate ML, in any region of interest, when earthquake data are scarce or unavailable. Both approaches show lower attenuation in southern Kansas than in the relations being used by the U.S. Geological Survey and Oklahoma Geological Survey to compute ML values in this area. This difference results in a systematic decrease of ~0.1 magnitude unit between our results and those reported in the southern Kansas Network catalog. We also find a dependence of ML on stress drop for earthquakes with corner frequencies near and below the Wood-Anderson instrumental corner of 1.25 Hz. The derived attenuation curve is consistent with a mean stress drop of 3-4 MPa for these earthquakes. [ABSTRACT FROM AUTHOR]

The occurrence of both right and left lateral senses of slip on planes subparallel to the Loma Prieta rupture can be explained by a uniform stress field which has an axis of maximum compression nearly normal to the main-shock rupture zone. The derived uniform stress field provides an effective representation of the postseismic stress field at Loma Prieta. A near-frictionless faulting on preexisting planes of weakness subparallel to the main rupture could also explain the apparent contradictory senses of slip.

We use broad-band stations of the 'Los Angeles Syncline Seismic Interferometry Experiment' (LASSIE) to perform a joint inversion of the Horizontal to Vertical spectral ratios (H/V) and multimode dispersion curves (phase and group velocity) for both Rayleigh and Love waves at each station of a dense line of sensors. The H/V of the autocorrelated signal at a seismic station is proportional to the ratio of the imaginary parts of the Green's function. The presence of low-frequency peaks (∼0.2 Hz) in H/V allows us to constrain the structure of the basin with high confidence to a depth of 6 km. The velocity models we obtain are broadly consistent with the SCEC CVM-H community model and agree well with known geological features. Because our approach differs substantially from previous modelling of crustal velocities in southern California, this research validates both the utility of the diffuse field H/V measurements for deep structural characterization and the predictive value of the CVM-H community velocity model in the Los Angeles region. We also analyse a lower frequency peak (∼0.03 Hz) in H/V and suggest it could be the signature of the Moho. Finally, we show that the independent comparison of the H and V components with their corresponding theoretical counterparts gives information about the degree of diffusivity of the ambient seismic field. [ABSTRACT FROM AUTHOR]

Stress drop has important predictive value for ground motion intensity when determined reliably. Published measurements of stress drop, however, are strongly variable, even for well‐studied earthquake populations. Much of that variability arises from uncertainty in the corner frequency. In this study, we use a spectral ratio method to estimate stress drop and focus on assessing the fit to the data to understand trade‐offs among the parameters of the spectral model. We consider empirical Green's function events located within a 2‐km radius for events in the magnitude range 4.5 ≥ MW ≥ 4 near Cushing, Oklahoma, using an accurately relocated earthquake catalog from May 2013 to November 2016. We form composite S coda wave spectral ratios between mainshock–eGf pairs of events. We fit the Brune spectral model to the data over the frequency range from 0.5 to 30 Hz. To examine the characteristics of the spectral ratios, we evaluate the trade‐off among three estimated parameters—two corner frequencies and the moment ratio—from 1,000 bootstrap resamples of the data. We discard event pairs with a large trade‐off between any two parameters and assess the distribution of the residuals to test the suitability of candidate eGf events. Through these processes, we reduce the standard deviation of earthquake stress drop estimates. The stress drop estimates after the selection are similar to those of natural earthquakes in the central USA. We conclude that careful eGf event selection has the potential to increase substantially the accuracy of stress drop estimates derived from spectral ratios. Key Points: We introduce three evaluation criteria for more reliable estimation of stress drop by the spectral ratio methodTrade‐off among parameters, distribution of the residual, and the trend of the residual are effective criteria for selecting appropriate eGf eventsWe obtain stable stress drop estimates of induced earthquakes in Oklahoma and find that they are comparable to those of tectonic earthquakes [ABSTRACT FROM AUTHOR]

Foreshocks provide valuable information on the nucleation process of an upcoming large earthquake. We applied high‐resolution similar‐waveform techniques for earthquake detection, location, and source parameter estimation to understand the space‐time evolution of a foreshock sequence and its relationship to the mainshock hypocenter. The 1999 Mw 7.1 Hector Mine, California, earthquake was preceded by 50 foreshocks (−0.4 ≤ M ≤ 3.7) during the 20 hr before the mainshock. Foreshock activity did not accelerate leading up to the mainshock. Their locations moved north with time, rupturing adjacent areas along the fault plane with little overlap, but remained within a compact <2 km3 volume. The mainshock initiated at a location where previous foreshocks had locally increased the shear stress. These observations are consistent with a triggered cascade of stress transfer, where previous foreshocks load adjacent fault patches to rupture as additional foreshocks, and eventually the mainshock. Plain Language Summary: Foreshocks are smaller earthquakes that happen near, but before, a large earthquake. We can use foreshocks as a tool to help understand how large earthquakes begin. One model of earthquake initiation holds that small foreshocks directly trigger other small foreshocks by changing the stress on the fault and, eventually, a large earthquake occurs if the conditions are favorable. Since earthquakes of all sizes would start out the same way, it is impossible to predict ahead of time which earthquakes grow large. An alternative explanation holds that smaller foreshocks do not directly trigger the large earthquake but instead are a symptom of something else happening underground, such as slow fault movement that does not release ground‐shaking waves. Several foreshocks occurred in the 20 hr before the 1999 magnitude 7.1 Hector Mine, California, earthquake. Analyzing records of the ground shaking from these foreshocks, we found that they are consistent with the former model, where each foreshock triggered the next one to happen like a line of falling dominoes. We discovered that the large earthquake started out the same way as the smaller foreshocks, so we could not tell ahead of time that it would eventually grow large. Key Points: Foreshocks moved north with time, toward eventual mainshock location; seismicity rate did not accelerate leading up to mainshockForeshocks ruptured adjacent areas along fault with little overlap; mainshock initiated where foreshocks locally increased stressHector Mine foreshocks and mainshock were triggered by a cascade of stress transfer [ABSTRACT FROM AUTHOR]

As the number of seismic sensors grows, it is becoming increasingly difficult for analysts to pick seismic phases manually and comprehensively, yet such efforts are fundamental to earthquake monitoring. Despite years of improvements in automatic phase picking, it is difficult to match the performance of experienced analysts. A more subtle issue is that different seismic analysts may pick phases differently, which can introduce bias into earthquake locations. We present a deep-neural-network-based arrival-time picking method called "PhaseNet" that picks the arrival times of both P and S waves. Deep neural networks have recently made rapid progress in feature learning, and with sufficient training, have achieved super-human performance in many applications. PhaseNet uses three-component seismic waveforms as input and generates probability distributions of P arrivals, S arrivals and noise as output. We engineer PhaseNet such that peaks in the probability distributions provide accurate arrival times for both P and S waves. PhaseNet is trained on the prodigious available data set provided by analyst-labelled P and S arrival times from the Northern California Earthquake Data Center. The data set we use contains more than 700 000 waveform samples extracted from over 30 yr of earthquake recordings. We demonstrate that PhaseNet achieves much higher picking accuracy and recall rate than existing methods when applied to the waveforms of known earthquakes, which has the potential to increase the number of S -wave observations dramatically over what is currently available. This will enable both improved locations and improved shear wave velocity models. [ABSTRACT FROM AUTHOR]