We demonstrate that the statistics of earthquake data in the global Cen- troid Moment Tensor (CMT) and National Earthquake Information Center (NEIC) catalogs and local California Council of the National Seismic System (CNSS) catalog are consistent with the idea that a single physical triggering mechanism is responsible for the occurrence of aftershocks, foreshocks, and multiplets. Specifically, we test the hypothesis that tectonic earthquakes usually show clustering only as a result of an initial earthquake triggering subsequent ones and that the magnitude of each trig- gered earthquake is entirely independent of the magnitude of the triggering earth- quake. Therefore a certain percentage of the time, as determined by the Gutenberg- Richter magnitude-frequency relationship, an earthquake should by chance be larger than or comparable in size to the earthquake that triggered it. This hypothesis predicts that the number of times foreshocks or multiplets are observed should be a fixed fraction of the number of aftershock observations. We find that this is indeed the case in the global CMT and NEIC catalogs; the average ratios between foreshock, aftershock, and multiplet rates are consistent with what would be predicted by the Gutenberg-Richter relationship with b 1. We give special attention to the Solomon Islands, where it has been claimed that unique fault structures lead to unusually high numbers of multiplets. We use Monte Carlo trials to demonstrate that the Solomon Islands multiplets may be explained simply by a high regional aftershock rate and earthquake density. We also verify our foreshock results from the more complete recordings of small earthquakes available in the California catalog and find that foreshock rates for a wide range of foreshock and mainshock magnitudes can be projected from aftershock rates using the Gutenberg-Richter relationship with b 1 and the relationship that the number of earthquakes triggered varies with triggering earthquake magnitude M as c10M, where c is a productivity constant and is equal to 1. Finally, we test an alternative model that proposes that foreshocks do not trigger their mainshocks but are instead triggered by the mainshock nucleation phase. In this model, the nucleation phase varies with mainshock magnitude, so we would expect mainshock magnitude to be correlated with the magnitude, number, or spatial extent of the foreshocks. We find no evidence for any of these correlations.

A Common Origin for Aftershocks, Foreshocks, and Mutliplets

Hydraulic Fracturing-Induced Seismicity

Seismicity in the Raton Basin over the past two decades suggests reactivation of basement faults due to waste‐water injection. In the summer of 2018, 96 short period three‐component nodal instruments were installed in a highly active region of the basin for a month. A machine‐learning based phase picker (PhaseNet) was adopted and identified millions of picks, which were associated into events using an automated algorithm—REAL (Rapid Earthquake Association and Location). After hypocenter relocation with hypoDD, the earthquake catalog contains 9,259 ML −2.2 to 3 earthquakes focused at depths of 4–6 km. Magnitude of completeness (Mc) varies from −1 at nighttime to −0.5 in daytime, likely reflecting noise variation modulated by wind. The clustered hypocenters with variable depths and focal mechanisms suggest a complex network of basement faults. Frequency‐magnitude statistics and the spatiotemporal evolution of seismicity are comparable to tectonic systems. Plain Language Summary Earthquakes induced by waste‐water injection are widely observed worldwide and have been occurring in the Raton Basin (located at the border of New Mexico and Colorado) for two decades. We deployed 96 short period seismic stations in the summer of 2018 to investigate the faults in the southern section of Raton Basin. Earthquake detection was performed with state‐of‐the‐art machine‐learning techniques, which led to a catalog with ~10,000 earthquakes with magnitude ranging from −2.2 to 3. Clusters of earthquakes were investigated in detail. We found that the orientation of the faults varies within the study region and that induced earthquakes may exhibit spatial‐temporal‐magnitude clustering just like tectonic ones. Successful application of the automated catalog‐building workflow also sheds light on the power of “hands‐free” processing of large‐volume seismic data.

Injection-Induced Earthquakes on Complex Fault Zones of the Raton Basin Illuminated by Machine-Learning Phase Picker and Dense Nodal Array


 The ever-increasing networks and quantity of seismic data drive the need for seamless and automatic workflows for rapid and accurate earthquake detection and location. In recent years, machine learning (ML)-based pickers have achieved remarkable accuracy and efficiency with generalization, and thus can significantly improve the earthquake location accuracy of previously developed sequential location methods. However, the inconsistent input or output (I/O) formats between multiple packages often limit their cross application. To reduce format barriers, we incorporated a widely used ML phase picker—PhaseNet—with several popular earthquake location methods and developed a “hands-free” end-to-end ML-based location workflow (named LOC-FLOW), which can be applied directly to continuous waveforms and build high-precision earthquake catalogs at local and regional scales. The renovated open-source package assembles several sequential algorithms including seismic first-arrival picking (PhaseNet and STA/LTA), phase association (REAL), absolute location (VELEST and HYPOINVERSE), and double-difference relative location (hypoDD and GrowClust). We provided different location strategies and I/O interfaces for format conversion to form a seamless earthquake location workflow. Different algorithms can be flexibly selected and/or combined. As an example, we apply LOC-FLOW to the 28 September 2004 Mw 6.0 Parkfield, California, earthquake sequence. LOC-FLOW accomplished seismic phase picking, association, velocity model updating, station correction, absolute location, and double-difference relocation for 16-day continuous seismic data. We detected and located 3.7 times (i.e., 4357) as many as earthquakes with cross-correlation double-difference locations from the Northern California Earthquake Data Center. Our study demonstrates that LOC-FLOW is capable of building high-precision earthquake catalogs efficiently and seamlessly from continuous seismic data.

LOC-FLOW: An End-to-End Machine Learning-Based High-Precision Earthquake Location Workflow

The 2019 M_w 7.1 Ridgecrest earthquake occurred on July 6th, preceded by the M_w 6.4 foreshock on July 4th, 2019. These two earthquakes occurred close in space and time with partially overlapping surface ruptures and aftershock patterns, raising the question of the relationship between the two events. Geological surveys and satellite observations provide important constraints on the surface traces of faulting. However, the subsurface fault geometries, which are important for understanding the regional stress field, earthquake initiation, propagation, and termination, are not well resolved. In this study, moment tensor solutions for 256 earthquakes in the 2019 Ridgecrest sequence were determined by waveform inversion using 3D velocity model. The obtained moment tensor solutions show rotations of the stresses after mainshock, indicating the ratio of mainshock stress drop to the background stress to be 0.5‐0.9. The obtained moment tensor catalog also facilitates a better understanding of the sub‐surface fault geometries, including (1) splay faults and antithetic faults in the northwest aftershock zone; (2) shallow flower structures near the M_w 7.1 epicenter; (3) sub‐parallel faults in the southeast aftershock zone. The aftershocks’ studies suggest the very complex surface ruptures near the 2019 Ridgecrest M_w 7.1 epicenter are near‐surface features that linked to a simple large through‐going fault at >5 km depth. We also found that the southeastern‐most aftershocks, which are located less than a kilometer from the Garlock fault trace, have significant different strike directions from that of the Garlock fault, indicating the central Garlock fault remains seismically quiet.

Seismotectonics and Fault Geometries of the 2019 Ridgecrest Sequence: Insight From Aftershock Moment Tensor Catalog Using 3-D Green's Functions

Rapid Characterization of the July 2019 Ridgecrest, California, Earthquake Sequence From Raw Seismic Data Using Machine-Learning Phase Picker

\n Microearthquake detection and location are critical for understanding earthquake mechanisms and mitigating seismic hazards. Match and locate (M&L) is an effective method for simultaneously detecting and locating small earthquakes. However, the heavy computational demands of the M&L make it challenging to apply to big data. In this article, we develop an improved M&L method—called graphics processing unit-based M&L (GPU-M&L). The GPU-M&L differs from the M&L in two ways: (1) adding weighting factor for each component of templates to improve the detection ability and (2) implementing the M&L method on GPU to accelerate the computation. Synthetic tests show the GPU-M&L can not only handle smaller earthquakes than the M&L but also perform 4.5 times faster than the M&L parallelly programed on central processing unit. As an example, we utilize the GPU-M&L to study the seismic activity during seven days after the 2015 Ms 5.8 Alxa, China, earthquake (from 15 to 21 April 2015). Using 38 cataloged earthquakes as templates, we detect ∼20 times more events than in the routine catalog. The distribution of those detected events, along with focal mechanisms of large events, suggests that the 2015 Ms 5.8 earthquake occurred on an east–west-trending hidden strike-slip fault.

Graphics Processing Unit‐Based Match and Locate (GPU‐M&L): An Improved Match and Locate Method and Its Application

On 21 May 2021, an MS 6.4 earthquake occurred in Yangbi, Yunnan province, China. A significant foreshock sequence occurred 3 days before the mainshock, which provides an opportunity to study earthquake nucleation. In this study, we adopt a template matching technique and a double‐difference location method to build a catalog for the foreshock sequence of the 2021 Yangbi MS 6.4 earthquake, which contains 1,086 events with high‐precision locations. The whole foreshock sequence can be grouped into four episodes, each with one MW > 3.7 principal foreshock. The foreshocks within the first episode initiated on a NW‐trending small fault, whereas the subsequent foreshocks within episodes 2–4 jumped onto a nearby major fault to gradually rupture toward NW direction. We identify three MW > 3.4 repeating earthquakes on the NW‐trending small fault in episode 1, implying that aseismic slip occurred in the early stage of the mainshock nucleation. Meanwhile, the rupture dimension estimation and stress disturbance analysis of the later foreshocks in episodes 2–4 reveal a cascading rupture process on the major fault, indicating that stress transfer may dominate the later stage of the mainshock nucleation. Hence, our study suggests that the two end‐member mechanisms of aseismic slip and stress transfer may be compossible in the nucleation process of the 2021 Yangbi MS 6.4 mainshock.

Min Liu

Papers

Rapid Characterization of the July 2019 Ridgecrest, California, Earthquake Sequence From Raw Seismic Data Using Machine-Learning Phase Picker

LOC-FLOW: An End-to-End Machine Learning-Based High-Precision Earthquake Location Workflow

Graphics Processing Unit‐Based Match and Locate (GPU‐M&L): An Improved Match and Locate Method and Its Application

Multistage Nucleation of the 2021 Yangbi MS 6.4 Earthquake, Yunnan, China and Its Foreshocks

Investigation of the 2013 Eryuan, Yunnan, China MS 5.5 Earthquake Sequence: Aftershock Migration, Seismogenic Structure and Hazard Implication