An official website of the United States government
Abstract The San Fernando Valley (SFV), part of the Los Angeles metropolitan area, is a seismically active urban environment. Large-magnitude earthquakes, such as the 1994 Mw 6.7 Northridge event that occurred on a blind fault beneath the valley, caused significant infrastructure damage in the region, underscoring the need for enhanced seismic monitoring to improve the identification of buried faults and hazard evaluation. Currently, the Southern California Earthquake Data Center operates four broadband instruments within the valley; however, the network’s ability to capture small earthquakes beneath the region may be limited. To demonstrate how this data gap can be filled, we use recordings from the SFV array, comprised of 140 nodal instruments with interstation distances ranging from 0.3 to 2.5 km that recorded for one month. High-anthropogenic noise levels in urbanized areas tend to conceal earthquake signals; therefore, we applied a previously developed machine learning model fine-tuned on similar waveforms to detect events and pick seismic phases. In a two-step event association workflow, isolated phase picks were first culled, which eliminated false positive detections and reduced computational runtime. We located 62 events within a 209 km radius of our array with magnitudes ranging from ML 0.13 to 4, including 36 new events that were undetected by the regional network. One event cluster reveals a previously unidentified (5.3 km × 4 km) blind fault zone located ∼5 km beneath the southern part of the valley. Seismicity from this zone is rare in the regional catalog (<3 events per year), despite producing a Mb 4.4 event in 2014. Our results highlight the benefits of detecting small-magnitude seismicity for hazard estimation. Temporary nodal arrays can identify critical gaps in regional monitoring and guide site selection for permanent stations. In addition, our workflow can be applied to complement seismic monitoring in other urban settings.
more »
« less
Rapid Estimation of Single-Station Earthquake Magnitudes with Machine Learning on a Global Scale
ABSTRACT The foundation of earthquake monitoring is the ability to rapidly detect, locate, and estimate the size of seismic sources. Earthquake magnitudes are particularly difficult to rapidly characterize because magnitude types are only applicable to specific magnitude ranges, and location errors propagate to substantial magnitude errors. We developed a method for rapid estimation of single-station earthquake magnitudes using raw three-component P waveforms observed at local to teleseismic distances, independent of prior size or location information. We used the MagNet regression model architecture (Mousavi and Beroza, 2020b), which combines convolutional and recurrent neural networks. We trained our model using ∼2.4 million P-phase arrivals labeled by the authoritative magnitude assigned by the U.S. Geological Survey. We tested input data parameters (e.g., window length) that could affect the performance of our model in near-real-time monitoring applications. At the longest waveform window length of 114 s, our model (Artificial Intelligence Magnitude [AIMag]) is accurate (median estimated magnitude within ±0.5 magnitude units from catalog magnitude) between M 2.3 and 7.6. However, magnitudes above M ∼7 are more underestimated as true magnitude increases. As the windows are shortened down to 1 s, the point at which higher magnitudes begin to be underestimated moves toward lower magnitudes, and the degree of underestimation increases. The over and underestimation of magnitudes for the smallest and largest earthquakes, respectively, are potentially related to the limited number of events in these ranges within the training data, as well as magnitude saturation effects related to not capturing the full source time function of large earthquakes. Importantly, AIMag can determine earthquake magnitudes with individual stations’ waveforms without instrument response correction or knowledge of an earthquake’s source-station distance. This work may enable monitoring agencies to more rapidly recognize large, potentially tsunamigenic global earthquakes from few stations, allowing for faster event processing and reporting. This is critical for timely warnings for seismic-related hazards.
more »
« less
- Award ID(s):
- 1835661
- PAR ID:
- 10576069
- Publisher / Repository:
- BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA
- Date Published:
- Journal Name:
- Bulletin of the Seismological Society of America
- Volume:
- 114
- Issue:
- 3
- ISSN:
- 0037-1106
- Page Range / eLocation ID:
- 1523 to 1538
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Earthquake early warning (EEW) systems aim to forecast the shaking intensity rapidly after an earthquake occurs and send warnings to affected areas before the onset of strong shaking. The system relies on rapid and accurate estimation of earthquake source parameters. However, it is known that source estimation for large ruptures in real‐time is challenging, and it often leads to magnitude underestimation. In a previous study, we showed that machine learning, HR‐GNSS, and realistic rupture synthetics can be used to reliably predict earthquake magnitude. This model, called Machine‐Learning Assessed Rapid Geodetic Earthquake model (M‐LARGE), can rapidly forecast large earthquake magnitudes with an accuracy of 99%. Here, we expand M‐LARGE to predict centroid location and fault size, enabling the construction of the fault rupture extent for forecasting shaking intensity using existing ground motion models. We test our model in the Chilean Subduction Zone with thousands of simulated and five real large earthquakes. The result achieves an average warning time of 40.5 s for shaking intensity MMI4+, surpassing the 34 s obtained by a similar GNSS EEW model. Our approach addresses a critical gap in existing EEW systems for large earthquakes by demonstrating real‐time fault tracking feasibility without saturation issues. This capability leads to timely and accurate ground motion forecasts and can support other methods, enhancing the overall effectiveness of EEW systems. Additionally, the ability to predict source parameters for real Chilean earthquakes implies that synthetic data, governed by our understanding of earthquake scaling, is consistent with the actual rupture processes.more » « less
-
ABSTRACT Earthquake ground motions in the vicinity of receivers couple with the atmosphere to generate pressure perturbations that are detectable by infrasound sensors. These so-called local infrasound signals traverse very short source-to-receiver paths, so that they often exhibit a remarkable correlation with seismic velocity waveforms at collocated seismic stations, and there exists a simple relationship between vertical seismic velocity and pressure time series. This study leverages the large regional network of infrasound sensors in Alaska to examine local infrasound from several light to great Alaska earthquakes. We estimate seismic velocity time series from infrasound pressure records and use these converted infrasound recordings to compute earthquake magnitudes. This technique has potential utility beyond the novelty of recording seismic velocities on pressure sensors. Because local infrasound amplitudes from ground motions are small, it is possible to recover seismic velocities at collocated sites where the broadband seismometers have clipped. Infrasound-derived earthquake magnitudes exhibit good agreement with seismically derived values. This proof-of-concept demonstration of computing seismic magnitudes from infrasound sensors illustrates that infrasound sensors may be utilized as proxy vertical-component seismometers, making a new data set available for existing seismic techniques. Because single-sensor infrasound stations are relatively inexpensive and are becoming ubiquitous, this technique could be used to augment existing regional seismic networks using a readily available sensor platform.more » « less
-
Fault damage zones can influence various aspects of the earthquake cycle, such as the recurrence intervals and magnitudes of large earthquakes. The properties and structure of fault damage zones are often characterized using dense arrays of seismic stations located directly above the faults. However, such arrays may not always be available. Hence, our research aims to develop a novel method to image fault damage zones using broadband stations at relatively larger distances. Previous kinematic simulations and a case study of the 2003 Big Bear earthquake sequence demonstrated that fault damage zones can act as effective waveguides, amplifying high-frequency waves along directions close to fault strike via multiple reflections within the fault damage zone. The amplified high-frequency energy can be observed by stacking P-wave spectra of earthquake clusters with highly-similar waveforms (Huang et al., 2016), and the frequency band which is amplified may be used to estimate the width and velocity contrast of the fault damage zone. We attempt to identify the high-frequency peak associated with fault zone waves in stacked spectra by conducting a large-scale study of small earthquakes (M1.5–3). We use high quality broadband data from seismic stations at hypocentral distances of 20-80 km in the 2019 Ridgecrest earthquake regions. First, we group the Ridgecrest earthquakes in clusters by their locations and their waveform similarity, and then stack their velocity spectra to average the source effects of individual earthquakes. Our results show that the stations close to the fault strike record more high-frequency energies around the characteristic frequency of fault zone reflections. We find that the increase in the amount of high-frequencies is consistent across clusters with average magnitudes ranging from 1.6-2.4, which suggests that the azimuthal variation in spectra is caused by fault zone amplification rather than rupture directivity. We will apply our method to other fault zones in California, in order to search for fault damage zone structures and estimate their material properties.more » « less
-
Fault damage zones can influence various aspects of the earthquake cycle, such as the recurrence intervals and magnitudes of large earthquakes. The properties and structure of fault damage zones are often characterized using dense arrays of seismic stations located directly above the faults. However, such arrays may not always be available. Hence, our research aims to develop a novel method to image fault damage zones using broadband stations at relatively larger distances. Previous kinematic simulations and a case study of the 2003 Big Bear earthquake sequence demonstrated that fault damage zones can act as effective waveguides, amplifying high-frequency waves along directions close to fault strike via multiple reflections within the fault damage zone. The amplified high-frequency energy can be observed using the stacked P-wave spectra of earthquake clusters with highly-similar waveforms (Huang et al., 2016). We attempt to identify the high-frequency peak associated with fault zone waves in stacked spectra by conducting a large-scale study of small earthquakes (M1.5–3). We use high quality broadband data from seismic stations at hypocentral distances of 20-100km in the 2004 Parkfield and 2019 Ridgecrest earthquake regions. First, we group earthquakes in clusters by their locations and their waveform similarity, and then stack their velocity spectra to average the source effects of individual earthquakes. We applied our method to the 2019 Ridgecrest earthquake sequence, and our preliminary results show that stations close to the fault strike tend to record more high-frequency energies around the characteristic frequency of fault zone reflections. The frequency bands in which amplified high-frequency energies are observed may be used to estimate the width and velocity contrast of the fault damage zone. We aim to develop a robust and versatile method that can be used to search for fault damage zone structures and estimate their material properties, in order to shed light on earthquake source processes.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1785/0120230171
