Applications of machine learning in seismology have greatly improved our capability of detecting earthquakes in large seismic data archives. Most of these efforts have been focused on continental shallow earthquakes, but here we introduce an integrated deep-learning-based workflow to detect deep earthquakes recorded by a temporary array of ocean-bottom seismographs (OBSs) and land-based stations in the Tonga subduction zone. We develop a new phase picker, PhaseNet-TF, to detect and pick P- and S-wave arrivals in the time–frequency domain. The frequency-domain information is critical for analysing OBS data, particularly the horizontal components, because they are contaminated by signals of ocean-bottom currents and other noise sources in certain frequency bands. PhaseNet-TF shows a much better performance in picking S waves at OBSs and land stations compared to its predecessor PhaseNet. The predicted phases are associated using an improved Gaussian Mixture Model Associator GaMMA-1D and then relocated with a double-difference package teletomoDD. We further enhance the model performance with a semi-supervised learning approach by iteratively refining labelled data and retraining PhaseNet-TF. This approach effectively suppresses false picks and significantly improves the detection of small earthquakes. The new catalogue of Tonga deep earthquakes contains more than 10 times more events compared to the reference catalogue that was analysed manually. This deep-learning-enhanced catalogue reveals Tonga seismicity in unprecedented detail, and better defines the lateral extent of the double-seismic zone at intermediate depths and the location of four large deep-focus earthquakes relative to background seismicity. It also offers new potential for deciphering deep earthquake mechanisms, refining tomographic models, and understanding of subduction processes.
Distributed Acoustic Sensing (DAS) is an emerging technology for earthquake monitoring and subsurface imaging. However, its distinct characteristics, such as unknown ground coupling and high noise level, pose challenges to signal processing. Existing machine learning models optimized for conventional seismic data struggle with DAS data due to its ultra-dense spatial sampling and limited manual labels. We introduce a semi-supervised learning approach to address the phase-picking task of DAS data. We use the pre-trained PhaseNet model to generate noisy labels of P/S arrivals in DAS data and apply the Gaussian mixture model phase association (GaMMA) method to refine these noisy labels and build training datasets. We develop PhaseNet-DAS, a deep learning model designed to process 2D spatio-temporal DAS data to achieve accurate phase picking and efficient earthquake detection. Our study demonstrates a method to develop deep learning models for DAS data, unlocking the potential of integrating DAS in enhancing earthquake monitoring.
more » « less- Award ID(s):
- 1848166
- PAR ID:
- 10479007
- Publisher / Repository:
- Nature Publishing Group
- Date Published:
- Journal Name:
- Nature Communications
- Volume:
- 14
- Issue:
- 1
- ISSN:
- 2041-1723
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
more » « less
SUMMARY -
Abstract Accurate and (near) real-time earthquake monitoring provides the spatial and temporal behaviors of earthquakes for understanding the nature of earthquakes, and also helps in regional seismic hazard assessments and mitigations. Because of the increase in both the quality and quantity of seismic data, an automated earthquake monitoring system is needed. Most of the traditional methods for detecting earthquake signals and picking phases are based on analyses of features in recordings of an individual earthquake and/or their differences from background noises. When seismicity is high, the seismograms are complicated, and, therefore, traditional analysis methods often fail. With the development of machine learning algorithms, earthquake signal detection and seismic phase picking can be more accurate using the features obtained from a large amount of earthquake recordings. We have developed an attention recurrent residual U-Net algorithm, and used data augmentation techniques to improve the accuracy of earthquake detection and seismic phase picking on complex seismograms that record multiple earthquakes. The use of probability functions of P and S arrivals and potential P and S arrival pairs of earthquakes can increase the computational efficiency and accuracy of backprojection for earthquake monitoring in large areas. We applied our workflow to monitor the earthquake activity in southern California during the 2019 Ridgecrest sequence. The distribution of earthquakes determined by our method is consistent with that in the Southern California Earthquake Data Center (SCEDC) catalog. In addition, the number of earthquakes in our catalog is more than three times that of the SCEDC catalog. Our method identifies additional earthquakes that are close in origin times and/or locations, and are not included in the SCEDC catalog. Our algorithm avoids misidentification of seismic phases for earthquake location. In general, our algorithm can provide reliable earthquake monitoring on a large area, even during a high seismicity period.more » « less
-
more » « less
Abstract Distributed Acoustic Sensing (DAS) is a promising technique to improve the rapid detection and characterization of earthquakes. Previous DAS studies mainly focus on the phase information but less on the amplitude information. In this study, we compile earthquake data from two DAS arrays in California, USA, and one submarine array in Sanriku, Japan. We develop a data‐driven method to obtain the first scaling relation between DAS amplitude and earthquake magnitude. Our results reveal that the earthquake amplitudes recorded by DAS in different regions follow a similar scaling relation. The scaling relation can provide a rapid earthquake magnitude estimation and effectively avoid uncertainties caused by the conversion to ground motions. Our results show that the scaling relation appears transferable to new regions with calibrations. The scaling relation highlights the great potential of DAS in earthquake source characterization and early warning.
-
null (Ed.)Abstract Seismograms are convolution results between seismic sources and the media that seismic waves propagate through, and, therefore, the primary observations for studying seismic source parameters and the Earth interior. The routine earthquake location and travel-time tomography rely on accurate seismic phase picks (e.g., P and S arrivals). As data increase, reliable automated seismic phase-picking methods are needed to analyze data and provide timely earthquake information. However, most traditional autopickers suffer from low signal-to-noise ratio and usually require additional efforts to tune hyperparameters for each case. In this study, we proposed a deep-learning approach that adapted soft attention gates (AGs) and recurrent-residual convolution units (RRCUs) into the backbone U-Net for seismic phase picking. The attention mechanism was implemented to suppress responses from waveforms irrelevant to seismic phases, and the cooperating RRCUs further enhanced temporal connections of seismograms at multiple scales. We used numerous earthquake recordings in Taiwan with diverse focal mechanisms, wide depth, and magnitude distributions, to train and test our model. Setting the picking errors within 0.1 s and predicted probability over 0.5, the AG with recurrent-residual convolution unit (ARRU) phase picker achieved the F1 score of 98.62% for P arrivals and 95.16% for S arrivals, and picking rates were 96.72% for P waves and 90.07% for S waves. The ARRU phase picker also shown a great generalization capability, when handling unseen data. When applied the model trained with Taiwan data to the southern California data, the ARRU phase picker shown no cognitive downgrade. Comparing with manual picks, the arrival times determined by the ARRU phase picker shown a higher consistency, which had been evaluated by a set of repeating earthquakes. The arrival picks with less human error could benefit studies, such as earthquake location and seismic tomography.more » « less
-
The longwall mining method is designed to optimize coal extraction through controlled roof caving, which inevitably induces seismicity. This research employs a distributed acoustic sensing (DAS) system incorporating a fire-safe fiber-optic cable strategically installed underground within an operational longwall coal mine. Despite lower sensitivity than traditional seismometers, DAS sensing technology benefits from dense sensor spacing and close proximity to the active face, where many microseismic events occur. To automatically detect seismic events within the voluminous DAS data records, we employ convolutional autoencoder deep learning models that can be used for anomaly (potential seismic event) detection in power spectral density (PSD) images of DAS recordings. The kernel density estimation (KDE) technique is used to calculate the probability density function (PDF) for the density scores of the latent space (representation of compressed data). We then use this calculated parameter as a threshold to distinguish between the PSD associated with background noise and with potential seismic events. The DAS monitoring system in conjunction with the developed deep learning model could enhance longwall coal mining safety and efficiency by offering valuable data from its densely deployed multichannel sensors near mining operations.more » « less
