Distributed acoustic sensing (DAS) systems, which use existing telecommunication fibers, offer high-resolution capabilities ideal for recording anthropogenic sources. However, the complexity of urban environments and the large amount of data recorded by DAS require automated methods to efficiently detect and categorize anthropogenic sources. We evaluate how well three machine learning models (k-nearest neighbor [k-NN], convolutional neural networks, and recurrent-convolutional neural networks) can identify various anthropogenic sources recorded by DAS. Our findings reveal that both k-NN and neural network methods perform well in high signal-to-noise ratio (SNR) settings. However, their accuracy decreases at SNRs < 4. We also use Kalman filtering, a form of Bayesian inference, on backprojected locations of these sources to recover locations that generally fall within standard smartphone Global Positioning System errors. By combining machine learning and Kalman filter results, we calculate a multidimensional model of moving anthropogenic sources. These results demonstrate the potential of DAS data in urban seismology for accurately identifying and locating such sources. Depending on the research objectives, these sources can be further studied or filtered out to improve the quality of seismic data for earthquake studies. Such methods provide a valuable tool for urban seismology and seismic hazard analysis.
The accurate estimation of seismic wave amplitude is vital to precisely determine the yield, magnitude, and event discrimination possible for a given network – a critical element in nuclear explosion monitoring. This task is complicated by several factors, including but not limited to radiation pattern, scattering effects, and crustal variations, which can lead to the attenuation or amplification of amplitude along a given raypath. In this report, we explore the novel application of machine learning to the task of seismic amplitude estimation by training a simple Artificial Neural Network (ANN) on an S-wave amplitude dataset from Lai et al. (2019). Attributes from this dataset used as input to the ANN included event-station distances, station locations (latitude, longitude), event locations (latitude, longitude), event depths, event magnitudes, radiation patterns, signal-to noise ratio (SNR) measurements (average-amplitude, peak-to-trough, maximum peak), and signal periods. We find that the trained ANN predicts S-wave amplitudes with a modest tendency toward underestimating the actual values, as indicated by a linear regression between predicted and actual data (slope: 0.892, intercept: -0.651). These results suggest that an ANN can perform this task, with potential for significant improvements through improved datasets, architectures, and parameter tuning.
Distributed Acoustic Sensing (DAS) can record acoustic wavefields at high sampling rates and with dense spatial resolution difficult to achieve with seismometers. Using optical scattering induced by cable deformation, DAS can record strain fields with spatial resolution of a few meters. However, many experiments utilizing DAS have relied on unused, dark telecommunication fibers. As a result, the geophysical community has not fully explored DAS survey parameters to characterize the ideal array design. This limits our understanding of guiding principles in array design to deploy DAS effectively and efficiently in the field. A better quantitative understanding of DAS array behavior can improve the quality of the data recorded by guiding the DAS array design. Here we use steered response functions, which account for DAS fiber’s directional sensitivity, as well as beamforming and back-projection results from forward modelling calculations to assess the performance of varying DAS array geometries to record regional and local sources. A regular heptagon DAS array demonstrated improved capabilities for recording regional sources over other polygonal arrays, with potential improvements in recording and locating local sources. These results help reveal DAS array performance as a function of geometry and can guide future DAS deployments.
The generation of synthetic seismograms through simulation is a fundamental tool of seismology required to run quantitative hypothesis tests. A variety of approaches have been developed throughout the seismological community and each has their own specific user interface based on their implementation. This causes a challenge to researchers who will need to learn new interfaces with each new software they wish to use and create substantial challenges when attempting to compare results from different tools. Here we provide a unified interface that facilitates interoperability amongst several simulation tools through a modern containerized Python package. Further, this package includes post-processing analysis modules designed to facilitate end-to-end analysis of synthetic seismograms. In this report we present the conceptual guidance and an example implementation of the new Waveform Simulation Framework.
Wuestefeld, Andreas; Spica, Zack J.; Aderhold, Kasey; Huang, Hsin-Hua; Ma, Kuo-Fong; Lai, Voon H.; Miller, Meghan; Urmantseva, Lena; Zapf, Daniel; Bowden, Daniel C.; Edme, Pascal; Kiers, Tjeerd; Rinaldi, Antonio P.; Tuinstra, Katinka; Jestin, Camille; Diaz-Meza, Sergio; Jousset, Philippe; Wollin, Christopher; Ugalde, Arantza; Ruiz Barajas, Sandra; Gaite, Beatriz; Currenti, Gilda; Prestifilippo, Michele; Araki, Eiichiro; Tonegawa, Takashi; De Ridder, Sjoerd; Nowacki, Andy; Lindner, Fabian; Schoenball, Martin; Wetter, Christoph; Zhu, Hong-Hu; Baird, Alan F.; Rorstadbotnen, Robin A.; Ajo-Franklin, Jonathan; Ma, Yuanyuan; Abbott, Robert; Hodgkinson, Kathleen M.; Porritt, Robert W.; Stanciu, Adrian C.; Podrasky, Agatha; Hill, David; Biondi, Biondo; Yuan, Siyuan; Bin LuoBin; Nikitin, Sergei; Morten, Jan P.; Dumitru, Vlad-Andrei; Lienhart, Werner; Cunningham, Erin; Wang, Herbert
During February 2023, a total of 32 individual distributed acoustic sensing (DAS) systems acted jointly as a global seismic monitoring network. The aim of this Global DAS Month campaign was to coordinate a diverse network of organizations, instruments, and file formats to gain knowledge and move toward the next generation of earthquake monitoring networks. During this campaign, 156 earthquakes of magnitude 5 or larger were reported by the U.S. Geological Survey and contributors shared data for 60 min after each event’s origin time. Participating systems represent a variety of manufacturers, a range of recording parameters, and varying cable emplacement settings (e.g., shallow burial, borehole, subaqueous, and dark fiber). Monitored cable lengths vary between 152 and 120,129 m, with channel spacing between 1 and 49 m. The data has a total size of 6.8 TB, and are available for free download. Finally, organizing and executing the Global DAS Month has produced a unique dataset for further exploration and highlighted areas of further development for the seismological community to address.
The SALSA3D tomographic model provides a crucial community resource for improving the quality (in terms of both accuracy and precision) of predictions of the traveltimes of seismic waves and therefore improving our ability to locate anthropogenic or natural seismic events. Constructing the requisite tomographic model requires addressing the challenges implied by a massive and growing dataset of traveltime measurements. This study explores one approach to tackle this challenge: the use of summary rays, which average traveltime measurements from sources within evenly spaced cells, thereby eliminating redundant data.
Distributed Acoustic Sensing (DAS) is a rapidly developing technology that can record acoustic wavefields at high sampling rates and with dense spatial spacing difficult to achieve with seismometers. However, the geophysical community has not fully explored DAS survey parameters to characterize the ideal array design. A better quantitative understanding of DAS array behavior prior to SPE Phase III acquisition can help improve the quality of the data recorded by guiding the DAS array design. Here we use array response functions as well as beamforming and backprojection results from forward modelling calculations to assess the performance of varying DAS array geometries to record regional and local sources. A seven-sided polygon DAS array demonstrated improved capabilities for recording regional sources over segmented linear arrays, with potential improvements in recording and locating local sources. These results help reveal DAS array performance as a function of geometry.
Distributed Acoustic Sensing (DAS) is an emerging technology capable of recording the acoustic wavefield at unprecedented spatial resolution. However, this new tool requires significant refinements before it becomes operational for explosion monitoring objectives. Recent studies have shown significant development of array processing with DAS data. In this report we explore three such array processing methods including DAS strain-rate data versus geophone measured ground motion, beamforming for event parameters, and machine learning based denoising. Prior to applying these algorithms to the Source Physics Experiment Phase II and Phase III data, we validate these methods through replication analysis.
Co-Author Voon Hui Lai, Australian National University, will be attending IUGG and presenting the work. The session of interested is "S05 - Advances in Earthquake and Explosion Monitoring Using Distributed Acoustic Sensing"
Accurate measurements of the arrival times of seismic waves are crucial for seismological analyses such as robust locations of earthquakes, characterization of seismic sources, and high-fidelity imaging of the Earth’s interior. However, these travel-time measurements can sometimes be contaminated by timing errors at the stations which record this data. In this study, we apply a classical approach, based on identifying time-dependence in measured body wave arrival times, to identify these timing errors in a dataset on the order of 107 individual measurements. We find timing deviations at a subset of the stations in our dataset and document the temporal location, extent, and severity of these errors, finding errors at 83 stations, and impacting ~100,000 measurements. This catalog of deviations may enable future investigators to obtain a more accurate dataset through the implementation of quality control measures to eliminate the contaminated data we have identified.
Seismic waveform data are generally contaminated by noise from various sources, which interfere with the signals of interest. In this study, we implemented and applied several noise suppression methods using data recorded by the regional network of the University of Utah Seismograph stations. The denoising methods, consisting of approaches based on nonlinear thresholding of continuous wavelet transforms (CWTs, e.g., Langston and Mousavi, 2019), convolutional neural network (CNN) denoising (Tibi et al., 2021), and frequency filtering, were all subjected to the same analyses and level of scrutiny. We found that for frequency filtering, the improvement in signal-to-noise ratio (SNR) decreases quickly with decreasing SNR of the input waveform, and that below an input SNR of about 32 dB the improvement is relatively marginal and nearly constant. In contrast, the SNR gains are low at high-input SNR and increase with decreasing input SNR to reach the top of the plateaus corresponding to gains of about 18 and 23 dB, respectively, for CWT and CNN denoising. The low gains at high-input SNRs for these methods can be explained by the fact that for an input waveform with already high SNR (low noise), only very little improvement can be achieved by denoising, if at all. Results involving 4780 constructed waveforms suggest that in terms of degree of fidelity for the denoised waveforms with respect to the ground truth seismograms, CNN denoising outperforms both CWT denoising and frequency filtering. Onset time picking analyses by an experienced expert analyst suggest that CNN denoising allows more picks to be made com-pared with frequency filtering or CWT denoising and is on par with the expert analyst’s processing that follows current operational procedure. The CWT techniques are more likely to introduce artifacts that made the waveforms unusable.
The SALSA3D project aims to improve our models used in travel time prediction. The current version uses tomographic modeling for propagation through the Earth’s mantle because of the large number (order of millions) of observations of seismic phases which primarily traverse the Earth’s mantle and the ability to pose the travel time problem as a set of linear equations. However, all seismic rays traverse the crust to reach receivers at the surface and therefore models of propagation through the crust are required. Therefore, the primary motivation for this study is to explore how to increase the scope of the SALSA3D project to phases which travel primarily through the crust. In this report, we evaluate new, machine learning based and physics-based methods to model these travel times for integration into the SALSA3D framework. Our results suggest that using our existing physics-based travel time tomography method is a viable approach for the regional to global scale, but better predictive capabilities can be achieved through a neural network trained on the region of interest for near-regional offsets. We suggest future iterations of SALSA3D should incorporate machine learning tools such as Physics-Informed Neural Networks or Bayesian Neural Networks.
