Characterizing Acoustic Signals and Searching for Precursors during the Laboratory Seismic Cycle Using Unsupervised Machine Learning

Bolton, David; Shokouhi, Parisa; Rouet‐Leduc, Bertrand; Hulbert, Claudia; Rivière, Jacques; Marone, Chris; Johnson, Paul A.

doi:10.1785/0220180367

Cited by 58 publications

(73 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…AE monitoring under controlled experimental conditions has been widely used to investigate rock fracturing (e.g., Lockner et al, 1991;Sellers et al, 2003;Johnson et al, 2008;Benson et al, 2008;Goebel et al, 2013;Bolton et al, 2019). Traditional AE technology is limited by data acquisition systems and computer storage capabilities, and it is mainly based on statistical parameters (e.g., event count, location, type of failure, and energy) that characterize the activity of AE (e.g., Ishida et al, 2017;Mair et al, 2007).…”

Section: Ae Monitoring Experimentsmentioning

confidence: 99%

Recent Advances and Challenges of Waveform‐Based Seismic Location Methods at Multiple Scales

Tan

Schwarz

et al. 2020

Reviews of Geophysics

129

View full text Add to dashboard Cite

Source locations provide fundamental information on earthquakes and lay the foundation for seismic monitoring at all scales. Seismic source location as a classical inverse problem has experienced significant methodological progress during the past century. Unlike the conventional traveltime‐based location methods that mainly utilize kinematic information, a new category of waveform‐based methods, including partial waveform stacking, time reverse imaging, wavefront tomography, and full waveform inversion, adapted from migration or stacking techniques in exploration seismology has emerged. Waveform‐based methods have shown promising results in characterizing weak seismic events at multiple scales, especially for abundant microearthquakes induced by hydraulic fracturing in unconventional and geothermal reservoirs or foreshock and aftershock activity potentially preceding tectonic earthquakes. This review presents a comprehensive summary of the current status of waveform‐based location methods, through elaboration of the methodological principles, categorization, and connections, as well as illustration of the applications to natural and induced/triggered seismicity, ranging from laboratory acoustic emission to field hydraulic fracturing‐induced seismicity, regional tectonic, and volcanic earthquakes. Taking into account recent developments in instrumentation and the increasing availability of more powerful computational resources, we highlight recent accomplishments and prevailing challenges of different waveform‐based location methods and what they promise to offer in the near future.

show abstract

Section: Ae Monitoring Experimentsmentioning

confidence: 99%

Recent Advances and Challenges of Waveform‐Based Seismic Location Methods at Multiple Scales

Tan

Schwarz

et al. 2020

Reviews of Geophysics

129

View full text Add to dashboard Cite

show abstract

“…It is evident that stick-slip failure (see arrow 3 on Figure 1) is preceded by a number of spikes of AD (see arrow 2 and similar symbols on Figure 1). These spikes appear as a result of micro failure events and may predict TTF [12,13]. Generally, the shorter the TTF the more frequent the AD spikes.…”

Section: Feature Engineeringmentioning

confidence: 99%

“…The variance of the seismic signal is the most important feature, although other statistical characteristics are also important [26,29,31]. The authors of [13] stressed that the kurtosis of the acoustic signal is an additional powerful feature for the prediction of TTF.…”

Section: Feature Engineeringmentioning

confidence: 99%

“…In general, the higher the number of features analyzed, the greater the time required to solve the model. For example, a set of 43 features [13] gives a total of 8.796 × 10 12 combinations, which would require a significant amount of time to find the best combination.…”

Section: Feature Engineeringmentioning

confidence: 99%

“…Therefore, predicting the time and magnitude of earthquakes is one of the fundamental goals of geoscience [12]. Despite long time-series of observations and field research, the accurate prediction of the size or timing of earthquakes remains a challenge [13,14]. Even more concerning is that devastating subduction earthquakes with magnitude 9.0 or more are considered unpredictable [14].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Machine Learning Modelling and Feature Engineering in Seismology Experiment

Брыков

Petryshynets

Pruncu

et al. 2020

Sensors

View full text Add to dashboard Cite

This article aims to discusses machine learning modelling using a dataset provided by the LANL (Los Alamos National Laboratory) earthquake prediction competition hosted by Kaggle. The data were obtained from a laboratory stick-slip friction experiment that mimics real earthquakes. Digitized acoustic signals were recorded against time to failure of a granular layer compressed between steel plates. In this work, machine learning was employed to develop models that could predict earthquakes. The aim is to highlight the importance and potential applicability of machine learning in seismology The XGBoost algorithm was used for modelling combined with 6-fold cross-validation and the mean absolute error (MAE) metric for model quality estimation. The backward feature elimination technique was used followed by the forward feature construction approach to find the best combination of features. The advantage of this feature engineering method is that it enables the best subset to be found from a relatively large set of features in a relatively short time. It was confirmed that the proper combination of statistical characteristics describing acoustic data can be used for effective prediction of time to failure. Additionally, statistical features based on the autocorrelation of acoustic data can also be used for further improvement of model quality. A total of 48 statistical features were considered. The best subset was determined as having 10 features. Its corresponding MAE was 1.913 s, which was stable to the third decimal point. The presented results can be used to develop artificial intelligence algorithms devoted to earthquake prediction.

show abstract

Unsupervised learning of seismic wavefield features: clustering continuous array seismic data during the 2009 L'Aquila earthquake

Shi

Seydoux

Poli

2020

Preprint

View full text Add to dashboard Cite

show abstract

Characterizing Acoustic Signals and Searching for Precursors during the Laboratory Seismic Cycle Using Unsupervised Machine Learning

Cited by 58 publications

References 40 publications

Recent Advances and Challenges of Waveform‐Based Seismic Location Methods at Multiple Scales

Recent Advances and Challenges of Waveform‐Based Seismic Location Methods at Multiple Scales

Machine Learning Modelling and Feature Engineering in Seismology Experiment

Unsupervised learning of seismic wavefield features: clustering continuous array seismic data during the 2009 L'Aquila earthquake

Contact Info

Product

Resources

About