Articles | Volume 13, issue 11
https://doi.org/10.5194/se-13-1721-2022
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/se-13-1721-2022
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Method article
| 
10 Nov 2022
Method article |
| 10 Nov 2022
A study on the effect of input data length on a deep-learning-based magnitude classifier
Megha Chakraborty
Frankfurt Institute for Advanced Studies, 60438 Frankfurt am Main,
Germany
Institute of Geosciences, Goethe University Frankfurt, 60438 Frankfurt
am Main, Germany
Wei Li
Frankfurt Institute for Advanced Studies, 60438 Frankfurt am Main,
Germany
Johannes Faber
Frankfurt Institute for Advanced Studies, 60438 Frankfurt am Main,
Germany
Institute for Theoretical Physics, Goethe University Frankfurt, 60438
Frankfurt am Main, Germany
Georg Rümpker
Frankfurt Institute for Advanced Studies, 60438 Frankfurt am Main,
Germany
Institute of Geosciences, Goethe University Frankfurt, 60438 Frankfurt
am Main, Germany
Horst Stoecker
Frankfurt Institute for Advanced Studies, 60438 Frankfurt am Main,
Germany
Institute for Theoretical Physics, Goethe University Frankfurt, 60438
Frankfurt am Main, Germany
Xidian-FIAS International Joint Research Center, Giersch Science
Center, 60438 Frankfurt am Main, Germany
GSI Helmholtzzentrum für Schwerionenforschung GmbH, 64291
Darmstadt, Germany
Nishtha Srivastava
CORRESPONDING AUTHOR
Frankfurt Institute for Advanced Studies, 60438 Frankfurt am Main,
Germany
Institute of Geosciences, Goethe University Frankfurt, 60438 Frankfurt
am Main, Germany
Related authors
Wei Li, Megha Chakraborty, Jonas Köhler, Claudia Quinteros Cartaya, Georg Rümpker, and Nishtha Srivastava
EGUsphere, https://doi.org/10.48550/arXiv.2211.09539, https://doi.org/10.48550/arXiv.2211.09539, 2023
Short summary
Short summary
Seismic phase picking & magnitude estimation are crucial components of real-time earthquake monitoring & early warning. Here, we test the potential of deep learning in real-time earthquake monitoring. We introduce DynaPicker which leverages dynamic convolutional neural networks for event detection & arrival time picking. Subsequently, we use a deep-learning model 'CREIME' for magnitude estimation. This workflow is tested on the continuous recording of the Turkey earthquake aftershock sequences.
Fabian Limberger, Georg Rümpker, Michael Lindenfeld, and Hagen Deckert
Solid Earth, 14, 859–869, https://doi.org/10.5194/se-14-859-2023, https://doi.org/10.5194/se-14-859-2023, 2023
Short summary
Short summary
Wind turbines that are located close to a seismometer produce ground tremors that can increase the noise level at the seismic station. Using numerical models, we analyse the effectivity of borehole installations to reduce this impact. We study effects of geophysical parameters on the borehole effectivity and validate our modelling approach with data from real boreholes. Boreholes are effective in reducing the impact of wind turbines; however, this depends on the wavelength of the seismic wave.
Wei Li, Megha Chakraborty, Jonas Köhler, Claudia Quinteros Cartaya, Georg Rümpker, and Nishtha Srivastava
EGUsphere, https://doi.org/10.48550/arXiv.2211.09539, https://doi.org/10.48550/arXiv.2211.09539, 2023
Short summary
Short summary
Seismic phase picking & magnitude estimation are crucial components of real-time earthquake monitoring & early warning. Here, we test the potential of deep learning in real-time earthquake monitoring. We introduce DynaPicker which leverages dynamic convolutional neural networks for event detection & arrival time picking. Subsequently, we use a deep-learning model 'CREIME' for magnitude estimation. This workflow is tested on the continuous recording of the Turkey earthquake aftershock sequences.
Carola Leva, Georg Rümpker, and Ingo Wölbern
Solid Earth, 13, 1243–1258, https://doi.org/10.5194/se-13-1243-2022, https://doi.org/10.5194/se-13-1243-2022, 2022
Short summary
Short summary
The seismicity of Fogo and Brava, Cape Verde, is dominated by volcano-tectonic earthquakes in the area of Brava and volcanic seismic signals, such as hybrid events, on Fogo. We locate these events using a multi-array analysis, which allows the localization of seismic events occurring outside the network and of volcanic signals lacking clear phases. We observe exceptionally high apparent velocities for the hybrid events located on Fogo. These velocities are likely caused by a complex ray path.
Fabian Limberger, Michael Lindenfeld, Hagen Deckert, and Georg Rümpker
Solid Earth, 12, 1851–1864, https://doi.org/10.5194/se-12-1851-2021, https://doi.org/10.5194/se-12-1851-2021, 2021
Short summary
Short summary
Frequency-dependent amplitude decays of seismic signals induced by wind turbines are determined from (up to) 6 months of continuous recordings measured along an 8 km profile located at a wind farm in Bavaria, Germany. The radiation pattern and amplitude decay of the induced signals are accounted for by an analytical approach that includes path and source effects. This approach is generalized to predict the characteristic seismic radiation patterns of arbitrary wind farm configurations.
Manvendra Singh and Georg Rümpker
Solid Earth, 11, 2557–2568, https://doi.org/10.5194/se-11-2557-2020, https://doi.org/10.5194/se-11-2557-2020, 2020
Short summary
Short summary
Using seismic array methods, 63 events were located in the Rodrigues–CIR region, not reported by any global network, most of them being off the ridge axis. The lack of seismicity along this section of the CIR, as observed from global data and this study, can possibly be attributed to the presence of partially molten mantle beneath Rodrigues Ridge. The results will be of interest for a broad range of geoscientists interested in the tectonic evolution of Indian Ocean and plume–crust interaction.
Carola Leva, Georg Rümpker, and Ingo Wölbern
Nat. Hazards Earth Syst. Sci., 20, 3627–3638, https://doi.org/10.5194/nhess-20-3627-2020, https://doi.org/10.5194/nhess-20-3627-2020, 2020
Short summary
Short summary
Often, an abrupt increase in shallow seismicity at volcanoes is seen as an indicator for magmatic intrusions into the upper crust. If no eruption occurs and the seismic activity stops, this is called a failed eruption. Here, we report a failed eruption of Brava, Cabo Verde, in August 2016. We remotely monitored the seismicity of Brava with a seismic array, operating from October 2015 to December 2016. Other episodes with increased seismicity around the island were also observed during the study.
Related subject area
Subject area: The evolving Earth surface | Editorial team: Seismics, seismology, paleoseismology, geoelectrics, and electromagnetics | Discipline: Seismology
A borehole trajectory inversion scheme to adjust the measurement geometry for 3D travel-time tomography on glaciers
Ocean bottom seismometer (OBS) noise reduction from horizontal and vertical components using harmonic–percussive separation algorithms
Towards real-time seismic monitoring of a geothermal plant using Distributed Acoustic Sensing
Upper-lithospheric structure of northeastern Venezuela from joint inversion of surface-wave dispersion and receiver functions
Multi-array analysis of volcano-seismic signals at Fogo and Brava, Cape Verde
Reflection imaging of complex geology in a crystalline environment using virtual-source seismology: case study from the Kylylahti polymetallic mine, Finland
The damaging character of shallow 20th century earthquakes in the Hainaut coal area (Belgium)
The effect of 2020 COVID-19 lockdown measures on seismic noise recorded in Romania
Accelerating Bayesian microseismic event location with deep learning
Strain to ground motion conversion of distributed acoustic sensing data for earthquake magnitude and stress drop determination
Regional centroid moment tensor inversion of small to moderate earthquakes in the Alps using the dense AlpArray seismic network: challenges and seismotectonic insights
Unprecedented quiescence in resource development area allows detection of long-lived latent seismicity
Seismic monitoring of urban activity in Barcelona during the COVID-19 lockdown
Seismic signature of the COVID-19 lockdown at the city scale: a case study with low-cost seismometers in the city of Querétaro, Mexico
Characterizing the oceanic ambient noise as recorded by the dense seismo-acoustic Kazakh network
Seismic evidence of the COVID-19 lockdown measures: a case study from eastern Sicily (Italy)
Sensing Earth and environment dynamics by telecommunication fiber-optic sensors: an urban experiment in Pennsylvania, USA
Effects of finite source rupture on landslide triggering: the 2016 Mw 7.1 Kumamoto earthquake
Sebastian Hellmann, Melchior Grab, Cedric Patzer, Andreas Bauder, and Hansruedi Maurer
Solid Earth, 14, 805–821, https://doi.org/10.5194/se-14-805-2023, https://doi.org/10.5194/se-14-805-2023, 2023
Short summary
Short summary
Acoustic waves are suitable to analyse the physical properties of the subsurface. For this purpose, boreholes are quite useful to deploy a source and receivers in the target area to get a comprehensive high-resolution dataset. However, when conducting such experiments in a subsurface such as glaciers that continuously move, the boreholes get deformed. In our study, we therefore developed a method that allows an analysis of the ice while considering deformations.
Zahra Zali, Theresa Rein, Frank Krüger, Matthias Ohrnberger, and Frank Scherbaum
Solid Earth, 14, 181–195, https://doi.org/10.5194/se-14-181-2023, https://doi.org/10.5194/se-14-181-2023, 2023
Short summary
Short summary
Investigation of the global Earth's structure benefits from the analysis of ocean bottom seismometer (OBS) data that allow an improved seismic illumination of dark spots of crustal and mantle structures in the oceanic regions of the Earth. However, recordings from the ocean bottom are often highly contaminated by noise. We developed an OBS noise reduction algorithm, which removes much of the oceanic noise while preserving the earthquake signal and does not introduce waveform distortion.
Jerome Azzola, Katja Thiemann, and Emmanuel Gaucher
EGUsphere, https://doi.org/10.5194/egusphere-2022-1417, https://doi.org/10.5194/egusphere-2022-1417, 2022
Preprint archived
Short summary
Short summary
Distributed Acoustic Sensing is applied to the micro-seismic monitoring of a geothermal plant. In this domain, the feasibility of managing the large flow of generated data and their suitability to monitor locally induced seismicity was yet to be assessed. The proposed monitoring system efficiently managed the acquisition, processing and saving of the data over a 6-month period. This testing period proved that the monitoring concept advantageously complements more classical monitoring networks.
Roberto Cabieces, Mariano S. Arnaiz-Rodríguez, Antonio Villaseñor, Elizabeth Berg, Andrés Olivar-Castaño, Sergi Ventosa, and Ana M. G. Ferreira
Solid Earth, 13, 1781–1801, https://doi.org/10.5194/se-13-1781-2022, https://doi.org/10.5194/se-13-1781-2022, 2022
Short summary
Short summary
This paper presents a new 3D shear-wave velocity model of the lithosphere of northeastern Venezuela, including new Moho and Vp / Vs maps. Data were retrieved from land and broadband ocean bottom seismometers from the BOLIVAR experiment.
Carola Leva, Georg Rümpker, and Ingo Wölbern
Solid Earth, 13, 1243–1258, https://doi.org/10.5194/se-13-1243-2022, https://doi.org/10.5194/se-13-1243-2022, 2022
Short summary
Short summary
The seismicity of Fogo and Brava, Cape Verde, is dominated by volcano-tectonic earthquakes in the area of Brava and volcanic seismic signals, such as hybrid events, on Fogo. We locate these events using a multi-array analysis, which allows the localization of seismic events occurring outside the network and of volcanic signals lacking clear phases. We observe exceptionally high apparent velocities for the hybrid events located on Fogo. These velocities are likely caused by a complex ray path.
Michal Chamarczuk, Michal Malinowski, Deyan Draganov, Emilia Koivisto, Suvi Heinonen, and Sanna Rötsä
Solid Earth, 13, 705–723, https://doi.org/10.5194/se-13-705-2022, https://doi.org/10.5194/se-13-705-2022, 2022
Short summary
Short summary
In passive seismic measurement, all noise sources from the environment, such as traffic, vibrations caused by distant excavation, and explosive work from underground mines, are utilized. In the Kylylahti experiment, receivers recorded ambient noise sources for 30 d. These recordings were subjected to data analysis and processing using novel methodology developed in our study and used for imaging the subsurface geology of the Kylylahti mine area.
Thierry Camelbeeck, Koen Van Noten, Thomas Lecocq, and Marc Hendrickx
Solid Earth, 13, 469–495, https://doi.org/10.5194/se-13-469-2022, https://doi.org/10.5194/se-13-469-2022, 2022
Short summary
Short summary
Over the 20th century, shallow damaging seismicity occurred in and near the Hainaut coal mining area in Belgium. We provide an overview of earthquake parameters and impacts, combining felt and damage testimonies and instrumental measurements. Shallower earthquakes have a depth and timing compatible with mining activity. The most damaging events occurred deeper than the mines but could still have been triggered by mining-caused crustal changes. Our modelling can be applied to other regions.
Bogdan Grecu, Felix Borleanu, Alexandru Tiganescu, Natalia Poiata, Raluca Dinescu, and Dragos Tataru
Solid Earth, 12, 2351–2368, https://doi.org/10.5194/se-12-2351-2021, https://doi.org/10.5194/se-12-2351-2021, 2021
Short summary
Short summary
The lockdown imposed in Romania to prevent the spread of COVID-19 has significantly impacted human activity across the country. By analyzing the ground vibrations recorded at seismic stations, we were able to monitor the changes in human activity before and during the lockdown.
The reduced human activity during the lockdown has also provided a good opportunity for stations sited in noisy urban areas to record earthquake signals that would not have been recorded under normal conditions.
Alessio Spurio Mancini, Davide Piras, Ana Margarida Godinho Ferreira, Michael Paul Hobson, and Benjamin Joachimi
Solid Earth, 12, 1683–1705, https://doi.org/10.5194/se-12-1683-2021, https://doi.org/10.5194/se-12-1683-2021, 2021
Short summary
Short summary
The localization of an earthquake is affected by many uncertainties. To correctly propagate these uncertainties into an estimate of the earthquake coordinates and their associated errors, many simulations of seismic waves are needed. This operation is computationally very intensive, hindering the feasibility of this approach. In this paper, we present a series of deep-learning methods to produce accurate seismic traces in a fraction of the time needed with standard methods.
Itzhak Lior, Anthony Sladen, Diego Mercerat, Jean-Paul Ampuero, Diane Rivet, and Serge Sambolian
Solid Earth, 12, 1421–1442, https://doi.org/10.5194/se-12-1421-2021, https://doi.org/10.5194/se-12-1421-2021, 2021
Short summary
Short summary
The increasing use of distributed acoustic sensing (DAS) inhibits the transformation of optical fibers into dense arrays of seismo-acoustic sensors. Here, DAS strain records are converted to ground motions using the waves' apparent velocity. An algorithm for velocity determination is presented, accounting for velocity variations between different seismic waves. The conversion allows for robust determination of fundamental source parameters, earthquake magnitude and stress drop.
Gesa Maria Petersen, Simone Cesca, Sebastian Heimann, Peter Niemz, Torsten Dahm, Daniela Kühn, Jörn Kummerow, Thomas Plenefisch, and the AlpArray and AlpArray-Swath-D working groups
Solid Earth, 12, 1233–1257, https://doi.org/10.5194/se-12-1233-2021, https://doi.org/10.5194/se-12-1233-2021, 2021
Short summary
Short summary
The Alpine mountains are known for a complex tectonic history. We shed light onto ongoing tectonic processes by studying rupture mechanisms of small to moderate earthquakes between 2016 and 2019 observed by the temporary AlpArray seismic network. The rupture processes of 75 earthquakes were analyzed, along with past earthquakes and deformation data. Our observations point at variations in the underlying tectonic processes and stress regimes across the Alps.
Rebecca O. Salvage and David W. Eaton
Solid Earth, 12, 765–783, https://doi.org/10.5194/se-12-765-2021, https://doi.org/10.5194/se-12-765-2021, 2021
Short summary
Short summary
Small earthquakes in Alberta and north-east British Columbia have been previously ascribed to industrial activities. The COVID-19 pandemic forced almost all these activities to stop for ~ 4 months. However, unexpectedly, earthquakes still occurred during this time. Some of these earthquakes may be natural and some the result of earthquakes > M6 occurring around the world. However, ~ 65 % of the earthquakes detected may be the remnants of previous fluid injection in the area (
latent seismicity).
Jordi Diaz, Mario Ruiz, and José-Antonio Jara
Solid Earth, 12, 725–739, https://doi.org/10.5194/se-12-725-2021, https://doi.org/10.5194/se-12-725-2021, 2021
Short summary
Short summary
During the COVID-19 pandemic lockdown, the city of Barcelona was covered by a network of 19 seismometers. The results confirm that the quieting of human activity during lockdown has resulted in a reduction of seismic vibrations. The different lockdown phases in Barcelona are recognized consistently at most of the seismic stations. Our contribution demonstrates that seismic noise can be used as a free and reliable tool to monitor human activity in urban environments.
Raphael S. M. De Plaen, Víctor Hugo Márquez-Ramírez, Xyoli Pérez-Campos, F. Ramón Zuñiga, Quetzalcoatl Rodríguez-Pérez, Juan Martín Gómez González, and Lucia Capra
Solid Earth, 12, 713–724, https://doi.org/10.5194/se-12-713-2021, https://doi.org/10.5194/se-12-713-2021, 2021
Short summary
Short summary
COVID-19 pandemic lockdowns in countries with a dominant informal economy have been a greater challenge than in other places. This motivated the monitoring of the mobility of populations with seismic noise throughout the various phases of lockdown and in the city of Querétaro (central Mexico). Our results emphasize the benefit of densifying urban seismic networks, even with low-cost instruments, to observe variations in mobility at the city scale over exclusively relying on mobile technology.
Alexandr Smirnov, Marine De Carlo, Alexis Le Pichon, Nikolai M. Shapiro, and Sergey Kulichkov
Solid Earth, 12, 503–520, https://doi.org/10.5194/se-12-503-2021, https://doi.org/10.5194/se-12-503-2021, 2021
Short summary
Short summary
Seismic and infrasound methods are techniques used to monitor natural events and explosions. At low frequencies, band signal can be dominated by microbaroms and microseisms. The noise observations in the Kazakh network are performed and compared with source and propagation modeling. The network is dense and well situated for studying very distant source regions of the ambient noise. The prospects are opening for the use of ocean noise in solid Earth and atmosphere tomography.
Andrea Cannata, Flavio Cannavò, Giuseppe Di Grazia, Marco Aliotta, Carmelo Cassisi, Raphael S. M. De Plaen, Stefano Gresta, Thomas Lecocq, Placido Montalto, and Mariangela Sciotto
Solid Earth, 12, 299–317, https://doi.org/10.5194/se-12-299-2021, https://doi.org/10.5194/se-12-299-2021, 2021
Short summary
Short summary
During the COVID-19 pandemic, most countries put in place social interventions, aimed at restricting human mobility, which caused a decrease in the seismic noise, generated by human activities and called anthropogenic seismic noise. In densely populated eastern Sicily, we observed a decrease in the seismic noise amplitude reaching 50 %. We found similarities between the temporal patterns of seismic noise and human mobility, as quantified by mobile-phone-derived data and ship traffic data.
Tieyuan Zhu, Junzhu Shen, and Eileen R. Martin
Solid Earth, 12, 219–235, https://doi.org/10.5194/se-12-219-2021, https://doi.org/10.5194/se-12-219-2021, 2021
Short summary
Short summary
We describe the Fiber Optic foR Environmental SEnsEing (FORESEE) project in Pennsylvania, USA, the first continuous-monitoring distributed acoustic sensing (DAS) fiber array in the eastern USA. With the success of collecting 1 year of continuous DAS recordings using nearly 5 km of telecommunication fiber underneath the university campus, we conclude that DAS along with telecommunication fiber will potentially serve the purpose of continuous near-surface seismic monitoring in populated areas.
Sebastian von Specht, Ugur Ozturk, Georg Veh, Fabrice Cotton, and Oliver Korup
Solid Earth, 10, 463–486, https://doi.org/10.5194/se-10-463-2019, https://doi.org/10.5194/se-10-463-2019, 2019
Short summary
Short summary
We show the landslide response to the 2016 Kumamoto earthquake (Mw 7.1) in central Kyushu (Japan). Landslides are concentrated to the northeast of the rupture, coinciding with the propagation direction of the earthquake. This azimuthal variation in the landslide concentration is linked to the seismic rupture process itself and not to classical landslide susceptibility factors. We propose a new ground-motion model that links the seismic radiation pattern with the landslide distribution.
Cited articles
Allen, R., Gasparini, P., Kamigaichi, O., and Böse, M.: The Status of
Earthquake Early Warning around the World: An Introductory Overview, Seismol. Res. Lett. 80, 682–693, https://doi.org/10.1785/gssrl.80.5.682, 2009.
Allen, R. and Kanamori, H.: The Potential for Earthquake Early Warning in
Southern California, Science, 300, 786–789, https://doi.org/10.1126/science.1080912,
2003.
Allen, R. M. and Melgar, D.: Earthquake Early Warning: Advances, Scientific
Challenges, and Societal Needs, Annu. Rev. Earth Planet Sc., 47, 361–388, https://doi.org/10.1146/annurev-earth-053018-060457, 2019.
Aly, M.: Survey on multiclass classification methods, Neural Netw., 19, 1–9,
2005.
Batista, G. E. A. P. A., Prati, R. C., and Monard, M. C.: A Study of the
Behavior of Several Methods for Balancing Machine Learning Training Data,
SIGKDD Explorations Newsletter, 6, 20–29, https://doi.org/10.1145/1007730.1007735,
2004.
Bengio, Y.: Practical Recommendations for Gradient-Based Training of Deep
Architectures, Neural Networks: Tricks of the Trade: Second Edition, Berlin, Heidelberg: Springer Berlin Heidelberg, 437–478, https://doi.org/10.1007/978-3-642-35289-8_26, 2012.
Chakraborty, M., Rümpker, G., Stöcker, H., Li, W., Faber, J., Fenner, D., Zhou, K., and Srivastava, N.: Real Time Magnitude Classification of Earthquake Waveforms using Deep Learning, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-15941, https://doi.org/10.5194/egusphere-egu21-15941, 2021.
Chung, D. H. and Bernreuter, D. L.: Regional relationships among earthquake
magnitude scales, Rev. Geophys., 19, 649–663, https://doi.org/10.1029/RG019i004p00649, 1981.
Collobert, R., Weston, J., Bottou, L., Karlen, M., Kavukcuoglu, K., and
Kuksa, P.: Natural language processing (almost) from scratch, J. Mach. Learn. Res., 12, 2493–2537, https://doi.org/10.5555/1953048.2078186, 2011.
Ekström, G. and Dziewonski, A.: Evidence of bias in estimations of
earthquake size, Nature, 332, 319–323,
https://doi.org/10.1038/332319a0, 1988.
Fawaz, H. I., Forestier, G., Weber, J., Idoumghar, L., and Muller, P. A.:
Deep learning for time series classification: a review, Data Min. Knowl. Disc., 33,
917–963, https://doi.org/10.1007/s10618-019-00619-1,
2019.
Goodfellow, I., Bengio, Y., and Courville, A.: Deep learning, MIT press., http://www.deeplearningbook.org (last access: 20 October 2022), 2016.
He, K., Ren, S., Sun, J., and Zhang, X.: Deep Residual Learning for Image
Recognition, Proc. Cvpr IEEE, 770–778, https://doi.org/10.1109/CVPR.2016.90, 2016.
Hinton, G., Deng, L., Yu, D., Dahl, G. E., Mohamed, A.R., Jaitly, N.,
Senior, A., Vanhoucke, V., Nguyen, P., Sainath, T. N., and Kingsbury, B.:
Deep Neural Networks for Acoustic Modeling in Speech Recognition: The
Shared Views of Four Research Groups, IEEE Signal Proc. Mag., 29, 82–97, https://doi.org/10.1109/MSP.2012.2205597, 2012.
Hochreiter, S. and Schmidhuber, J.: Long Short-Term Memory, Neural Comput., 9, 1735–1780, 1997.
Howell Jr., B. F.: On the saturation of earthquake magnitudes, B. Seismol. Soc. Am., 71,
1401–1422, https://doi.org/10.1785/BSSA0710051401, 1981.
Jin, X., Zhang, H., Li, J., Wei, Y., and Ma, Q.: Earthquake magnitude
estimation using the τc and Pd method for earthquake early
warning systems, Earthq. Sci., 26, 23–31, https://doi.org/10.1007/s11589-013-0005-4, 2013.
Kanamori, H.: Magnitude scale and quantification of earthquakes,
Tectonophysics, 93, 185–199, https://doi.org/10.1016/0040-1951(83)90273-1,
1983.
Kanamori, H.: Real-time seismology and earthquake damage mitigation,
Annu. Rev. Earth Planet Sc., 33, 195–214, https://doi.org/10.1146/annurev.earth.33.092203.122626, 2005.
Kanamori, H. and Stewart, G. S.: Seismological aspects of the Guatemala
Earthquake of February 4, 1976, J. Geophys. Res.-Sol. Ea., 83, 3427–3434, https://doi.org/10.1029/JB083iB07p03427, 1978.
Kingma, D. P. and Ba, J.: Adam: A Method for Stochastic Optimization, 3rd International Conference on Learning Representations, ICLR 2015,
San Diego, CA, USA, 7–9 May 2015, Conference Track Proceedings, 2015.
Kiranyaz, S., Avci, O., Abdeljaber, O., Ince, T., Gabbouj, M., and Inman, D. J.: 1D convolu tional neural networks and applications: A survey,
Mech. Sys. Signal Proc., 151, 107398,
https://doi.org/10.1016/j.ymssp.2020.107398, 2021.
Kong, Q., Trugman, D. T., Ross, Z. E., Bianco, M. J., Meade, B. J., and
Gerstoft, P.: Machine Learning in Seismology: Turning Data into Insights, Seismol. Res. Lett., 90, 3–14, https://doi.org/10.1785/0220180259,
2018.
Krawczyk, B.: Learning from imbalanced data: open challenges and future
directions, Prog. Artif., 5, 221–232, https://doi.org/10.1007/s13748-016-0094-0, 2016.
Krizhevsky, A., Sutskever, I., and Hinton, G. E.: ImageNet Classification
with Deep Convolutional Neural Networks, Communications of the ACM, 60, 84–90, https://doi.org/10.1145/3065386, 2017.
Kuyuk, H. S. and Susumu, O.: Real-time classification of earthquake using
deep learning, Proc. Comput. Sci., 140, 298–305, https://doi.org/10.1016/j.procs.2018.10.316, 2018.
LeCun, Y., Bengio, Y., and Hinton, G.: Deep learning, Nature, 521, 436–444, https://doi.org/10.1038/nature14539, 2015.
Li, W., Chakraborty, M., Fenner, D., Faber, J., Zhou, K., Ruempker, G.,
Stoecker, H., and Srivastava, N.: EPick: Multi-Class Attention-based U-shaped Neural Network for Earthquake Detection and Seismic Phase Picking, arXiv [preprint],
https://doi.org/10.48550/arXiv.2109.02567, 6 September 2021.
Liao, W. Y., Lee, E.J., Mu, D., Chen, P., and Rau, R. J.: ARRU Phase Picker:
Attention Recurrent-Residual U-Net for Picking Seismic P- and S-Phase
Arrivals, Seismol. Res. Lett., 92, 2410–2428, https://doi.org/10.1785/0220200382, 2021.
Lomax, A., Michelini, A., and Jozinović, D.: An Investigation of Rapid
Earthquake Characterization Using Single-Station Waveforms and a
Convolutional Neural Network, Seismol. Res. Lett., 90, 517–529,
https://doi.org/10.1785/0220180311, 2019.
Madhyastha, P. and Jain, R.: On Model Stability as a Function of Random Seed, arXiv [preprint],
https://doi.org/10.48550/arXiv.1909.10447, 2019.
Meier, M. A., Ross, Z. E., Ramachandran, A., Balakrishna, A., Nair, S.,
Kundzicz, P., Li, Z., Andrews, J., Hauksson, E., and Yue, Y.: Reliable
Real-Time Seismic Signal/Noise Discrimination With Machine Learning,
J. Geophys. Res. Sol.-Ea., 124, 788–800, https://doi.org/10.1029/2018JB016661, 2019.
Mikolov, T., Deoras, A., Povey, D., Burget, L., and Černocký, J.:
Strategies for training large scale neural network language models,
2011 IEEE Workshop on Automatic Speech Recognition Understanding, 196–201, https://doi.org/10.1109/ASRU.2011.6163930, 2011.
Mousavi, S. M. and Beroza, G. C.: A machine-learning approach for
earthquake magnitude estimation, Geophys. Res. Lett., 47, e2019GL085976,
https://doi.org/10.1029/2019GL085976, 2020.
Mousavi, S. M., Sheng, Y., Zhu, W., and Beroza, G. C.: STanford EArthquake
Dataset (STEAD): A Global Data Set of Seismic Signals for AI, IEEE Access, 7, 179464–179476, https://doi.org/10.1109/ACCESS.2019.2947848, 2019 (data available at: https://github.com/smousavi05/STEAD, last access: 20 October 2022).
Mousavi, S. M., Ellsworth, W. L., Zhu, W., Chuang, L. Y., and Beroza, G. C.:
Earthquake transformer – an attentive deep-learning model for simultaneous
earthquake detection and phase picking, Nat. Commun., 11, 3952, https://doi.org/10.1038/s41467-020-17591-w, 2020.
Münchmeyer, J., Bindi, D., Leser, U., and Tilmann, F.: The transformer
earthquake alerting model: a new versatile approach to earthquake early
warning, Geophys. J. Int., 225, 646–656, https://doi.org/10.1093/gji/ggaa609, 2020.
Murphy, K. P.: Machine learning: a probabilistic perspective, MIT press, ISBN 9780262018029, 2012.
Nagi, J., Ducatelle, F., Di Caro, G. A., Cireşan, D., Meier, U., Giusti,
A., Nagi, F., Schmidhuber, J., and Gambardella, L. M.: Max-pooling
convolutional neural networks for vision-based hand gesture recognition, 2011 IEEE International Conference on Signal and Image Processing Applications (ICSIPA), 342–347, https://doi.org/10.1109/ICSIPA.2011.6144164, 2011.
Nakamura, Y.: On the Urgent Earthquake Detection and Alarm System
(UrEDAS), 9th world conference on earthquake engineering, VII, B7, 673–678, 1988.
Panakkat, A. and Adeli, H.: Recurrent neural network for approximate
earthquake time and location prediction using multiple seismicity
indicators, Comput.-Aided Civ. Infrastruct. Eng., 24, 280–292, https://doi.org/10.1111/j.1467-8667.2009.00595.x, 2009.
Perol, T., Gharbi, M., and Denolle, M.: Convolutional neural network for
earthquake detection and location, Sci. Adv., 4, e1700578, https://doi.org/10.1126/sciadv.1700578, 2018.
Peters, M. E., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., and
Zettlemoyer, L.: Deep contextualized word representations, Association for Computational Linguistics,
2227–2237, https://doi.org/10.18653/v1/N18-1202, 2018.
Prechelt, L.: Early Stopping – But When?, Neural Networks: Tricks of the Trade:
Second Edition, Springer Berlin Heidelberg, 53–67, https://doi.org/10.1007/978-3-642-35289-8_5, 2012.
Ross, Z. E., Meier, M. A., and Hauksson, E.: P wave arrival picking and
first-motion polarity determination with deep learning, J. Geophys. Res. Sol.-Ea., 123, 5120–5129, https://doi.org/10.1029/2017JB015251, 2018.
Saad, O. M., Hafez, A. G., and Soliman, M. S.: Deep Learning Approach for
Earthquake Parameters Classification in Earthquake Early Warning System,
IEEE Geosci. Remote Sens. Lett., 18, 1293–1297, https://doi.org/10.1109/LGRS.2020.2998580, 2020.
Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., and
Salakhutdinov, R.: Dropout: A Simple Way to Prevent Neural Networks from
Overfitting, J. Mach. Learn. Res., 15, 1929–1958,
2014.
Ting, K. M.: Confusion Matrix, Encyclopedia of Machine Learning and Data Mining, Boston, MA: Springer US, 260–260, https://doi.org/10.1007/978-1-4899-7687-1_50, 2017.
Wang, J. and Teng, T. L.: Artificial neural network-based seismic
detector, B. Seismol. Soc. Am., 85, 308–319, https://doi.org/10.1785/BSSA0850010308, 1995.
Wen, Q., Sun, L., Yang, F., Song, X., Gao, J., Wang, X., and Xu, H.: Time
Series Data Augmentation for Deep Learning: A Survey, Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, https://doi.org/10.24963/ijcai.2021/631, 2021.
Wu, Y. M. and Zhao, L.: Magnitude estimation using the first three seconds
P-wave amplitude in earthquake early warning, Geophys. Res. Lett., 33, L16312, https://doi.org/10.1029/2006GL026871, 2006.
Zhu, W. and Beroza, G. C.: PhaseNet: a deep-neural-network-based seismic
arrival-time picking method, Geophys. J. Int., 216, 261–273, https://doi.org/10.1093/gji/ggy423, 2019.
Ziv, A.: New frequency-based real-time magnitude proxy for earthquake early
warning, Geophys. Res. Lett., 41, 7035–7040, https://doi.org/10.1002/2014GL061564, 2014.
Short summary
Earthquake magnitude is a crucial parameter in defining its damage potential, and hence its speedy determination is essential to issue an early warning in regions close to the epicentre. This study summarises our findings in an attempt to apply deep-learning-based classifiers to earthquake waveforms, particularly with respect to finding an optimum length of input data. We conclude that the input length has no significant effect on the model accuracy, which varies between 90 %–94 %.
Earthquake magnitude is a crucial parameter in defining its damage potential, and hence its...
