Long Short-Term Memory Networks for Pattern Recognition of Synthetical Complete Earthquake Catalog

Exploring the spatiotemporal distribution of earthquake activity, especially earthquake migration of fault systems, can greatly to understand the basic mechanics of earthquakes and the assessment of earthquake risk. By establishing a three-dimensional strike-slip fault model, to derive the stress response and fault slip along the fault under regional stress conditions. Our study helps to create a long-term, complete earthquake catalog. We modelled Long-Short Term Memory (LSTM) networks for pattern recognition of the synthetical earthquake catalog. The performance of the models was compared using the mean-square error (MSE). Our results showed clearly the application of LSTM showed a meaningful result of 0.08% in the MSE values. Our best model can predict the time and magnitude of the earthquakes with a magnitude greater than Mw = 6.5 with a similar clustering period. These results showed conclusively that applying LSTM in a spatiotemporal series prediction provides a potential application in the study of earthquake mechanics and forecasting of major earthquake events.

[1]  Y. Kagan,et al.  Earthquakes Cannot Be Predicted , 1997, Science.

[2]  Nitish Srivastava,et al.  Improving neural networks by preventing co-adaptation of feature detectors , 2012, ArXiv.

[3]  Chun-yong Wang,et al.  Evidence for mechanically coupled lithosphere in central Asia and resulting implications , 2008 .

[4]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[5]  I. Main,et al.  A Poisson model for earthquake frequency uncertainties in seismic hazard analysis , 2008, 0807.2396.

[6]  Yosihiko Ogata,et al.  Statistical Models for Earthquake Occurrences and Residual Analysis for Point Processes , 1988 .

[7]  D. Sornette,et al.  Global Earthquake Forecasting System (GEFS): The challenges ahead , 2021 .

[8]  G. Luo,et al.  Seismicity, Fault Slip Rates, and Fault Interactions in a Fault System , 2020, Journal of Geophysical Research: Solid Earth.

[9]  Jiankun He,et al.  3D Finite‐Element Modeling of Earthquake Interaction and Stress Accumulation on Main Active Faults around the Northeastern Tibetan Plateau Edge in the Past ∼100 Years , 2015 .

[10]  Yosihiko Ogata,et al.  Statistical model for standard seismicity and detection of anomalies by residual analysis , 1989 .

[11]  T. Utsu.  Aftershocks and Earthquake Statistics(2) : Further Investigation of Aftershocks and Other Earthquake Sequences Based on a New Classification of Earthquake Sequences , 1971 .

[12]  S. Stein,et al.  Long aftershock sequences in North China and Central US: implications for hazard assessment in mid-continents , 2014 .

[13]  Yoshua Bengio,et al.  Learning long-term dependencies with gradient descent is difficult , 1994, IEEE Trans. Neural Networks.

[14]  Peizhen Zhang,et al.  FEM simulation of interseismic and coseismic deformation associated with the 2008 Wenchuan Earthquake , 2013 .

[15]  Y. Ogata.  Space-Time Point-Process Models for Earthquake Occurrences , 1998 .

[16]  Michael I. Jordan,et al.  Machine learning: Trends, perspectives, and prospects , 2015, Science.

[17]  D. L. Anderson,et al.  Theoretical Basis of Some Empirical Relations in Seismology by Hiroo Kanamori And , 1975 .

[18]  Y. Kagan.  Seismic moment distribution revisited: I. Statistical results , 2002 .

[19]  Yuji Kim,et al.  Geophysical inversion versus machine learning in inverse problems , 2018, The Leading Edge.

[20]  Yaolin Shi,et al.  Influence of fault geometry and fault interaction on strain partitioning within western Sichuan and its adjacent region , 2010 .

[21]  A. Freed.  EARTHQUAKE TRIGGERING BY STATIC, DYNAMIC, AND POSTSEISMIC STRESS TRANSFER , 2005 .

[22]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[23]  Gernot Beer,et al.  Numerical methods in rock mechanics , 1990 .

[24]  Yan Y. Kagan,et al.  Statistical distributions of earthquake numbers: consequence of branching process , 2009, 0909.2372.

[25]  T. Utsu.  A statistical study on the occurrence of aftershocks. , 1961 .

[26]  M. Marques,et al.  Recent advances and applications of machine learning in solid-state materials science , 2019, npj Computational Materials.

[27]  Maarten V. de Hoop,et al.  Machine learning for data-driven discovery in solid Earth geoscience , 2019, Science.

[28]  G. Luo,et al.  Multi-timescale mechanical coupling between the San Jacinto fault and the San Andreas fault, southern California , 2012 .

[29]  Danijel Schorlemmer,et al.  The Statistical Power of Testing Probabilistic Seismic-Hazard Assessments , 2014 .

[30]  Y. Kagan.  Earthquake slip distribution: A statistical model , 2005 .

[31]  Francesco Mulargia,et al.  Why is Probabilistic Seismic Hazard Analysis (PSHA) still used? , 2017, Physics of the Earth and Planetary Interiors.

[32]  Vito Latora,et al.  Power-law time distribution of large earthquakes. , 2003, Physical review letters.

[33]  Tiago P. Peixoto,et al.  Spatiotemporal correlations of aftershock sequences , 2010, 1004.1924.

[34]  G. Luo,et al.  Stress evolution and fault interactions before and after the 2008 Great Wenchuan earthquake , 2010 .

[35]  G. King,et al.  STATIC STRESS CHANGES AND THE TRIGGERING OF EARTHQUAKES , 1994 .

[36]  Carmine Galasso,et al.  Earthquake early warning: Recent advances and perspectives , 2020, Earth-Science Reviews.

[37]  Jürgen Schmidhuber,et al.  Framewise phoneme classification with bidirectional LSTM and other neural network architectures , 2005, Neural Networks.

[38]  Mian Liu,et al.  A 3-D viscoelastoplastic model for simulating long-term slip on non-planar faults , 2009 .

[39]  Y. Ogata.  ESTIMATION OF THE PARAMETERS IN THE MODIFIED OMORI FORMULA FOR AFTERSHOCK FREQUENCIES BY THE MAXIMUM LIKELIHOOD PROCEDURE , 1983 .