Forecasting strong subsequent earthquakes in Japan using an improved version of NESTORE machine learning algorithm

Abstract

In this study, the advanced machine learning algorithm NESTORE (Next STrOng Related Earthquake) was applied to the Japan Meteorological Agency catalog (1973–2024). It calculates the probability that the aftershocks will reach or exceed a magnitude equal to the magnitude of the mainshock minus one and classifies the clusters as type A or type B, depending on whether this condition is met or not. It has been shown useful in the tests in Italy, western Slovenia, Greece, and California. Due to Japan’s high and complex seismic activity, new algorithms were developed to complement NESTORE: a hybrid cluster identification method, which uses both ETAS-based stochastic declustering and deterministic graph-based selection, and REPENESE (RElevant features, class imbalance PErcentage, NEighbour detection, SElection), an algorithm for detecting outliers in skewed class distributions, which takes in account if one class has a larger number of samples with respect to the other (class imbalance).

Keywords

Machine learning / Cluster identification / ETAS / Strong aftershock / Japan / Outliers detection

References

E.-A. Anyfadi, S. Gentili, P. Brondi, F. Vallianatos. Forecasting strong subsequent earthquakes in Greece with the machine learning algorithm NESTORE. Entropy, 25 (5) (2023), p. 797,

D.F. Argus, R.G. Gordon, C. DeMets. Geologically current motion of 56 plates relative to the no-net-rotation reference frame. Geochem. Geophys. Geosyst., 12 (2011), p. 11,

M. Båth. Lateral inhomogeneities in the upper mantle. Tectonophysics, 2 (1965), pp. 483-514,

Brondi, P., Gentili, S., Di Giovambattista R., 2023. Forecasting strong aftershocks in the Italian territory: A National and Regional application for NESTOREv1.0, EGU General Assembly 2023, Vienna, Austria, 23–28 April 2023.

Brondi, P., Gentili, S., Di Giovambattista R., 2024. Forecasting strong subsequent events in the Italian territory: a National and Regional application for NESTOREv1.0. Nat. Hazards.

K. Dascher-Cousineau, O. Shchur, E.E. Brodsky, S. Günnemann. Using deep learning for flexible and scalable earthquake forecasting. Geophys. Res. Lett., 50 (17) (2023), Article e2023GL103909,

S.D. Davis, C. Frohlich. Single-link cluster analysis, synthetic earthquake catalogues, and aftershock identification. Geophys. J. Int., 104 (2) (1991), pp. 289-306,

L. de Arcangelis, C. Godano, E. Lippiello. The overlap of aftershock coda waves and short-term postseismic forecasting. J. Geophys. Res. Solid Earth, 123 (7) (2018), pp. 5661-5674,

C. DeMets, R.G. Gordon, D.F. Argus. Geologically current plate motions. Geophys. J. Int., 181 (1) (2010), pp. 1-80,

P.M.R. DeVries, F. Viégas, M. Wattenberg, B.J. Meade. Deep learning of aftershock patterns following large earthquakes. Nature, 560 (2018), pp. 632-634,

J.P. Egan. Signal Detection Theory and ROC Analysis, Series in Cognition and Perception. Academic Press, New York (1975)

M. Ester, H.-P. Kriegel, J. Sander, X. Xu. A density-based algorithm for discovering clusters in large spatial databases with noise. Proc. 2nd Int. Conf. on Knowledge Discovery and Data Mining (1996)

T. Fawcett. An introduction to ROC analysis. Pattern Recogn. Lett., 27 (8) (2006), pp. 861-874,

C. Frohlich, S.D. Davis. Single-Link Cluster Analysis as a Method to Evaluate Spatial and Temporal Properties of Earthquake Catalogues. Geophys. J. Int., 100 (1) (1990), pp. 19-32,

Y. Fukushima, T. Nishikawa, Y. Kano. High probability of successive occurrence of Nankai megathrust earthquakes. Sci. Rep., 13 (2023), p. 63,

J.K. Gardner, L. Knopoff. Is the sequence of earthquakes in Southern California, with aftershocks removed, Poissonian?. Bullet. Seismol. Soc. Am., 64 (5) (1974), pp. 1363-1367,

S. Gentili. Radiated energy evolution during seismic sequences. Phys. Earth Planet. Inter., 196–197 (2012), pp. 49-61,

S. Gentili, G. Bressan. The partitioning of radiated energy and the largest aftershock of seismic sequences occurred in the northeastern Italy and western Slovenia. J. Seismol., 12 (2008), pp. 343-354,

S. Gentili, R. Di Giovambattista. Pattern recognition approach to the subsequent event of damaging earthquakes in Italy. Phys. Earth Planet. Inter., 266 (2017), pp. 1-17,

S. Gentili, R. Di Giovambattista. Forecasting strong aftershocks in earthquake clusters from northeastern Italy and western Slovenia. Phys. Earth Planet. Inter., 303 (2020), Article 106483,

S. Gentili, R. Di Giovambattista. Forecasting strong subsequent earthquakes in California clusters by machine learning. Phys. Earth Planet. Inter., 327 (2022), Article 106879,

S. Gentili, P. Brondi, R. Di Giovambattista. NESTOREv1.0: A MATLAB package for strong forthcoming earthquake forecasting. Seismol. Res. Lett., 94 (4) (2023), pp. 2003-2013,

S. Gentili, P. Brondi, G. Rossi, M. Sugan, G. Petrillo, J. Zhuang, S. Campanella. Seismic clusters and fluids diffusion: a lesson from the 2018 Molise (Southern Italy) earthquake sequence. Earth Planets Space, 76 (2024), p. 157,

L. Gulia, S. Wiemer, G. Vannucci. Prospective evaluation of the foreshock traffic light system in Ridgecrest and implications for aftershock hazard assessment. Seismol. Res. Lett., 91 (5) (2020), pp. 2828-2842,

F. Hirose, K. Maeda. Earthquake forecast models for inland Japan based on the G-R law and the modified G-R law. Earth Planets Space, 63 (2011), pp. 239-260,

M. Hoshiba, O. Kamigaichi, M. Saito, S. Tsukada, N. Hamada. Earthquake Early Warning Starts Nationwide in Japan. Eos Trans. Am. Geoph. Un., 89 (8) (2008), pp. 73-74,

T. Iidaka, K. Obara. Shear-wave splitting in a region with newly-activated seismicity after the 2011 Tohoku earthquake. Earth Planets Space, 65 (2013), pp. 1059-1064,

T. Ishibe, K. Shimazaki, K. Satake, H. Tsuruoka. Change in seismicity beneath the Tokyo metropolitan area due to the 2011 off the Pacific coast of Tohoku Earthquake. Earth Planets Space, 63 (2011), p. 40,

Japan Meteorological Agency, 2024. The seismological bulletin of Japan. https://www.data.jma.go.jp/svd/eqev/data/bulletin/index_e.html.

Y. Kagan. Seismic moment distribution revisited: I. Statistical Results. Geophys. J. Int., 148 (2002), pp. 520-541,

J. Kaizer, I. Kontuľ, P.P. Povinec. Impact of the Fukushima accident on 3H and 14C environmental levels: A review of ten years of investigation. Molecules, 28 (6) (2023), p. 2548,

V.I. Keilis-Borok, V.G. Kossobokov. Time of increased probability for the great earthquakes of the world. Comput. Seismol, 19 (1986), pp. 48-58

Y. Kodera, N. Hayashimoto, K. Tamaribuchi, K. Noguchi, K. Moriwaki, R. Takahashi, M. Morimoto, K. Okamoto, M. Hoshiba. Developments of the nationwide earthquake early warning system in Japan after the 2011 Mw 9.0 Tohoku-Oki earthquake. Front. Earth Sci., 9 (2021), Article 726045,

T. Lay. A review of the rupture characteristics of the 2011 Tohoku-oki Mw 9.1 earthquake. Tectonophysics, 733 (2018), pp. 4-36,

E. Lippiello, G. Petrillo, F. Landes, A. Rosso. The genesis of aftershocks in Spring slider Models. Stat. Methods Modeling Seismog., 1 (2021), pp. 131-151,

E. Lippiello, G. Petrillo. b-more-incomplete and b-more-positive: Insights on a robust estimator of magnitude distribution. J. Geophys. Res. : Solid Earth, 129 (2024), Article e2023JB027849,

B. Lolli, P. Gasperini. Aftershocks hazard in Italy Part I: estimation of time magnitude distribution model parameters and computation of probabilities of occurrence. J. Seismol., 7 (2003), pp. 235-257,

J. Luo, J. Zhuang. Three regimes of the distribution of the largest event in the critical ETAS model. Bull. Seismol. Soc. Am., 106 (3) (2016), pp. 1364-1369,

C. Molkenthin, R.V. Donner, S. Reich, G. Zoller, S. Hainzl, M. Holschneider, M. Opper. GP-ETAS: semiparametric Bayesian inference for the spatio-temporal epidemic type aftershock sequence model. Stat. Comput., 32 (2022), p. 29,

F. Musmeci, D. Vere-Jones. A space-time clustering model for historical earthquakes. Ann. Inst. Stat. Math., 44 (1992), pp. 1-11,

K.Z. Nanjo, T. Ishibe, H. Tsuruoka, D. Schorlemmer, Y. Ishigaki, N. Hirata. Analysis of the completeness magnitude and seismic network coverage of Japan. Bull. Seismol. Soc. Am., 100 (6) (2010), pp. 3261-3268,

Y. Ogata. Statistical models for earthquake occurrences and residual analysis for point processes. J. Am. Stat. Asso., 83 (401) (1988), pp. 9-27,

Y. Ogata. Space-time point-process models for earthquake occurrences. Annals Institute Stat. Math., 50 (1998), pp. 379-402,

Y. Ogata, J. Zhuang. Space-time ETAS model and an improved extension. Tectonophysics, 413 (1–2) (2006), pp. 13-23,

T. Omi, Y. Ogata, Y. Hirata, K. Aihara. Forecasting large aftershocks within one day after the main shock. Sci. Rep., 3 (2013), p. 2218,

T. Parsons, R. Console, G. Falcone, M. Murru, K. Yamashina. Comparison of characteristic and Gutenberg—Richter models for time-dependent M ≥ 7.9 earthquake probability in the Nankai-Tokai subduction zone, Japan. Geophys. J. Int., 190 (3) (2012), pp. 1673-1688,

G. Petrillo, T. Kumazawa, F. Napolitano, P. Capuano, J. Zhuang. Fluids‐triggered swarm sequence supported by a nonstationary epidemic‐like description of seismicity. Seismol. Res. Lett., 95 (6) (2024), pp. 3207-3220,

G. Petrillo, J. Zhuang. Bayesian earthquake forecasting approach based on the epidemic type aftershock sequence model. Earth Planets Space, 76 (2024), p. 78,

G. Petrillo, E. Lippiello. Testing of the foreshock hypothesis within an epidemic like description of seismicity. Geophys. J. Int., 225 (2) (2021), pp. 1236-1257,

G. Petrillo, E. Lippiello. Incorporating foreshocks in an epidemic-like description of seismic occurrence in Italy. Appl. Sci., 13 (8) (2023), p. 4891,

G. Petrillo, E. Lippiello, F.P. Landes, A. Rosso. The influence of the brittle-ductile transition zone on aftershock and foreshock occurrence. Nat. Commun., 11 (2020), p. 3010,

G. Petrillo, A. Rosso, E. Lippiello. Testing of the seismic gap hypothesis in a model with realistic earthquake statistics. J. Geophys. Res. Solid Earth, 127 (2022), Article e2021JB023542,

G. Petrillo, J. Zhuang. The debate on the earthquake magnitude correlations: A meta-analysis. Sci. Rep., 12 (1) (2022), p. 20683,

G. Petrillo, J. Zhuang. Verifying the magnitude dependence in earthquake occurrence. Phys. Rev. Lett., 131 (15) (2023), Article 154101,

P. Reasenberg. Second-order moment of central California seismicity, 1969–1982. J. Geophys. Res., 90 (B7) (1985), pp. 5479-5495,

P.A. Reasenberg, L.M. Jones. Earthquake Aftershocks: Update. Science, 265 (1994), pp. 1251-1252,

F. Romano, E. Trasatti, S. Lorito, C. Piromallo, A. Piatanesi, Y. Ito, D. Zhao, K. Hirata, P. Lanucara, M. Cocco. Structural control on the Tohoku earthquake rupture process investigated by 3D FEM, tsunami and geodetic data. Sci. Rep., 4 (2014), p. 5631,

G.J. Ross. Bayesian estimation of the ETAS model for earthquake occurrences. Bull. Seismol. Soc. Am., 111 (3) (2021), pp. 1473-1480,

J.G. Ross, A. Kolev. Semiparametric Bayesian forecasting of SpatioTemporal earthquake occurrences. Ann. Appl. Stat., 16 (4) (2022), pp. 2083-2100,

P.J. Rousseeuw, M. Hubert. Robust statistics for outlier detection. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 1 (1) (2011), pp. 73-79,

A. Saichev, D. Sornette. Distribution of the largest aftershocks in branching models of triggered seismicity: Theory of the universal Båth law. Phys. Rev. E., 71 (5) (2005), Article 056127,

V.M. Schimmenti, G. Petrillo, A. Rosso, F.P. Landes. Assessing the predicting power of GPS data for aftershocks forecasting. Seismol. Res. Lett., 95 (6) (2024), pp. 3243-3249,

R. Shcherbakov, K. Goda, A. Ivanian, G.M. Atkinson. Aftershock statistics of major subduction earthquakes. Bull. Seismol. Soc. Am., 103 (6) (2013), pp. 3222-3234,

R. Shcherbakov, J. Zhuang, G. Zöller, Y. Ogata. Forecasting the magnitude of the largest expected earthquake. Nat. Commun., 10 (2019), p. 4051,

I. Spassiani, G. Petrillo, J. Zhuang. Distribution related to all samples and extreme events in the ETAS cluster. Seismol. Res. Lett., 95 (6) (2024), pp. 3234-3242,

S. Stockman, D.J. Lawson, M.J. Werner. Forecasting the 2016–2017 Central Apennines earthquake sequence with a neural point process. Earth's Future, 11 (2023), Article e2023EF003777,

J.A. Sweets, R.M. Dawes, J. Monahan. Better decisions through science. Sci. Am., 283 (2000), pp. 82-87

K. Tamaribuchi. Evaluation of automatic hypocenter determination in the JMA unified catalog. Earth Planets Space, 70 (2018), p. 141,

K. Tamaribuchi, F. Hirose, A. Noda, Y. Iwasaki, K. Iwakiri, H. Ueno. Noise classification for the unified earthquake catalog using ensemble learning: the enhanced image of seismic activity along the Japan Trench by the S-net seafloor network. Earth Planets Space, 73, 91 (2021),

T. Ueda, A. Kato. Aftershocks following the 2011 Tohoku-Oki earthquake driven by both stress transfer and afterslip. Prog. Earth Planet. Sci., 10 (2023), p. 31,

R.A. Uhrhammer. Characteristics of northern and central California seismicity. Earthquake Notes, 57 (1986), pp. 21-37

T. Utsu. Catalog of large earthquakes in the region of Japan from 1885 through 1980. Bull. Earthq. Res. Inst., Tokyo Univ., 57 (1982), pp. 401-463 (in Japanese)

A. Van Horne, H. Sato, T. Ishiyama. Evolution of the Sea of Japan back-arc and some unsolved issues. Tectonophysics, 710–711 (2017), pp. 6-20,

A. Veen, F.P. Schoenberg. Estimation of space–time branching process models in seismology using an EM–type algorithm. J. Am. Stat. Asso., 103 (482) (2008), pp. 614-624,

D. Vere-Jones, J. Zhuang. Distribution of the largest event in the critical epidemic-type aftershock-sequence model. Phys. Rev. E., 78 (4) (2008), p. 7102,

I.A. Vorobieva, G.F. Panza. Prediction of the occurrence of related strong earthquakes in Italy. Pure Applied Geophys., 141 (1993), pp. 25-41,

Z. Wang, D. Zhao. 3D anisotropic structure of the Japan subduction zone. Sci. Adv., 7 (4) (2021), Article eabc9620,

S. Wiemer, M. Wyss. Minimum magnitude of completeness in earthquake catalogs: Examples from Alaska, the Western United States, and Japan. Bull. Seismol. Soc. Am., 90 (4) (2000), pp. 859-869,

S. Wiemer. A software package to analyze seismicity: ZMAP. Seismol. Res. Lett., 72 (3) (2001), pp. 373-382,

A. Witze. Risk of human-triggered earthquakes laid out in biggest-ever database. Nature, 2017 (2017), p. 22693,

J. Wu, J. Suppe, R. Lu, R. Kanda. Philippine Sea and East Asian plate tectonics since 52 Ma constrained by new subducted slab reconstruction methods. J. Geophys. Res. Solid Earth, 121 (6) (2016), pp. 4670-4741,

Y. Xu, R. Goodacre. On splitting training and validation set: A comparative study of cross-validation, bootstrap and systematic sampling for estimating the generalization performance of supervised learning. J. Anal. Test, 2 (2018), pp. 249-262,

J. Zhuang, Y. Ogata, D. Vere-Jones. Stochastic declustering of space-time earthquake occurrences. J. Am. Stat. Ass., 97 (458) (2002), pp. 369-380,

J. Zhuang, Y. Ogata. Properties of the probability distribution associated with the largest event in an earthquake cluster and their implications to foreshocks. Phys. Rev. E., 73 (4) (2006), p. 6134,

J. Zhuang, W.T. Ogata. Data completeness of the Kumamoto earthquake sequence in the JMA catalog and its influence on the estimation of the ETAS parameters. Earth Planets Space, 69 (2017), p. 36,

O. Zlydenko, G. Elidan, A. Hassidim, D. Kukliansky, Y. Matias, B. Meade, A. Molchanov, S. Nevo, Y. Bar-Sinai. A neural encoder for earthquake rate forecasting. Sci. Rep., 13 (2023), p. 12350,