Deep learning of aftershock patterns following large earthquakes

Abstract

Aftershocks are a response to changes in stress generated by large earthquakes and represent the most common observations of the triggering of earthquakes. The maximum magnitude of aftershocks and their temporal decay are well described by empirical laws (such as Bath's law and Omori's law), but explaining and forecasting the spatial distribution of aftershocks is more difficult. Coulomb failure stress change is perhaps the most widely used criterion to explain the spatial distributions of aftershocks, but its applicability has been disputed. Here we use a deep-learning approach to identify a static-stress-based criterion that forecasts aftershock locations without prior assumptions about fault orientation. We show that a neural network trained on more than 131,000 mainshock-aftershock pairs can predict the locations of aftershocks in an independent test dataset of more than 30,000 mainshock-aftershock pairs more accurately (area under curve of 0.849) than can classic Coulomb failure stress change (area under curve of 0.583). We find that the learned aftershock pattern is physically interpretable: the maximum change in shear stress, the von Mises yield criterion (a scaled version of the second invariant of the deviatoric stress-change tensor) and the sum of the absolute values of the independent components of the stress-change tensor each explain more than 98 per cent of the variance in the neural-network prediction. This machine-learning-driven insight provides improved forecasts of aftershock locations and identifies physical quantities that may control earthquake triggering during the most active part of the seismic cycle.

Machine Learning Can Predict the Timing and Size of Analog Earthquakes

“…Geodesy also yields useful information for understanding megathrust dynamics. The recent progress in seismic and geodetic identification of earthquake precursory phenomena is accompanied by a rapidly increasing number of studies where machine learning (ML) algorithms have been used for earthquake related problems, such as predicting laboratory earthquakes (Rouet-Leduc et al, 2017), estimating lab-scale fault friction (Rouet-Leduc, Hulbert, Bolton, et al, 2018), predicting GPS displacement rates associated to slow slip events (Rouet-Leduc, Hulbert, & Johnson, 2018), and forecast of aftershock locations (DeVries et al, 2018). Moreover, the recent identification of transients (i.e., interseismic accelerations and decelerations) in the geodetic time series prior to large earthquakes (e.g., Mavrommatis et al, 2014) suggests that the continental surface velocity is likely a good indicator of when a given portion of the megathrust is ready to fail.…”

Fault Slip Model of the 2018 Mw 6.6 Hokkaido Eastern Iburi, Japan, Earthquake Estimated from Satellite Radar and GPS Measurements

“…Felzer and Brodsky [51] inferred that aftershocks may be driven by a mix of both dynamic and static stress changes. Furthermore, shear stress changes and the invariants of the stress tensor could better explain the aftershock patterns than could the ∆CSC, as reported by several previous studies [52][53][54]. Thus, further research is also needed to reveal the dominant triggering mechanism of the aftershocks following the 2018 Hokkaido eastern Iburi event.…”

1.1 The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design

“…Other fields that have been improved by the use of deep learning-based approaches include quantum chemistry [Gilmer et al 2017], earthquake prediction [DeVries et al 2018], flood forecasting [Nevo 2019], genomics [Poplin et al 2018], protein folding [Evans et al 2018], high energy physics [Baldi et al 2014], and agriculture [Ramcharan et al 2017].…”