Machine Learning Reveals the State of Intermittent Frictional Dynamics in a Sheared Granular Fault

Seismogenic plate boundaries are posited to behave in a similar manner to a densely packed granular medium, where fault and block systems rapidly rearrange the distribution of forces within themselves, as particles do in slowly sheared granular systems. We use machine learning to show that statistical features of velocity signals from individual particles in a simulated sheared granular fault contain information regarding the instantaneous global state of intermittent frictional stick-slip dynamics. We demonstrate that combining features built from the signals of more particles can improve the accuracy of the global model and discuss the physical basis behind the decrease in error. We show that the statistical features such as median and higher moments of the signals that represent the particle displacement in the direction of shearing are among the best predictive features. Our work provides novel insights into the applications of machine learning in studying frictional processes occurring in geophysical systems. Plain Language Summary Records of previous earthquakes do not provide adequate data for scientists to predict future earthquakes with sufficient certainty. In this study, we use computer simulations representing earthquakes as frictional slips and record hundreds of scaled earthquakes. We employ machine learning, an artificial intelligence technique, to estimate the fault friction. In machine learning, the computer is trained to establish a relation between emitted seismic signals and fault friction. Subsequently, when the trained model is applied to new seismic data, it can accurately estimate the fault friction. The similarities between our model and field-scale observations from real faults suggest that an extension of our approach may have potential of estimating the friction of geological faults leading to prediction of real earthquakes.