13:45 〜 14:00
[SSS06-01] Machine Learning and Deep Learning Predicts Meter-Scale Laboratory Earthquakes
★Invited Papers
キーワード：機械学習・深層学習、1mスケール巨大岩石実験における地震発生予測、数値シミュレーションを用いた岩石実験の再現、局所的なクリープや剪断応力の発展と予測可能性の関係
In recent years, there has been a growing interest in utilizing machine learning (ML) or deep learning (DL) to investigate the predictability of shear-slip failures, known as laboratory quakes, in centimeter-scale rock friction experiments. The ML/DL models are typically trained by the continuous acoustic emissions and successfully predict linear scale time-to-failure. However, the applicability of ML/DL to larger-scale laboratory quakes and natural earthquakes, where important timescales vary by orders of magnitude, remains uncertain. In this study, we apply an advanced ML/DL approaches to a meter-scale laboratory quake data, characterized by accelerating foreshock activity manifesting as increasing numbers of tiny acoustic emission events. Unlike the previous studies, we established catalog-based characterization of shear-slip and demonstrate that the trained models, relying on the sparse event catalog, can accurately predict the time-to-failure of meter-scale mainshocks, from tens of seconds to milliseconds before the upcoming main quakes. These timescales correspond to approximately decades down to weeks in the context of large earthquakes which is rarely addressed in the previous studies. By comparing our results with a dynamic model of shear failures that replicates the experimental data, we suggest that our method effectively tracks the evolution of shear stress on locally creeping fault, rather than macroscopic shear stress, indirectly through the acoustic emission events, and it enables ML/DL to predict both numerical and laboratory quakes. These findings provide critical insights into the role of local creep (slowly slipping) activity on account of fault heterogeneity and related seismic precursors that may facilitate short-term forecasting of earthquakes in nature.