← Back to Sessions

Complex Multi-Scale Preparatory Processes of Large Stick-Slip Events in Laboratory Experiments

Description: 

Earthquake forecasting relies on identification of distinctive patterns of precursory parameters that might precede large earthquakes. Laboratory stick-slip experiments provide an analog of the seismic cycle with dynamic slip and Acoustic Emissions (AE) characterizing brittle failure events before, during and after large lab earthquakes in controlled conditions. Recent laboratory studies showed that the ML/AI techniques may open new avenues towards laboratory earthquake prediction on smooth faults. However, not enough work with deep learning techniques has been yet done on rough and heterogeneous faults.

In this study laboratory stick-slip experiments were performed in triaxial pressure vessel on Westerly Granite samples at constant displacement rate on rough pre-fractured fault. The experiments produced complex slip patterns including fast and slow, large and small slips of the fault surface, and confined slips producing AE data bursts without externally measurable slip. Preparatory and nucleation processes were tracked with an ensemble of 16 features characterizing damage and stress evolution, localization and cluster processes, earthquake interactions, and local micromechanics and stress heterogeneity. At different experimental boundary conditions, we decompose observable complex spatio-temporal trends in features identifying effects of persistent fault roughness, a cross-play of local and global damage, and local stress evolution approaching system-size events. The observed trends highlight localization processes, and the nucleation of the system size events is masked or governed by interaction of different spatial scales during the preparatory process, which are also affected by the boundary conditions. The results provide a pool of reliable, physics-based characteristics that signify processes leading to system-size earthquakes, which could be used in development of ML/AI aided earthquake forecasting.

Session: Earthquake Preparation Across Scales: Reconciling Geophysical Observations With Laboratory and Theory

Type: Oral

Date: 4/20/2023

Presentation Time: 08:15 AM (local time)

Presenting Author: Grzegorz Kwiatek

Student Presenter: No

Invited Presentation: 

Authors