Seismic imaging in the laboratory: Reframing fault stability using elastic-wave observables

A growing body of geophysical observations, numerical simulations, and theoretical studies indicates that the evolution of the internal structure of fault zones strongly influences fault slip behavior. However, the theoretical frameworks most commonly used to describe fault stability—such as rate-and-state friction—were originally formulated to represent frictional sliding at idealized laboratory interfaces, in which the fault is treated as an effectively two-dimensional boundary with implicitly prescribed contact-scale processes. As a result, these models do not explicitly account for the space–time evolution of fault-zone structure, including damage, granular reorganization, and fluid-mediated processes. Moreover, the state variables used to represent contact evolution are phenomenological and are only weakly constrained by seismological observations, limiting the ability of these formulations to be rigorously applied across spatial and temporal scales.

Here, we propose an experimentally derived theoretical framework that reformulates fault stability in terms of internal variables directly linked to elastic-wave observables. Using double direct shear experiments on gouge layers under controlled boundary conditions, we combine mechanical measurements with active ultrasonic probing. Full waveform inversion is employed to reconstruct one-dimensional profiles of shear modulus and attenuation across the entire sample during normal and shear loading, stable sliding, and stick–slip events.

Ultrasonic waves induce only a small perturbation in strain and therefore probe the linearized constitutive response of the system without modifying its internal state. In this context, effective elastic stiffness and attenuation can be treated as internal variables that encode the evolving fabric and organization of the fault zone. The inverted profiles reveal spatially localized regions within the gouge where elastic properties evolve during slip instabilities, enabling a data-driven identification of the dynamically active fault region, distinct from the mechanically inactive surrounding material.

Based on these observations, we reframe the classical stiffness competition problem that defines the criteria for slip instability entirely in terms of observable quantities. Specifically, we propose to substitute the phenomenological state variable with the retrieved effective viscoelastic properties. Because elastic wave propagation obeys the same governing equations across laboratory and geophysical scales, this framework provides a physically grounded pathway for connecting laboratory experiments, numerical models, and seismological imaging of natural faults. More broadly, it represents a step toward a theory of fault mechanics grounded in seismological observables and geologically relevant fault-zone structures.