Foreshock-induced slip transients set mainshock nucleation timing | Nature

Download PDF

Subjects

- Geophysics

- Seismology

- Surfaces, interfaces and thin films

Abstract

Foreshocks are sometimes observed before earthquakes1,2,3,4,5,6,7,8,9,10,11,12,13, yet their role in controlling rupture nucleation remains unclear1,11,14. Classical models often assume that nucleation arises from slow, quasi-static slip governed primarily by fault weakening15,16,17,18,19,20,21, typically neglecting impulsive precursory events. Here we show, using laboratory experiments and a rate-and-state-based Griffith-like rupture framework22, that foreshocks, when they occur at the onset of or during nucleation, can fundamentally regulate earthquake initiation. We find that the slip burst induced by foreshocks imparts a transient sliding velocity, Vmin, whose magnitude is set by foreshock size and which robustly predicts both nucleation duration and spatial length. Larger foreshocks generate higher Vmin and trigger a more rapid transition to dynamic rupture, whereas smaller foreshocks produce long-duration quasi-static growth and very small impulses lead to ruptures entirely arresting. Extending our theoretical framework to tectonic faults, we show that foreshock and associated slow-slip sequences preceding natural earthquakes seem to follow the same scaling. These observations allow us to constrain realistic characteristic nucleation slip distances of 0.3–3.0 mm, orders of magnitude smaller than those inferred for dynamic rupture23. Our results demonstrate that foreshock-induced transients set the timing and potential detectability of earthquake nucleation24.

Main

Understanding the nucleation of frictional ruptures is a central challenge in earthquake mechanics18. Beyond earthquake physics, nucleation processes are generic to frictional instabilities. Nucleation refers to the early processes that initiate fault rupture, marking the transition from a stable, locked fault to an unstable, dynamically slipping one. Coseismic rupture is theoretically expected to be preceded by quasi-static slip, growing in space and time until it reaches a critical size that may trigger dynamic failure15,16,17,18,19,20,21. This behaviour is observed in both rate-and-state16,20,22,25 and slip-weakening models15,17,19,26. These foundational studies suggest that the nucleation length, a critical measure of the size of the slipping zone before dynamic rupture, is primarily governed by the fracture energy of the fault interface, Gc, and the weakening rate within the nucleation region19. However, the physical parameters that control the duration and detectability of this stable phase remain poorly understood25,27. As a result, it is still unclear under which conditions geophysical observations might reveal precursory signals. Some earthquakes seem to be preceded by a prolonged nucleation stage2,3,4,6,8,24, potentially accompanied by foreshocks1,3,4,6,8,28, whereas many others exhibit no detectable dynamic precursor activity at all. These contrasting observations highlight the need for further investigation into the various physical mechanisms that may govern earthquake nucleation. Key open questions include: (1) whether all earthquakes exhibit quasi-static nucleation; (2) how foreshocks affect nucleation; and (3) whether the slip distance governing dynamic rupture also applies to nucleation.

We report laboratory experiments that, for the first time to our knowledge, directly capture how foreshocks control the nucleation of earthquakes. A small initial slip event sets the transient sliding velocity, which in turn dictates the duration and scale of the nucleation phase before dynamic rupture. Our results are supported by a theoretical framework based on a rate-dependent Griffith-like equation of motion (EoM), formulated using rate-and-state friction. Finally, we extend this theoretical framework to natural earthquakes preceded by foreshocks, enabling constraint of the characteristic nucleation parameters of large earthquakes, challenging standard estimates.

Two decades of nucleation duration

We conducted experiments using a biaxial loading apparatus (Fig. 1a; see Methods for details). Three representative nucleation phases from a single experiment are shown in Fig. 1b–e. Despite having nearly identical initial stress conditions (Fig. 1b), the duration of the nucleation phase varied by more than an order of magnitude, ranging from 0.6 to 12.7 ms (Fig. 1c–e), and reached maximum values of about 80 ms across the experimental suite, highlighting the intrinsic complexity of nucleation dynamics.

Fig. 1: Large variations in nucleation duration despite similar stress states.The alternative text for this image may have been generated using AI.Full size image

a, The experimental set-up of frictional stick-slip experiments. Two PMMA blocks (D) with lengths, heights and widths of (x, y, z) = (40, 10, 1) cm and (x, y, z) = (45, 10, 1.8) cm are contained in a steel sample holder (C). A normal load, FN, is first applied by three vertical h