📄 earth_science

Immediate foreshock–mainshock arise from a paused model

By analyzing immediate foreshock–mainshock pairs in Japan, this study reveals that foreshocks exhibit reduced stress drops and spectral similarities to their mainshocks, supporting a "paused-rupture" model where both events represent successive phases of a single dynamic rupture process.

Original authors: Xueting Wei, Xiaofei Chen

Published 2026-08-05

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Xueting Wei, Xiaofei Chen

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Earth's crust as a giant, tangled ball of yarn, where the threads are massive fault lines. Sometimes, these threads get stuck, building up tension like a stretched rubber band. When they finally snap, the energy releases in a violent shudder we call an earthquake. Scientists have long been obsessed with the very first split-second of that snap: how does a tiny, invisible slip turn into a planet-shaking disaster? For decades, they've watched the "foreshocks"—the little tremors that sometimes happen right before the big one—trying to figure out if they are just random noise, a slow buildup of pressure, or a chain reaction where one small rock knocks over the next. Understanding this isn't just about satisfying curiosity; it's about knowing if the ground beneath us is just having a bad day or if it's about to let go completely.

Now, picture a group of scientists in Japan, a place so packed with earthquake sensors it's like having a microphone on every single street corner. They decided to zoom in on the most dramatic, split-second drama: pairs of earthquakes that happened less than a minute apart, often within the same tiny patch of ground. They were looking for a specific kind of "immediate foreshock" that behaves strangely. Usually, if a small earthquake happens right before a big one, you'd expect the big one to be a much bigger, louder version of the small one. But these scientists found something weird: sometimes, the little foreshock and the big mainshock sounded almost identical in their "pitch," even though the mainshock was 1 to 2 times larger in size. It's like hearing a whisper and a shout that sound like they come from the exact same throat, with the shout only happening a few seconds later.

To solve this mystery, the team used a clever trick called "relocation," which is like using a super-precise GPS to pin down exactly where the earthquakes happened. They found that in these strange pairs, the little foreshock and the big mainshock were practically hugging the same spot, sometimes just 64 meters apart. Then, they compared the "stress drop"—a measure of how much energy was released—of these foreshocks to other normal earthquakes of the same size. The result was shocking: these immediate foreshocks were releasing way less energy, about 3 to 10 times less than their normal neighbors. It was as if the fault tried to break, got stuck, and then, after a brief pause, finally snapped with full force.

This led the researchers to propose a new idea they call the "paused model." Imagine a car trying to drive up a steep hill. It revs its engine (the foreshock), starts to move, but then the wheels spin and it almost stops, hovering on the edge of failure. For a few seconds, it just sits there, trembling. Then, suddenly, it finds just enough grip to surge forward, accelerating into a full-speed run (the mainshock). In their computer simulations, the scientists recreated this exact scenario. They built a virtual fault and watched as a rupture started, slowed down to a near-halt, paused for a few seconds, and then re-accelerated into a massive, self-sustaining earthquake.

The paper suggests that these immediate foreshocks aren't separate events that trigger the big one; instead, they are the first, stuttering phase of the same single rupture. The fault didn't just "break" twice; it started to break, paused, and then finished the job. The authors note that this specific "pause-and-go" behavior seems to happen only under very specific conditions, which might explain why we don't see it in every earthquake sequence. While their computer models successfully reproduced the strange timing and energy levels they observed in Japan, this remains a strong suggestion based on simulations and data analysis, not a final, unchangeable law of nature. But it offers a fascinating new way to think about those terrifying seconds before the ground shakes: it might not be a countdown to a new event, but a single, dramatic hesitation before the real thing.

Technical Summary: Immediate Foreshock–Mainshock Arise from a Paused Model

Problem Statement
The physics of earthquake initiation remains a central unknown in seismology. While foreshocks provide a rare observational window into the onset of failure, the mechanisms governing "immediate" foreshocks—those occurring seconds before a mainshock within tens of meters of the hypocenter—are poorly understood. Conventional models struggle to explain these sequences: the "preslip" model suggests aseismic nucleation triggering small instabilities, while the "cascade" model posits neighbor-to-neighbor stress transfer. However, immediate foreshocks exhibit behaviors difficult to reconcile with these frameworks, such as multi-second pauses between events and the re-initiation of failure on the same fault patch with significantly different source characteristics.

Methodology
The authors utilized Japan's dense seismic networks (Hi-net and F-net) to analyze earthquake pairs from 2002 to 2022. The study focused on pairs with inter-event times $\le$ 60 s and hypocentral separations $\le$ 5 km.

Data Selection: 1,644 foreshock–mainshock pairs (where the later event is larger) and 1,585 consecutive pairs were identified.
Spectral Analysis: Apparent corner frequencies ( $f_{app}$ ) were estimated for S-waves using waveforms from the three nearest F-net stations. The study compared the ratio of mainshock to foreshock corner frequencies against magnitude differences.
Relocation and Stress Drop Estimation: Four representative pairs were selected for high-precision relocation using the double-difference algorithm (HypoDD). Empirical Green's Function (EGF) analysis was employed to estimate stress drops. This involved comparing immediate foreshocks with nearby isolated events of similar magnitude, using a shared, smaller-magnitude EGF to minimize path effects and inversion bias.
Dynamic Simulations: To explain the observations, the authors developed a "paused-rupture" model using 3D dynamic rupture simulations based on the boundary integral equation method (BIEM) with rate-and-state friction. The model featured a circular velocity-weakening patch embedded in a velocity-strengthening medium to simulate rupture arrest and re-initiation.

Key Results

Weak Scaling of Time and Magnitude: Unlike consecutive pairs, where inter-event time scales linearly with magnitude difference, immediate foreshock–mainshock intervals show only a weak positive association with magnitude difference, clustering predominantly in the 0–10 s range regardless of magnitude contrast.
Spectral Anomalies: Some immediate foreshocks share apparent corner frequencies with their mainshocks despite magnitude contrasts of 1–2. This implies a departure from constant stress drop scaling.
Spatial Coincidence: Relocations of anomalous pairs (those with high corner-frequency ratios) revealed that the foreshock and mainshock hypocenters nearly coincide (separations of 465 m and 64 m), with the foreshock rupture area lying within the mainshock rupture area.
Reduced Foreshock Stress Drops: EGF analysis of representative pairs showed that immediate foreshocks have stress drops 3 to 10 times lower than nearby isolated events of comparable magnitude. For example, in one pair, the foreshock stress drop was 0.043 MPa compared to 0.247 MPa for a similar isolated event.
Paused Rupture Simulation: The dynamic simulations successfully reproduced the observed sequence. A rupture initiated, decelerated, and nearly arrested (producing a low-stress-drop "foreshock" phase), followed by a brief pause (approx. 7 s in the model) before re-accelerating into a self-sustaining, higher-stress-drop "mainshock" rupture. The model generated two distinct moment-rate peaks separated by a near-zero slip interval.

Key Contributions
The paper proposes a "paused model" for immediate foreshock–mainshock sequences, suggesting that these events are not distinct triggered earthquakes but rather successive phases of a single dynamic rupture process.

Mechanism Identification: The study identifies a rupture-type transition from a "slow self-arresting" phase to a "sub-Rayleigh" rupture, punctuated by a brief pause.
Physical Explanation: This model explains the anomalous low stress drops of immediate foreshocks and their spectral similarity to mainshocks without invoking extreme fault heterogeneity or complex fluid-flow transients.
Unification of Observations: It reconciles the short time intervals, spatial coincidence, and distinct source parameters (magnitude vs. stress drop) that conventional preslip or cascade models struggle to accommodate simultaneously.

Significance
The authors claim that immediate foreshocks may constitute the initial, partially completed slip episode of the mainshock itself. This challenges the view that foreshocks are merely precursors or independent triggering events. Instead, the onset of the mainshock may already constitute a dynamic slip event that temporarily arrests before re-initiating. This framework suggests that the size of the mainshock is governed by the location of strength barriers or velocity-strengthening boundaries, while the initial foreshock scales with the nucleation zone. The findings imply that the transition between rupture types is a critical mechanism in earthquake initiation, occurring when frictional parameters lie near a critical boundary between slow self-arresting and sub-Rayleigh regimes.

More like this