Pre-Seismic Quiescence Detected by K–R Critical Slowing-Down Indicators: Independent Replication in Japan and Chile Subduction Zone Catalogs

Why It Matters

The replicated CSD quiescence offers a new, statistically robust precursor that could enhance short‑term seismic risk assessment across major subduction zones.

Key Takeaways

• CSD₅₀ drops ~‑18% before M≥6.0 quakes
• Signal replicated in Japan and Chile subduction zones
• Variance reduction drives observed critical slowing‑down suppression
• No concurrent b‑value change; CSD distinct metric
• CSD₁₀₀ AUC 0.55, modest forecasting improvement

Pulse Analysis

Critical slowing‑down (CSD) has long been a theoretical hallmark of systems approaching a tipping point, but its practical detection in seismology has been elusive. The K‑R excitation‑regulation framework bridges this gap by converting rolling magnitude windows into ODE‑derived CSD metrics, enabling researchers to isolate subtle quiescence patterns that precede large earthquakes. By demonstrating statistically significant CSD₅₀ suppression in two independent subduction‑zone catalogs—Japan and Chile—the study moves beyond anecdotal evidence, offering a reproducible signal that survives rigorous false‑discovery rate and permutation testing.

The authors’ synthetic validation isolates variance reduction as the primary driver of the observed CSD suppression, with a combined rate‑and‑variance scenario matching real‑world effect sizes. This distinction is crucial because traditional seismicity metrics, such as the Gutenberg‑Richter b‑value, remained unchanged, indicating that CSD captures a different facet of the fault‑locking process. Rate‑and‑state friction simulations further corroborate the finding, showing pronounced CSD drops during locking phases. By confirming that the signal is temporally anchored and not an artifact of catalog filtering, the research strengthens the case for CSD as a complementary diagnostic tool in earthquake physics.

While the AUC of 0.55 for CSD₁₀₀ forecasts suggests modest predictive power, the real value lies in augmenting existing hazard models with an independent early‑warning indicator. The study’s transparent methodology and cross‑catalog replication set a benchmark for future operational testing, especially in regions with dense seismic monitoring. However, the authors rightly note limitations: the signal’s spatial universality and real‑time applicability remain unproven. Continued integration of CSD metrics with machine‑learning pipelines and high‑resolution fault simulations could eventually refine short‑term risk estimates, offering policymakers a more nuanced tool for emergency preparedness.