Magnitude Variance Compression Before Large Earthquakes: A Unified Three-Force State Index with Prospective Validation, Nucleation Zone Localization, and a Proposed Monitoring Protocol for Japan and Turkey

Abstract Whether measurable statistical changes in earthquake catalogs precede large earthquakes remains one of the most debated questions in seismology. This study introduces and validates a unified seismic state index, Ψ = C × Q × (1 + S), that combines three independently measurable quantities—magnitude variance compression (C), seismicity rate quiescence (Q), and magnitude distribution skewness anomaly (S)—into a single dimensionless number characterizing the pre-seismic state of a fault zone. The method is built on a simple observation: the rolling standard deviation of earthquake magnitudes, computed within local spatial windows of 10–300 km around future mainshock epicenters, decreases systematically by 6–22% in the 7–21 days before large earthquakes. This suppression strengthens monotonically as the measurement radius decreases toward the epicenter—a spatial gradient that is the hallmark of a localized physical process rather than a statistical artifact. At near-field distances (10–50 km), a distinctive two-phase temporal pattern emerges: compression at 7–21 days followed by a variance increase at 1–3 days, consistent with aseismic preparation followed by foreshock-cascade onset. Using the JMA unified hypocenter catalog (2000–2023, 178,577 events, 288 isolated M ≥ 6.0 mainshocks) as the primary dataset, USGS ComCat data for Japan (46,515 events, 2000–2026) for consistency verification, and USGS ComCat data for Turkey (7,238 events, 2000–2026) for cross-regime validation, this study demonstrates: (1) walk-forward prospective testing with frozen 2000–2009 training parameters achieves AUC = 0.761 on 71 test-period mainshocks; (2) the Ψ index achieves AUC = 1.000 in discriminating detected from non-detected pre-earthquake states across eight events in two tectonic regimes; (3) a complementary lock-to-load ratio R = (C × Q)/(1 + E) independently achieves AUC = 1.000; (4) blind tests on post-catalog earthquakes detect the 2024 Noto M7.6 (703 fatalities, 95% pre-event compression), the 2025 Aomori M7.6, the 2020 Elazig M6.7, and the 2011 Van M7.1; (5) fine-resolution spatial scanning identifies future nucleation zones within 200 km in 50% of retrospective tests, including the 2011 Tohoku M9.1 at 62 km accuracy 14 months in advance. A two-force physical model is proposed: Force 1 (compression) arises from stress homogenization as aseismic loading synchronizes nearby faults; Force 2 (energy accumulation) arises from continued tectonic loading on the locked fault, manifesting as rightward skewness in the magnitude distribution. Rupture occurs when Force 2 overwhelms Force 1. The S4→S1 post-event reset observed universally confirms the physical reality of the pre-event compression. The method does not predict individual earthquakes; it identifies regions where the seismic system is simultaneously compressed, quiet, and energy-loaded. Current assessment identifies Van (Turkey, Ψ = 0.718), Erzincan (Turkey, Ψ = 0.644), and the Kanto corridor (Japan, Ψ = 0.45–0.59) as the highest-Ψ regions globally.