(2000–2025, Mc = 5.0)
(Guerrero–Michoacán interface)
within 100 km of Saladita
2000s vs 2020s
La Saladita sits roughly 50 km from the Guerrero subduction interface, where the Cocos plate dives under North America at about 6 cm/year. That convergence has been generating large earthquakes for as long as the record goes back. This analysis fits Gutenberg-Richter scaling to the modern USGS catalog, reconstructs the historical M7+ recurrence across the broader Guerrero–Michoacán arc, and derives a first-order Poisson hazard estimate for the immediate area.
What follows is exploratory. The data are real and the methods are standard, but the catalog is uneven, the magnitude bins are coarse, and no formal PSHA (Probabilistic Seismic Hazard Analysis) has been attempted. Read the caveats section. The honest answer to "how dangerous is the ground here?" is: more active than most coastal areas people surf at, and the instrumental record is short relative to the fault's memory.
The magnitude distribution is normal for a subduction zone.
Gutenberg-Richter law states that log10(N) = a − b·M, where N is the number of earthquakes above magnitude M. The b-value encodes the relative frequency of small vs. large events. A lower b means proportionally more large events. Global subduction zones average b ≈ 1.0.
b = 1.094 ± 0.070 (95% CI: 0.96–1.23). Consistent with the global subduction-zone average.
Fitted by Aki (1965) maximum-likelihood estimator on 245 M≥5 events from the 2000–2025 USGS catalog. Magnitude of completeness set conservatively at Mc = 5.0 to avoid M3–4 detection bias from network coverage changes.
Decade-by-decade fit shows no significant trend. CIs overlap at every window:
No statistically significant b-value trend. The 2020s window is too short to interpret independently.
A systematically decreasing b over time would be one signal of strain accumulation toward a large rupture — but 25 years of data and n = 39–104 per decade cannot detect that reliably. Decade CIs all overlap. The current b is consistent with a normally stressed subduction zone, not with anomalous quiescence or precursory strain loading detectable by this method.
The interface has ruptured M7+ every 6 to 17 years.
The Guerrero seismic gap — roughly the segment between Acapulco and the Michoacán border — is one of the most-studied subduction gaps in the world. It has not produced a gap-spanning M8+ since 1911. Neighboring segments (Michoacán, Oaxaca) have ruptured M7.5–8.1 in the meantime. The timeline below shows M7+ events within the broader Guerrero–Michoacán box (16–18.5°N, 103–99°W), which is the relevant hazard context for Saladita.
11 major rupture clusters since 1909. Mean recurrence: 11 years. Elapsed since last: 4 years (2022 Aguililla M7.6).
The most recent event — the 2022 M7.6 near Aguililla, Michoacán — ruptured the western end of the interface. The 2014 M7.2 near Coyuquilla Norte (53 km from Saladita) was the closest large rupture in the modern record. Elapsed time since the last event is short relative to mean recurrence, but recurrence models on individual fault segments carry large uncertainty.
The central Guerrero segment — Saladita's immediate neighbor — has not produced a gap-spanning M8+ in over 100 years.
The "Guerrero gap" refers specifically to the segment between roughly Papanoa and Acapulco that has been anomalously quiet for interface-type megathrust ruptures since 1911. Episodic slow-slip events (2006, 2009–10, 2014) release strain aseismically, which some researchers argue delays a large rupture; others argue it partially loads adjacent locked portions. There is no consensus prediction. The gap has been "overdue" by simple recurrence models for decades.
The 2014 Guerrero SSE (slow-slip event) was one of the largest ever recorded globally, with Mw equivalent ~7.5 in geodetic moment, though it produced only scattered M5 seismicity. Whether SSEs deplete the seismogenic potential of the central Guerrero gap or simply redistribute it to adjacent locked segments is an active research question.
The M5+ rate has been stable but sequence-driven.
Annual M5+ event counts within 500 km, averaged by decade. The early 2000s peak reflects dense aftershock sequences following the 2002 Nuxco M6.8 and the 2003 Colima M7.6, not a background rate change.
No secular increase or decrease in M5+ rate. Variation tracks major aftershock sequences, not background stress change.
The 2020–2025 shortfall is most likely a quiet inter-sequence period following the 2021 and 2022 events, not a signal of anomalous quiescence. Sequence clustering is the dominant driver of year-to-year variation in this catalog.
Roughly 1-in-9 annual odds of a significant nearby earthquake.
Eight M≥6.5 events occurred within 100 km of Saladita between 1960 and 2025 (66 years). Under a Poisson model — stationary rate, independent events — the implied probabilities are:
The 8 events within 100 km since 1960:
11% annual probability is background rate. It does not mean an earthquake is "coming" — or not coming.
A Poisson process has no memory. The 4-year gap since the 2022 Aguililla M7.6 does not itself make the next year more or less likely than any other. The 11% figure is the best single-number summary of long-run frequency given the observed catalog, under the assumption that the rate is stationary. Slow-slip events, coupling changes, and sequence structure all violate that assumption to varying degrees. For design and planning purposes, the UNAM/CFE design spectra for this region encode far more complete hazard information.
b-value uses binned magnitude proxies, not individual event records. The 2000–2025 JS artifact provides annual counts by magnitude class (M3+, M5+, M6+), not per-event magnitudes. The Aki MLE here uses midpoint magnitudes within each bin as a proxy. This approximation is adequate for a first-order estimate but will slightly smooth the frequency-magnitude relation. b-value is also sensitive to the choice of Mc; a lower Mc would draw in incomplete M3–4 counts and distort the fit upward.
Guerrero gap event list is location-filtered only. Many events in the list are intraslab (deep) rather than interface-type ruptures. The 1964 M7.3 at 93 km depth, for example, is not a thrust-interface rupture in the traditional sense. Focal mechanism data are available for modern events but were not incorporated. The recurrence intervals above mix interface and intraslab events within the geographic box.
Pre-1960 catalog completeness is uncertain for M<7. The 1943 M7.4 near Palos Blancos (22 km from Saladita) is included in the gap timeline but was excluded from the Poisson rate estimate, which uses only 1960–2025 data. Events below M7 in 1900–1960 are likely underrepresented; this makes the historical recurrence list incomplete at the lower-magnitude end.
Poisson model is a simplification. Earthquake sequences are clustered, not Poisson-distributed. After a large event, short-term probability temporarily rises (Omori's law aftershock decay); between major sequences it may be lower than the long-run average. The 11% annual figure is a long-run average only.
This is not a design-basis hazard assessment. For construction, infrastructure, or emergency planning, use the official UNAM seismic hazard maps and Mexico City CFE design spectra. This analysis is exploratory — it describes what the data say in plain terms, not what an engineer should use to size a foundation.
b ≈ 1.09 — normal subduction regime. M7+ events recur across this interface roughly every decade. Poisson model: ~11% annual chance of M≥6.5 within 100 km. No anomalous trend detectable in 25 years of modern data. The Guerrero gap remains geologically unresolved.Synthesis · June 2026 · Exploratory analysis — not a hazard map
Sources and methods
Catalog: USGS FDSNWS Earthquake Catalog (fdsnws/event/1/query). M3.0+ events 2000–2025 within 500 km of 17.5897°N, 101.4317°W: 6,588 events. M5.5+ events 1900–1999, same radius: 233 events. Haversine distance used for radius filtering.
b-value: Aki (1965) MLE — b = log10(e) / (mean(M) − Mc). Mc = 5.0 chosen conservatively; maximum-curvature method on binned data would place Mc at the frequency peak but M3–4 detection is network-dependent and biases the curvature estimate. Magnitude inputs are bin midpoints (M5.35 for M5–6 class, M6.5 for M6+ class) — an acknowledged approximation. σb = b / √n (Aki 1965). 95% CI = b ± 1.96σ.
Guerrero gap recurrence: M≥7.0 events within the box 16–18.5°N, 103–99°W, 1909–2025, from combined historical and modern catalog. No focal-mechanism filter applied. Closely spaced sequence pairs (1911 doublet, 1985 main + aftershock) treated as single sequences for the inter-event interval calculation.
Poisson hazard: λ = n(M≥6.5, within 100 km) / T where T = 66 years (1960–2025). P(at least one event in t years) = 1 − exp(−λt). Pre-1960 events excluded for completeness. Scripts: scripts/analyze_seismic_hazard.py → functions/api/_findings_seismic_hazard.js.
Analysis run: June 2026