ENSO Teleconnection to Global Seismicity — Replication Study

Author: TerraPulse Lab

Status: Complete

Created: 2026-03-22

Dataset: 138K earthquakes (2021–2026) + 913 monthly ONI values (1950–2026)

Abstract

We test the hypothesis that El Niño-Southern Oscillation (ENSO) phases modulate global earthquake rates through crustal loading from oceanic mass redistribution. Using 138,276 USGS M4.0+ earthquakes and 58 months of overlapping ONI data, we find no significant linear correlation between ONI values and monthly earthquake counts (Pearson , ). The chi-squared test shows significant rate differences between ENSO phases (, ), but this reflects La Niña having slightly lower rates — opposite to the crustal loading prediction. The magnitude distribution shows a small but statistically significant difference between El Niño and La Niña phases (K-S ), though the effect size is negligible (mean magnitude difference: 0.01). We find no evidence supporting the ENSO-seismicity teleconnection hypothesis.

Hypothesis

El Niño events redistribute oceanic mass — sea levels rise in the eastern Pacific and fall in the western Pacific. This redistribution could alter crustal loading, potentially modulating earthquake rates through:

1. Flexural stress changes in oceanic lithosphere
2. Pore pressure variations in subduction zones
3. Isostatic adjustment of continental margins

If valid, we would expect higher earthquake rates during El Niño () or a lagged response within 6–12 months.

Data Sources

Methodology

Phase Classification

ONI values classify ENSO state:

• El Niño: (12 months in overlap)
• Neutral: (25 months)
• La Niña: (21 months)

Statistical Tests

1. Pearson and Spearman correlation between monthly ONI and earthquake count
2. Lagged correlation (ONI leads earthquakes by 0–12 months)
3. Chi-squared test for rate equality across phases
4. Kolmogorov-Smirnov test on magnitude distributions by phase

Findings

No Linear Correlation

The correlation is effectively zero. ENSO phase has no measurable linear effect on earthquake frequency.

No Lagged Signal

No lag achieves statistical significance. The negative trend at 8–10 months is suggestive but — well above any reasonable threshold.

Rate Differences (Significant but Wrong Direction)

Chi-squared: ,

The rate differences are significant, but La Niña has the lowest rate — the opposite of what crustal loading theory predicts (it should be El Niño with more earthquakes from eastern Pacific sea level rise).

Magnitude Distribution (Statistically Different, Negligible Effect)

K-S test: El Niño vs La Niña, ,

The magnitude distributions are statistically different at , but the effect size is negligible:

• El Niño mean magnitude:
• La Niña mean magnitude:

A difference of 0.01 magnitude units is below measurement precision and has no physical significance.

Conclusion

The ENSO-seismicity teleconnection hypothesis is not supported by our data. Across 58 months of overlapping data with 138K M4+ earthquakes:

1. No linear correlation (Pearson )
2. No lagged correlation (all )
3. Rate differences exist but contradict the hypothesis direction
4. Magnitude distribution differences are negligible ()

This is consistent with the broader seismological consensus that earthquake occurrence is primarily governed by tectonic stress accumulation and release, not by surface mass loading from ENSO-driven sea level changes. The stress perturbations from oceanic mass redistribution (~10 cm sea level change) are several orders of magnitude below typical tectonic stress drops (~MPa).

Limitations

• 58 months of overlap limits statistical power for detecting small effects
• Only M4+ events analyzed; lower magnitudes might show different patterns
• No spatial stratification (Pacific Ring of Fire vs interior)
• ENSO effects may be masked by aftershock sequences

Visualizations

• Interactive Dashboard
• Static Figure