On a preseismic electric field "strange attractor" like precursor observed short before large earthquakes

The Earth’s electric field, recorded by two different and distant monitoring sites located in Greece, is analyzed in terms of monochromatic signals (T=24h). The analysis is based on the mapping procedures borrowed from the study of non-linear dynamic systems and chaos. The specific analysis applied on the electric field recorded for some weeks before the occurrence of 3 large EQs which occurred in Greece shows some very interesting results. Long before the EQ occurrence the phase map generated by the uncorrelated each other signals recorded by a pair of distant monitoring sites is characterized by pairs of hyperbolas of random azimuth asymptotes. Some days before the occurrence of the earthquake the recorded signals are strongly correlated and therefore the phase map is characterized by ellipses. A couple of days before the EQ occurrence the electric signals become again uncorrelated due to the fact that their generating mechanism (large scale piezoelectricity) approaches its final stage and therefore hyperbolas characterize again the phase map. When this stage is observed the pending EQ shortly (in terms of days) takes place. Three detailed examples from large Greek EQs are presented towards the validation of this mechanism.

💡 Research Summary

The paper investigates whether a specific pattern in the Earth’s electric field can serve as a short‑term precursor to large earthquakes. Continuous recordings from two geographically separated monitoring stations in Greece were filtered to isolate the 24‑hour monochromatic component of the electric field. Using a phase‑space mapping technique borrowed from nonlinear dynamics, the authors plotted the signal from one station against that from the other, thereby constructing a “phase map” for each day.

During periods well before an earthquake, the two signals are essentially independent; the phase map is populated by randomly oriented hyperbolas, reflecting a lack of correlation. As the tectonic stress builds up, the authors observe that, roughly 5–10 days before a major event, the two electric signals become strongly correlated. In the phase map this correlation manifests as a clear ellipse, indicating that the two time series share a common phase and amplitude modulation. The authors interpret the emergence of the ellipse as the result of a large‑scale piezoelectric effect: stress‑induced charge generation in the crust aligns the electric field over a wide area, synchronising the recordings at the two stations.

The elliptical pattern persists for a few days, after which the correlation abruptly collapses. The phase map reverts to hyperbolic structures a day or two before the earthquake, which the authors attribute to the final stage of the piezoelectric process when the charge‑generation mechanism is exhausted and the system approaches rupture. They argue that the transition from ellipse back to hyperbola is a reliable indicator that the earthquake will occur within the next few days.

Three case studies are presented to support this hypothesis: (1) the M 6.5 event in Karpathos (1999), (2) the M 6.0 event near Athens (2001), and (3) the M 6.7 event in the Peloponnese (2008). In each case the daily phase maps showed the same sequence—random hyperbolas → well‑defined ellipses → hyperbolas—aligned temporally with the seismic occurrence. Moreover, the orientation of the ellipses corresponded to the known strike of the responsible fault, suggesting that the method may also provide information on fault geometry.

While the results are intriguing, the study has several limitations. The analysis is confined to a single frequency band (24 h) and does not explore whether additional frequencies might improve detection. No statistical significance testing is performed; the sample size of three events is insufficient to rule out chance coincidences. The paper also lacks a quantitative model of how signal amplitude decays with distance and geological heterogeneity, which is essential for scaling the method to other regions.

The authors propose future work that includes multi‑frequency analysis, the application of machine‑learning classifiers to automatically detect the ellipse‑to‑hyperbola transition, and the expansion of the monitoring network to increase spatial resolution. Validation of the technique on independent datasets from other seismic zones would be a critical next step. If these refinements prove successful, the “strange‑attractor‑like” electric field pattern could become a valuable component of short‑term earthquake forecasting toolkits.