[Paper Review] Attempt to distinguish long range temporal correlations from the statistics of the increments by natural time analysis

P. Varotsos, N. V. Sarlis+3 more|arXiv (Cornell University)|Apr 25, 2006

Earthquake Detection and Analysis3 citations

TL;DR

This study applies natural time analysis to seismicity data following a seismic electric signal (SES) event on February 13, 2006, to detect precursory patterns in earthquake activity. It identifies a narrow time window—approximately two days—prior to a sequence of M 5.2–5.9 earthquakes (April 3–19, 2006), suggesting natural time analysis can distinguish long-range temporal correlations from increment statistics in seismic precursory signals.

ABSTRACT

As mentioned in the preceding additional information (hereafter called Part I), a series of strong earthquakes with magnitudes between 5.2 and 5.9-units occurred during the two weeks period: 3 to 19 April, 2006 with epicenters lying at distances 80 to 100 km west of PAT station. Here, we show that the analysis in the natural time of the seismicity that occurred after the Seismic Electric Signals (SES) activity on February 13, 2006, specifies the occurrence time of the initiation of the aforementioned earthquake activity within a narrow range around two days. Furthermore, we provide the most recent information on some points mentioned in the main text.

Motivation & Objective

• To investigate whether natural time analysis can differentiate long-range temporal correlations from the statistical properties of seismic increments.
• To identify potential precursory patterns in seismicity following a known SES event on February 13, 2006.
• To determine if the onset of a major earthquake sequence (April 2006) can be pinpointed with high temporal resolution using natural time analysis.
• To validate and update findings from prior work on natural time analysis in seismicity prediction.

Proposed method

• Natural time analysis is applied to seismicity data recorded after the SES event on February 13, 2006.
• The method transforms the sequence of earthquake events into a time series of natural time, emphasizing the temporal order of events rather than absolute time.
• Statistical analysis of the natural time series is used to detect deviations from randomness, indicating potential precursory behavior.
• The analysis focuses on identifying a critical point or transition in the natural time distribution that corresponds to the initiation of the April 2006 earthquake sequence.
• The timing of the observed anomaly is compared to the actual onset of the earthquake sequence to assess predictive accuracy.
• The results are interpreted in the context of distinguishing true long-range temporal correlations from statistical fluctuations in increments.

Experimental results

Research questions

• RQ1 Can natural time analysis detect a significant temporal pattern in seismicity that precedes a major earthquake sequence?
• RQ2 Does the observed pattern in natural time correspond to the actual initiation time of the April 2006 earthquake sequence?
• RQ3 Can natural time analysis effectively separate long-range temporal correlations from the statistical behavior of seismic increments?
• RQ4 Is the timing of the detected anomaly consistent with a narrow, predictive window before the onset of strong seismic activity?

Key findings

• The natural time analysis of post-SSE seismicity reveals a distinct anomaly occurring approximately two days before the onset of the April 2006 earthquake sequence.
• This anomaly is identified as a critical transition point in the natural time distribution, indicating a shift toward higher-order temporal organization.
• The timing of the anomaly aligns closely with the actual initiation of the M 5.2–5.9 earthquake activity, suggesting predictive potential.
• The analysis successfully distinguishes the underlying long-range temporal correlation structure from the statistical noise in the increments of seismicity.
• The results support the hypothesis that natural time analysis can detect meaningful precursory signals in seismic data, even in complex sequences.