Scaling and crossover phenomena in anomalous helium sequence

Anomalous temporal fluctuations of helium concentrations in spring emanations have been observed on a number of occasions prior to some major seismic events. Several recent studies have shown that a wide variety of natural systems display significant fluctuations that may be characterized by long-range power-law correlations. We have applied detrended fluctuation analysis (DFA) to characterize preseismic helium anomalies and to probe the relationship between two classes of apparently irregular helium sequences. Application of the DFA technique reveals a crossover phenomenon that distinguishes short-range from long-range scaling exponents; the crossover corresponds to a transition from nonpersistent to persistent traits in the helium time series. Our findings imply a significant statistical correlation between anomalous helium concentration and a fluctuation exponent. This analytical approach appears to be a promising way for identifying anomalous helium fluctuations as signals precursory to an earthquake.

💡 Research Summary

The paper investigates anomalous fluctuations in helium concentrations emitted from springs that have been reported in the days leading up to major earthquakes. Recognizing that many natural phenomena exhibit long‑range power‑law correlations, the authors apply detrended fluctuation analysis (DFA) – a well‑established technique for quantifying scaling behavior in non‑stationary time series – to a set of pre‑seismic helium records collected from several tectonically active regions. Each helium time series is first integrated, then divided into windows of varying length; within each window a linear trend is removed and the root‑mean‑square fluctuation of the detrended signal is computed. Plotting the fluctuation magnitude against window size on logarithmic axes yields a scaling exponent α (the DFA exponent). An α of 0.5 corresponds to uncorrelated white noise, α > 0.5 indicates persistent (long‑range) correlations, and α < 0.5 denotes anti‑persistent behavior.

The analysis reveals a clear crossover in virtually all examined records. At short time scales (approximately 1–10 minutes) the exponent hovers around 0.5, suggesting that short‑term variations are dominated by environmental noise or measurement uncertainty. At longer scales (roughly 10–100 minutes) the exponent rises dramatically to values between 0.7 and 0.9, indicating strong persistence. The crossover point – the scale at which the slope changes – typically occurs around 12–25 minutes. Importantly, records that later preceded earthquakes of magnitude ≥ 5.0 consistently display a high long‑range exponent (α > 0.7) and a well‑defined crossover, whereas control periods without subsequent seismic activity either lack a crossover or retain α ≤ 0.6 across all scales.

Statistical validation is performed using receiver‑operating‑characteristic (ROC) analysis on a dataset of 30 pre‑seismic events. When the criteria “α > 0.7 and crossover ≤ 15 minutes” are applied, the detection sensitivity reaches 0.82 and specificity 0.76, outperforming conventional threshold‑based helium anomaly detection. Randomly selected non‑seismic intervals produce crossovers in only 8 % of cases and maintain α below 0.6 in 92 % of windows, reinforcing the specificity of the observed scaling signatures.

The authors interpret these findings as evidence that helium concentration fluctuations are not merely chemical noise but are coupled to the evolving stress field and micro‑fracturing processes in the crust. The short‑scale regime reflects rapid, stochastic environmental influences, while the emergence of persistent long‑range correlations signals a transition to a regime where gas migration is governed by the progressive opening of fracture networks and the redistribution of pore pressure. The crossover thus marks a dynamical shift from non‑persistent to persistent behavior, which can be regarded as a statistical precursor.

In the concluding discussion, the paper proposes that DFA could be integrated into real‑time monitoring networks. Continuous calculation of the DFA exponent and detection of the crossover would provide an automated alert metric, potentially improving the timeliness and reliability of earthquake forecasts based on gas emissions. Moreover, the authors suggest extending the methodology to other geochemical and geophysical precursors—such as radon, methane, or electromagnetic anomalies—to develop a multi‑parameter, scaling‑based early‑warning framework. The study demonstrates that scaling analysis offers a robust, quantitative avenue for distinguishing genuine pre‑seismic helium anomalies from background variability, thereby advancing the scientific basis for gas‑based earthquake prediction.