Predictive modelling of radon variations in time series using wavelets, multiple linear regression and ARIMA.

Radon-222 (222Rn) constitutes a naturally occurring radioactive gas that emerges as a fundamental byproduct of the uranium decay series within the Earth's lithosphere. Prior research has shown that this specific isotope functions as a versatile geophysical tracer for identifying clandestine underground faults and mapping complex geological formations across diverse terrains. Scientists frequently deploy these gas measurements during uranium surveys and as a primary metric for forecasting significant seismic events in tectonically active regions. The observation of distinct abnormalities in radon time-series (RTS) data has historically preceded major tectonic shifts, suggesting its potential as a reliable earthquake precursor. Climatological variables, including ambient temperature, atmospheric pressure, and relative humidity, exert significant influence on the exhalation and transport of these radioactive particles through porous soil and rock matrices. Understanding these environmental interactions is essential for distinguishing between meteorologically induced fluctuations and genuine geophysical signals originating from deep within the crust. This absence of evidence motivated a deeper investigation into the simulation techniques required to extract meaningful physical information from these multifaceted environmental datasets.

This investigation evaluates complex radon time-series (RTS) data alongside specific climatological factors to extract relevant physical information through advanced simulation techniques and mathematical modeling. The researchers sought to characterize the precise linear and non-linear interactions between radioactive gas concentrations and environmental variables like atmospheric pressure and humidity. Determining the most effective mathematical model for forecasting these variations over uncertain temporal periods remained a central objective of the scientific effort to improve predictive accuracy. The study aimed to refine the utility of these radioactive tracers as precursors for seismic activity to enhance disaster risk reduction and community resilience. Aligning these predictive capabilities with international safety standards for infrastructure and environmental health guided the development of the analytical framework used throughout the study. By identifying the optimal simulation method, the team intended to provide a robust tool for monitoring geological stability and atmospheric safety. The team focused on identifying a specific model that minimizes information loss while maximizing the accuracy of long-term concentration predictions in various environmental contexts.

The investigative process involved applying wavelet-based regression (WBR) to decompose the temporal data and analyze the behavior of the radioactive gas against temperature. Multiple linear regression (MLR) provided a secondary statistical framework for assessing the positive relationships between the tracer gas and environmental parameters like pressure and humidity. The researchers implemented the autoregressive integrated moving average (ARIMA) model to evaluate patterns, trends, and stationarity within the comprehensive radon time-series (RTS) dataset. Model selection relied on identifying the specific architecture that yielded the lowest values for the Akaike information criterion (AIC) and the Bayesian information criterion (BIC). These statistical metrics ensured the selection of a robust predictive framework capable of handling the complexities of extended temporal periods and fluctuating environmental conditions. The experimental design integrated temperature, pressure, and humidity as independent variables to simulate the real-world atmospheric interactions that govern gas release from the subsurface. This multi-model approach allowed for a comparative analysis of how different mathematical structures capture the nuances of radioactive gas dynamics.

The autoregressive integrated moving average (ARIMA) model outperformed alternative simulation techniques in predicting radioactive gas concentrations over extended durations with high statistical confidence. Wavelet-based regression (WBR) analysis revealed that the gas exhibits strictly linear behavior when correlated with ambient temperature variations across the studied timeframes. Non-linear behaviors became significantly more evident when the researchers analyzed the relationship between the tracer gas and both atmospheric pressure and relative humidity. Multiple linear regression (MLR) confirmed a distinct positive correlation between the gas levels and the measured atmospheric pressure and humidity within the temporal dataset. Significant anomalies in the temporal data appeared consistently before the occurrence of seismic events, validating the use of these tracers as precursors for tectonic activity. The identification of the lowest AIC and BIC values provided quantitative evidence for the superior predictive accuracy of the selected statistical model compared to linear alternatives. These results demonstrate that the chosen modeling approach effectively captures the complex interplay between geophysical signals and meteorological noise.

These findings enhance the reliability of using radioactive gas fluctuations as viable earthquake precursors for global disaster risk reduction and seismic resilience in vulnerable regions. The study demonstrates that advanced statistical modeling improves the understanding of environmental health factors, directly supporting the objectives of United Nations Sustainable Development Goal (SDG) 3. Implementing these predictive frameworks facilitates infrastructure safety by providing early warnings of geological instability, which aligns with the goals of SDG 9. Future disaster mitigation efforts can leverage these optimized models to enhance seismic resilience and reduce the impact of tectonic events on human populations. The integration of climatological variables into time-series analysis provides a more comprehensive view of the dynamics governing geophysical tracer transport in the atmosphere. This research contributes to the application of advanced technologies for infrastructure safety and the enhancement of disaster risk reduction strategies globally. The researchers conclude that the ARIMA model provides a superior framework for environmental monitoring and the early detection of seismic precursors.