Modeling radon time series on the North Anatolian Fault Zone, Turkiye: Fourier transforms and Monte Carlo simulations

Time series studies depend mostly on stochastic models for radon seasonal, annual or temporal variability explanations. Others solve radon transport steady state equation analytically to explain radon variability with soil depth. In order to understand radon variability comprehensively, there is the need for a model which encapsulates most required information about its seasonal variability from different aspects. In this paper, soil radon time series is modeled on the North Anatolian Fault Zone in Turkiye. The general PDE representing radon soil dynamics and boundary conditions is employed successfully through a hybrid regression model, which captures and forecasts radon seasonal anomaly as well as its depth profile. An efficient model can be used to estimate other radon-related variables like diffusion rate and velocity. The model evaluation criteria facilitated the forecast of almost 86% variation of radon concentration with an RMSE value of 9.4 Bqm−3, which is reasonable considering the nature of data used. This model is simple and can provide a realistic statistical outcome on any radon data. The relationship between Rn and soil temperature was also investigated. Radon seasonal anomaly is observed to attain its maximum and minimum values in summer and winter seasons. A solid correlation is obtained between radon and soil temperature at various depths. The radon anomaly in normal conditions is found to correlate strongly with the model. Monte Carlo simulation procedure is affected by taking the mean of 300 simulation paths within $$\pm \,2\sigma$$ ± 2 σ from the regression curve with practically acceptable results.