Nonlinear dynamics and predictability limits of indoor radon under temperature and humidity variability

Indoor radon variability is shaped by coupled physical processes that limit short-term predictability and challenge conventional statistical modeling. We analyzed high-resolution 10-min indoor radon measurements (4842 samples, October-November 2023) at the Centre de Recherche Nucléaire d'Alger, alongside simultaneous indoor temperature and relative humidity records. Nonlinear time-series metrics, including Kolmogorov complexity (KC), the Kolmogorov complexity spectrum (KCS), and the largest Lyapunov exponent (LLE), were applied to characterize dynamical structure. The KC spectrum indicates that radon exhibits non-zero algorithmic complexity over a broader range of normalized amplitude thresholds than temperature or humidity, reflecting enhanced multiscale structural heterogeneity. Radon also shows a larger LLE (λ = 0.0826 per 10-min interval; 95% CI: 0.0728-0.0924), corresponding to a characteristic Lyapunov divergence timescale ∼ 2 h. Because the observations were obtained from a single indoor location over approximately one month, this estimated predictability timescale should not be generalized to indoor radon dynamics beyond the analyzed conditions. By comparison, temperature and humidity display lower dynamical instability and longer predictability horizons. These findings demonstrate that indoor radon has stronger short-term instability and reduced deterministic predictability relative to microclimatic variables. The results provide a quantitative framework for assessing structural complexity and intrinsic forecast limits in radon time series, offering practical insights for predictive modeling and ventilation-based mitigation strategies.