Investigating radon and TEC anomalies relative to earthquakes via AI models

Abstract

Simultaneously studying earthquakes, Radon gas (Rn), and the Ionospheric total electron content (TEC) provides insight into the Lithosphere-Atmosphere-Ionosphere coupling (LAIC) phenomenon. The potential responses of Rn and TEC to earthquakes along the North Anatolian Fault (NAF) from 2007 to 2009 were examined using statistical and machine/deep learning methods, such as ARIMA, LSTMs, ANNs, SVMs, and STUMPY. The study was divided into four regions and the best performing models were used for anomaly detection. Results showed that the correlation between Radon-TEC changes and earthquakes does not always occur, but Rn anomalies are followed by TEC anomalies relative to earthquakes. Additionally, Rn variations were found to be more sensitive to seismic activity than TEC variations.