A MLP neural network as an investigator of TEC time series to detect seismo-ionospheric anomalies

Anomaly detection is extremely important for earthquake parameters estimation. In this paper, an application of Artificial Neural Networks (ANNs) in the earthquake precursor’s domain has been developed. This study is concerned with investigating the Total Electron Content (TEC) time series by using a Multi-Layer Perceptron (MLP) neural network to detect seismo-ionospheric anomalous variations induced by the powerful Tohoku earthquake of March 11, 2011.The duration of TEC time series dataset is 120 days at time resolution of 2 h. The results show that the MLP presents anomalies better than referenced and conventional methods such as Auto-Regressive Integrated Moving Average (ARIMA) technique. In this study, also the detected TEC anomalies using the proposed method, are compared to the previous results (Akhoondzadeh, 2012) dealing with the observed TEC anomalies by applying the mean, median, wavelet and Kalman filter methods. The MLP detected anomalies are similar to those detected using the previous methods applied on the same case study. The results indicate that a MLP feed-forward neural network can be a suitable non-parametric method to detect changes of a non linear time series such as variations of earthquake precursors.

► An application of Artificial Neural Networks (ANNs) in the earthquake precursor’s domain has been developed.
► Powerful Tohoku earthquake of March 11, 2011 has been selected as case study.
► The Total Electron Content (TEC) time series have been investigated by using a Multi-Layer Perceptron (MLP) neural network to detect seismo-ionospheric anomalous variations.
► A MLP neural network can be a suitable non-parametric method to detect unusual changes of earthquake precursors time series.