Singular spectrum analysis of ionospheric anomalies preceding great earthquakes: Case studies of Kaikoura and Fukushima earthquakes

To study seismo-ionospheric coupling, the singular spectrum analysis (SSA) method was introduced to detect ionospheric anomalies preceding great earthquakes. The SSA method was used to decompose the total electronic content (TEC) series, and the reconstructed components (RCs) were combined to obtain the TEC principal components using w-correlation. The background TEC value during the quiet solar-terrestrial environment period was used to calculate the standard deviation, which was subsequently used to determine the upper and lower bounds of the background values. The root-mean-square error of the global ionosphere map (GIM) published by the International GNSS Service (IGS) near the earthquake epicenter was extracted and averaged to obtain the threshold, which was then used to suppress the systematic error effect during the detection of pre earthquake TEC anomalies process. This new method was used to detect ionospheric anomalies prior to the occurrence of the Mw7.8 Kaikoura earthquake in New Zealand on 13 November 2016 and the Mw6.9 Fukushima earthquake in Japan on 21 November 2016. The TEC data were further analysed with the sliding interquartile range (IQR) method to verify the reliability and feasibility of the SSA. The results showed that the TEC anomalies over these two epicenters were clearly detected on 25 and 28 October and 10 November. Our results showed that compared with IQR, SSA could detect TEC anomalies with a larger magnitude, making TEC anomalies caused by external factors more visible. Furthermore, based on the analysis of the solar-terrestrial environment using the data of F10.7, Dst and Kp indices, it is found that magnetic storms occurred during these three days. To identify and extract potential precursor information on seismo-ionospheric anomalies, the global distribution of TEC anomalous areas with SSA and IQR was also analysed. The results indicated the presence ofobvious anomalous characteristics of the seismic-ionospheric coupling effect in the preparation zones detected by SSA. In addition, the relation between TEC anomaly and Dst index detected by the cross-wavelet analysis indicated that ionospheric anomalies outside the preparation zones detected by IQR were caused by magnetic storms. In contrast, the ranges and amplitudes of TEC anomalies detected by SSA in these zones were much smaller than those with IQR. Overall, these findings demonstrate that SSA can effectively eliminate the effect of solar-geomagnetic activity and extract seismic-ionospheric coupling anomalies.