[M-IS04] Interdisciplinary studies on pre-earthquake processes

[MIS04-03] ULF seismo-magnetic signal extraction and evaluation

*Hongyan Chen¹、Peng Han¹、Katsumi Hattori^2,3 (1.Southern University of Science and Technology, Shenzhen, China、2.Graduate School of Science, Chiba University, Chiba, Japan、3.Center for Environmental Remote Sensing, Chiba University, Chiba, Japan)

キーワード：ULF seismo-magnetic changes, Signal extraction and evaluation, Time-series analysis, Forecasting and prediction

The early research on the anomaly of seismo-magnetic signals mainly focused on the spectrum analysis of the original data of the station, and then PCA and other methods were developed. In recent years, with the application of far reference stations, the anomaly index of P value for the ratio of the observed data energy to its model energy has been established, and the anomaly index P value is dimensionless. It is worth thinking about whether the actual abnormal waveform can be obtained. However, the seismo-magnetic signal is weak, it is usually mixed with the space sources' signal and human noises, so the critical problem is distinguishing the seismo-magnetic signal from the complex environment. It is a potential method to extract the signal of observation stations by using the reference station data based on the interstation transfer function. In view of the previous studies, which did not use strict multivariate coherence analysis for the transfer function, this study proposed an interstation transfer function algorithm based on wavelet multivariate coherence to establish the electromagnetic estimated values of observation stations, and the transfer function in the high-frequency part can also be accurately calculated. The phase difference is estimated well and the actual waveform data of the seismo-magnetic signal can be obtained.
For Kakioka (KAK) station, the data from the KNY and MMB stations are used as reference stations to calculate the interstation transfer function and obtain the estimated values of the three components of the KAK station. By showing the geomagnetic pulsation caused by apparent global disturbance at both the observed and reference stations, the geomagnetic pulsation signal disappears in the residual between the observed and the estimated value. By adopting the SEA method, the anomalies of the seismo-magnetic signal before the earthquake show significance. On the basis of correlation analysis, the Molchan diagram was used to evaluate the precursor information in the observed data. The residual values between vertical observed and estimated values are significantly increased in the total residual, the 50s, 100s, 200s, and 400s periodic of the total residual 6-15 days before the earthquake. The forecasting efficiency of the above parameters is better than that of the random prediction, and the forecasting efficiency of the total residual and 400s is the best.