Reduction Of Geomagnetic Effects (Periods T < 1000 s) From Geomagnetic And Geoelectrical Potential Difference Data

Electromagnetic phenomena preceding large earthquakes have been reported in various frequency ranges and they are considered as candidates for the short-term prediction. ULF electromagnetic phenomena are most promising among them because of the deep skin depth. In order to verify earthquake-related electromagnetic phenomena and clarify the possible physical mechanisms, a network observation has been established in Japan. The observed ULF magnetic and electric potential data are superposition of signals: (1)global signals originated from the external source field associated with the solar-terrestrial interaction like geomagnetic pulsations and their inductive field, which appears simultaneously in scale of hundreds of km, (2) regional (a few tens of km) signals such as artificial noises associated with the leakage current from DC-driven trains, and earthquake-related signals, (3) local (less than few kms) signals around the sensors. The signals associated with the crustal activity are very weak in general, and therefore the sophisticated signal separation is important. As for the ULF geomagnetic data, we have already developed effective methods such as polarization analysis, principal component analysis, and direction fractal analysis. These methods detect signal characteristics. In order to clarify physical mechanisms of earthquake-related ULF signals, time series analysis to identify waveform is required. In this aim, we developed the method for elimination of the most intense external source fields originated from solar-terrestrial interaction. It is interstation transfer function (ISTF) method. The ISTF method is based on the correlation between a site and a quiet remote reference station. Once interstation transfer function is estimated with high accuracy, which is considered to be invariant in time, it is possible to estimate the ideal external source field and their inductive variations at the site. Therefore, the resudials between the observed and estimated variations at the site provide only regional and local variations and earthquake-related phenomena are included. Generally, ISTF approaches were based on the conventional Fourier transform, but we use the wavelet transform due to existence of transient changes. In this paper, the proposed method has been applied to the data observed at the sites in Boso Peninsula, Japan and we use geomagnetic data observed at Kakioka Magnetic Observatory, Japan Meteorological Agency, as the remote reference data. We applied the proposed method to the geomagnetic and geoelectrical potential data to evaluate the elimination of global geomagnetic changes and their inductive fields in the frequency range of 0.001 to 1 Hz. The obtained performance is found to be quite good and changes associated with geomagnetic pulsations are effectively eliminated.