Singular spectral analysis and principal component analysis for signal discrimination of ULF geomagnetic data associated with 2000 Izu Island Earthquake Swarm

In order to extract any ULF signature associated with earthquakes, the singular spectral analysis (SSA) and the principal component analysis (PCA) have been performed to investigate a possibility of discrimination of signals from different sources (geomagnetic variation, artificial noise, and the other sources (earthquake-related ULF emissions)). We apply SSA to the time series data observed at closely spaced stations; Seikoshi (SKS), Mochikoshi (MCK), and Kamo (KAM) stations. Then, PCA is applied to the time series data sets filtered at 0.01 Hz of NS component at three stations. In order to remove the most intense signal like the first principal component, we make the differential data sets of SKS–MCK and MCK–KAM for the above data. The major findings are summarized as follows. (1) It is important to apply simultaneously SSA and PCA, because SSA gives the structure of signals and the number of sensors for PCA is estimated. This makes the results more convincing. (2) There is a significant advantage using PCA with differential data sets of filtered (0.01 Hz band) signal between SKS–KAM and MCK–KAM in NS component for removing the most intense signal like global variation (solar–terrestrial interaction). This yields that the anomalous changes in the second principal component were detected more clearly, and the contribution of the second principal component is found to be 20–40%. It is enough to prove mathematical accuracy of the signal. Further application is required to accumulate events. These facts demonstrate a possibility of monitoring the crustal activity with using the SSA and PCA.