13:45 〜 15:15
[MIS04-P03] Further Investigation on the ULF geomagnetic changes as the Earthquake Precursor, at Kakioka, Kanto, Japan
キーワード：ULF電磁波現象、スーパーポーズド・エポック解析、モルチャンのエラーダイアグラム
To mitigate earthquake disasters, the establishment of the short-term earthquake forecast technique is important. To achieve this, the ultra-low frequency (ULF) geomagnetic field changes has been studied and statistical significance in correlation and precursor characteristics in ROC analysis have been presented, It means that the ULF geomagnetic measurement has a potential parameter for the short-term earthquake forecast. In previous studies, the data observed Kakioka station during 2001-2010 operated by JMA has been utilized. In order to clarify ULF seismo-magnetic phenomena, we have studied the geomagnetic data observed at the Kakioka (KAK) station, Kanto, Japan, during 2015-2020. The magnetic data and earthquake catalogs obtained from Japan Meteorological Agency. The same approach with the previous studies have been performed, To select earthquakes, we use the Es parameter which considers the magnitude and distance of an earthquake simultaneously to select statistical samples. As the frequency of original geomagnetic data is 1Hz, the method of wavelet transform was implemented, and extract the signals at the frequency around 0.01Hz. Ground-based ULF geomagnetic data are a superposition of several signals: global magnetic perturbations, artificial noises, and magnetic signals possibly due to underground activities. To minimize artificial noises, we selected the midnight time data (JST 1:30-3:30). And to reduce the influence of global magnetic perturbations, the station Kanoya (KNY) was chosen as reference station. Then the statistical method of superposed epoch analysis (SEA) was adopted to highlight the weak but significant signal from noisy data. Finally, to verify the usefulness of the prediction model compared with random prediction. We evaluated the precursory information of ULF geomagnetic signals for local sizable earthquakes using statistical method receiver operating characteristic (ROC) curve. In ROC curve, the closer to the upper left corner, the better performing for this prediction model. The results will be shown in the presentation.