海底電磁場観測における高品質データ部の抽出方法の検討

抄録

We propose a method for estimating high-quality magnetotelluric impedances by evaluating data objectively from long-term seafloor electromagnetic observations. The marine magnetotelluric method, based on natural fluctuations of geomagnetic and electric fields measured on the seafloor, elucidates the distribution of electrical resistivity in the solid Earth. Recently, enormous data are acquired due to increase of observation site and sampling rate, so that the selection and discarding of data are required before analyzing to obtain the target’s EM responses more accurately. However, these processes usually depend on empirical methods. In this study, we addressed this issue by analyzing data acquired at the Kumano-nada area, Japan. We evaluated the quality of the multi-component electric field data using different combinations of electrodes. Then, we divided the time-series data at certain time intervals and evaluated the quality of the observed data for each section. The quality of the observed data was determined based on errors in the impedance phase and the value of multiple coherencies. Using the best combination of electrodes and the better section of the observed data sets, we successfully estimated magnetotelluric impedances of better quality than those obtained by the conventional analysis. Our proposed method enables automatic selection of high-quality observed data effectively, facilitating the high-quality MT response functions required for imaging resistivity structures.