SEGJ Technical Conference


Time series analysis for magnetotelluric data using cepstrum


Abstract
Magnetotelluric method is an electromagnetic method for deep subsurface exploration using a perturbation of natural electromagnetic field. The most serious problem of MT data analysis at a surrounding region of urban area is to reduce a heavy noise from observed data. A new approach to reduce a heavy noise of time series data is presented in this paper. The cepstrum analysis was used to remove a noise from raw MT data which was observed at Ito campus of Kyushu University. As a result of experiment using field data, it was found that cepstrum analysis was effective to get smooth MT impedance after removing a noise of MT data.