DC, EM & NMR technologies

[DE-04] Numerical experiment for processing noisy magnetotelluric data based on independence of signal sources and continuity of response functions

*Hiroki Ogawa^1,2, Koichi Asamori², Takumi Ueda¹ (1. Waseda Univ. (Japan), 2. JAEA (Japan))

In the magnetotelluric (MT) method, observed data consist of the sum of several types of signals including the natural signal and artificial noises. The application of the technique of blind source separation (BSS) to MT data has been reported to extract responses of the natural signal from observed data that are often regarded as convolutively mixed signals. In performing BSS, it is usual to reconstruct the MT data after transforming the observed data into several components of signals and eliminating the specific components which correspond to noise. However, improper subtraction of values from separated signals can lead to the loss of useful values of natural signal or missing a large amount of noise-affected values, which may result in failure in deriving true MT responses. There are few cases of verifying the validity of how to subtract values of noise from separated signals and reconstruct MT data.
This study presents a new attempt to enhance the quality of MT responses in the time-frequency domain by utilizing the method of frequency-domain independent component analysis (FDICA). The MT response functions, which are represented as the ratio of horizontal electric to magnetic field components, are substantially constant over the relatively short period of time. Moreover, the MT response functions vary smoothly in the direction of frequency. An evaluation factor with respect to these two characteristics of the natural signal was introduced when determining which values to subtract prior to the reconstruction of the observed data. In the numerical experiment, the effectiveness of the proposed method is illustrated by applying the method to a set of MT time series that contains synthetic strong noise generated over the whole observation time.