A combinatorial filtering method for magnetotelluric data series with strong interference

In highly industrialized areas, magnetotelluric (MT)-induced variations are contaminated by strong manmade noise signals. A method is described as an alternative approach for noise removal, based on a combination of empirical mode decomposition (EMD) with independent component analysis (ICA). The filtering procedure takes advantage of the fact that data are analyzed through different scale levels, which requires a minimum of human intervention and leaves good data sections unchanged. Principle and steps of method are discussed, and de-noising results are evaluated by some parameters. After the filtering stage, data is processed in the frequency domain to yield two sets of reliable MT transfer functions and the result was compared with that of the EMD-Wavelet method. Simulated signal and measured MT data series are processed. The results show that this procedure can lead to greatly improved apparent resistivity and phase curves after processing. Point defects are filtered out to eliminate their deleterious influence, which yields reliable estimates of the MT transfer functions. The EMD-ICA method provides a new method for the de-noising of MT data series under the condition of low SNR.