Abstract: Aiming at the problem that the magnetotelluric signal in the ore concentration is affected by the environmental noise and the human noise pollution, a combinatorial method based on the EMD and mathematical morphology filter is described for filtering of MT time-series. The principle and steps of the method are given, and the de-noising effect of this method is evaluated. Comparing with wavelet transform, the simulation test results proved the reliability of this method, and the field measurement data of a certain mine area is processed. The results show that the combined filtering method makes full use of the multi-scale decomposition and reconfigurable characteristic of EMD and the advantages of mathematical morphology filter. While filtering out the noise, the useful information is retained as much as possible for the MT signal. After de-noising, the variance of the response curve is reduced to 1/2, which provides a guarantee for the correct data processing and geological interpretation.

Key words: empirical mode decomposition(EMD), mathematical morphological filtering, adaptive filtering, magnetotelluric signal, de-noising, ore concentration area