Magnetotelluric Noise Suppression Based on Impulsive Atoms and NPSO-OMP Algorithm

Abstract

The magnetotelluric (MT) method is a mainstream geophysical exploration method widely used in deep mineral resource exploration and other fields. However, the MT signal is extremely vulnerable to noise since it employs natural electromagnetic fields as a source. In order to improve the signal-to-noise ratio, a novel MT noise suppression method based on impulsive atoms and a niche particle swarm optimization-orthogonal matching pursuit (NPSO-OMP) algorithm is proposed. First, an over-complete dictionary composed of impulsive atoms is well-developed to match different types of cultural noise. Then, the sparse decomposition process of OMP is optimized with NPSO. Finally, the original MT data are de-noised using the well-developed dictionary and the NPSO-OMP algorithm. By processing the synthetic and measured MT data, and comparing the experimental results with those by traditional methods, we found that the proposed method can effectively suppress the high-intensity impulsive noise in the MT signal under the premise of retaining useful information. The continuity of apparent resistivity and the phase curve are significantly improved, and the results are verified by the remote reference method and robust statistic estimation. When the observation data are subjected to persistent, strongly correlated noise pollution, our method can obtain better results than the remote reference method and robust statistic estimation. Thus, when it is difficult to obtain high-quality MT response curves through the remote reference method and robust statistic estimation, our method is a promising alternative.