AUDIO MAGNETOTELLURIC SIGNAL-NOISE IDENTIFICATION AND SEPARATION BASED ON MULTIFRACTAL SPECTRUM AND MATCHING PURSUIT

Abstract

To avoid the blindness of the overall de-noising method and retain useful low frequency signals that are not over processed, we proposed a novel audio magnetotelluric (AMT) signal-noise identification and separation method based on multifractal spectrum and matching pursuit. We extracted two sets of multifractal spectrum characteristic from AMT time-series data to analyze the singularity. We used a support vector machine approach to learn the multifractal spectrum characteristics in a sample’s library and generate a model of support vector machine to distinguish between sections with and without interference in the measured AMT data. The matching pursuit algorithm was used to separate only those sections identified as having interference. Experimental results showed that the proposed method can effectively identify interference in the EMTF mathematical model and measured AMT data. Sections without interference were accurately preserved and reconstructed AMT signals were close to the natural electromagnetic field. The resulting apparent resistivity-phase curve is more continuous and smooth, and effectively improves the quality of AMT data. Moreover, the proposed method provides more reliable AMT data for subsequent electromagnetic inversion.