Application of local mean decomposition and wavelet threshold in magnetotelluric noise suppression

李晋,彭冲,汤井田,燕欢,蔡剑华

DOI: https://doi.org/10.13465/j.cnki.jvs.2017.05.021

2017-01-01

Abstract:The magnetotelluric sounding method is a method based on the principle of electromagnetic induction,to detect electrical properties and distribution characteristics of underground rock layers by use of the natural alternating electromagnetic field.However,the natural electromagnetic field has a wider frequency band range and weak signals,and these signals are easy to be disturbed by all kinds of electromagnetic noise in the actual measurement,so the interpretation level of the subsequent electromagnetic inversion method is seriously effected.In order to solve this problem,combining the adaptability of local mean decomposition (LMD) with the multi-resolution of wavelet,here the magnetotelluric noise suppression method based on local mean decomposition and wavelet threshold was proposed.First of all,LMD was used to divide a noisy signal into a number of product functiion (PF) components.Then,according to magnetotelluric signal-tonoise characters,PF1 component was retained,the appropriate wavelet threshold was chosen to deuoise the resteach PF component.Finally,the magnetotelluric useful signal was obtained with superposition and reconstruction.Using a computer to simulate typical strong interferences,the denoising performances of the proposed method were studied under the conditions of different wavelet functions,decomposition layers and threshold modes,and the method was applied to process the magnetotelluric data measured in ore concentration areas.The results showed that the proposed method can better extract the outline features of large-scale strong interferences superimposed on weak magnetotelluric signals,and the apparent resistivity curve is more smooth and continuous;the quality of magnetotelluric data is improved significantly in lower frequency bands.