Noise Reduction Method for Long Period Magnetotelluric Signals Based on VMD-Wavelet Thresholding

Abstract

This article proposes a noise reduction method for long-period magnetotelluric (MT) data that combines Variational Mode Decomposition (VMD) with wavelet thresholding (WT). Firstly, the signal is decomposed and combined with correlation coefficients to screen component signals. Then, the outliers of the component signals are identified and replaced using the median absolute deviation. The filtered component signals are denoised and reconstructed using wavelet thresholding to obtain denoised magnetotelluric signals.