De-Noising Algorithm for Magnetotelluric Signal Based on Mathematical Morphology Filtering

Abstract

In this paper, an effective de-noising algorithm based on mathematical morphology filtering for magnetotelluric sounding data is presented. Magnetotelluric signals are nonlinear, non-stationary, non-minimum phase, they do not meet the basic requirements of the Fourier transform based on the traditional power spectrum estimation. Mathematical morphology filtering is a new signal analysis method developed in recent years for dealing with non-linear, non-stationary signal. This paper briefly introduce the mathematical morphology filtering basic principles and algorithms. According to the properties of structuring elements, the mathematical morphology filtering is designed. Analysis structuring elements type selection program by filtering performance. Based on the measured signal processing, we discussed its application in magnetotelluric sounding data processing and strong interference separation. Experimental results indicate that the proposed method is feasible and can effectively eliminate larger scale disturbance and baseline drift of magnetotelluric sounding data. In addition, the method is efficient to keep the main characteristics of the original signals, and is helpful to improve signal quality and information interpretability for magnetotelluric sounding data.