Noise Suppression Method for Magnetic Resonance Sounding Signals Based on Double Singular Value Decomposition

Abstract

AbstractMagnetic resonance sounding (MRS) is a geophysical method that directly detects, evaluates, and monitors groundwater resource. The amplitude of the MRS signal detected by the instrument is on the order of nanovolts, resulting in very sensitive to environmental noise and power line harmonics. The singular value decomposition (SVD) method separates the signal from the noise based on the different component contributions to the singular values between the MRS signal and noise. In this paper, we propose a noise suppression method based on double SVD (DSVD) for the reliably extraction of an MRS signal with high‐level noise. The first SVD process is to extract the harmonics from the noise‐only data, in which the MRS signal is removed by a band‐stop filter. After subtracting the extracted harmonics from the measured data, we use the SVD algorithm a second time to obtain the MRS signal with further suppression of random noise. From the synthetic results with different signal‐to‐noise ratios, we conclude that the DSVD method improves the signal‐to‐noise ratio by more than 28 dB, and the fitting errors of the initial amplitude and relaxation time are ±2% and ±3%, respectively. Moreover, we analyze the selection criterion for the two key parameters of the algorithm, the delay step size, and the decomposition order. The processing of measured field data further validates the effectiveness of the proposed algorithm. Finally, discussions with other denoising algorithms show that the DSVD algorithm has better performance in a variety of noise case.