Cancellation of varying harmonic noise in magnetic resonance sounding signals

Abstract

Magnetic resonance sounding (MRS) is a promising geophysical method for direct detection and quantification of groundwater. However, practical application of MRS is greatly limited because it is vulnerable to electromagnetic interference. Most traditional denoising methods for MRS assume that the harmonic noise is stable within one measurement period. In this work, the harmonic noise in MRS is analyzed based on the short-time power spectral density (PSD). In the primary and reference coils, the harmonic noise is often found to be nonstationary, and its variations in both coils may be asynchronous. A frame-based denoising method is introduced based on multichannel Wiener filtering in the frequency domain. By establishing harmonic noise models with different variabilities, the frame-based denoising method is compared with traditional denoising methods when dealing with a variety of noise types. The results show that the frame-based method improves denoising when the harmonic noise has asynchronous variations and multiple sources. In addition, we analyzed the factors influencing this new approach and found that when the harmonic noise is stable, a longer frame length produces better denoising. When the harmonic noise changes asynchronously or multiple noise sources with different fundamental frequencies are present, there is an optimal range of frame lengths from approximately 0.08–0.2 s. The deviation between the Larmor frequency and the power frequency also affects denoising. This study has demonstrated improvements in denoising during short-time variations in noise and with multiple noise sources that will undoubtedly expand the scope of MRS applications.