Presenter Information

Annette Hein, University of Wyoming

Department

Department of Geology and Geophysics

First Advisor

Andrew Parsekian

Description

Surface nuclear magnetic resonance (NMR) is a unique geophysical method due to its direct sensitivity to water. An NMR measurement produces a sounding that shows volumetric water content as a function of depth, which is useful for hydrogeology. A key limitation to overcome is the difficulty of obtaining usable data from surface NMR measurements in environments with anthropogenic electromagnetic noise, particularly constant frequency sources such as powerlines. Noise from these sources is typically much larger in magnitude than the desired NMR water signals, and it significantly corrupts the data. Previous research has explored a variety of methods for removing the noise, usually based on prior knowledge of the noise sources. Here, I take a different approach based on prior knowledge of the NMR water signal, which has a well-defined mathematical form. This form can be exploited to identify and remove noise from the data. I present a method of removing certain types of noise by utilizing frequency domain symmetry of surface NMR signals to reconstruct portions of the spectrum corrupted by frequency-domain peaks. This procedure is simple, does not introduce errors into the dataset, and requires no prior knowledge about the noise source. Modeling and field examples show that the noise-reduction procedure decreases the effects of powerline harmonics on the water content inversion and makes the inversion more accurate than before.

Comments

EPSCoR Community College Transition Program

Included in

Education Commons

Share

COinS
 

Removing Harmonic Noise from Geophysical Surface Nuclear Magnetic Resonance Measurements

Surface nuclear magnetic resonance (NMR) is a unique geophysical method due to its direct sensitivity to water. An NMR measurement produces a sounding that shows volumetric water content as a function of depth, which is useful for hydrogeology. A key limitation to overcome is the difficulty of obtaining usable data from surface NMR measurements in environments with anthropogenic electromagnetic noise, particularly constant frequency sources such as powerlines. Noise from these sources is typically much larger in magnitude than the desired NMR water signals, and it significantly corrupts the data. Previous research has explored a variety of methods for removing the noise, usually based on prior knowledge of the noise sources. Here, I take a different approach based on prior knowledge of the NMR water signal, which has a well-defined mathematical form. This form can be exploited to identify and remove noise from the data. I present a method of removing certain types of noise by utilizing frequency domain symmetry of surface NMR signals to reconstruct portions of the spectrum corrupted by frequency-domain peaks. This procedure is simple, does not introduce errors into the dataset, and requires no prior knowledge about the noise source. Modeling and field examples show that the noise-reduction procedure decreases the effects of powerline harmonics on the water content inversion and makes the inversion more accurate than before.