Variational Mode Decomposition Based Processing of Surface NMR Data

TY - GEN

T1 - Variational Mode Decomposition Based Processing of Surface NMR Data

AU - Naveed, Khuram

AU - Larsen, Jakob Juul

PY - 2024

Y1 - 2024

N2 - Surface nuclear magnetic resonance (sNMR) method provides direct sensitivity to groundwater, but the method is often challenged by the presence of an overwhelming level of noise in acquired sNMR data. That is because the free induction decay (FID) signal from water molecules has several times lower amplitude than the coincident noise potentials. A model-based subtraction approach is often used to mitigate the noise problem. While this approach caters for powerline harmonics it cannot cater for random noise and may leave traces of subtracted noise due to errors in the estimation of noise parameters. This work explores the use of variational mode decomposition (VMD) for extracting the FID signal which happens fit the definition of an IMF. Hence, an intelligent framework around VMD is suggested to selects decomposition parameters in the spectral domain with an objective to render the FID signal as a mode. Preliminary results are presented to demonstrate the efficacy of our approach over the benchmark method.

AB - Surface nuclear magnetic resonance (sNMR) method provides direct sensitivity to groundwater, but the method is often challenged by the presence of an overwhelming level of noise in acquired sNMR data. That is because the free induction decay (FID) signal from water molecules has several times lower amplitude than the coincident noise potentials. A model-based subtraction approach is often used to mitigate the noise problem. While this approach caters for powerline harmonics it cannot cater for random noise and may leave traces of subtracted noise due to errors in the estimation of noise parameters. This work explores the use of variational mode decomposition (VMD) for extracting the FID signal which happens fit the definition of an IMF. Hence, an intelligent framework around VMD is suggested to selects decomposition parameters in the spectral domain with an objective to render the FID signal as a mode. Preliminary results are presented to demonstrate the efficacy of our approach over the benchmark method.

UR - http://www.scopus.com/inward/record.url?scp=85214819907&partnerID=8YFLogxK

U2 - 10.3997/2214-4609.202420100

DO - 10.3997/2214-4609.202420100

M3 - Article in proceedings

BT - NSG2024, 30th European Meeting of Environmental and Engineering Geophysics

PB - European Association of Geoscientists and Engineers

T2 - NSG 2024 30th European Meeting of Environmental and Engineering Geophysics,

Y2 - 8 September 2024 through 12 September 2024

ER -