A hybrid denoising method for low-field nuclear magnetic resonance data

Area of Science:

• Geophysics
• Petroleum Engineering
• Signal Processing

Background:

• Low-field nuclear magnetic resonance (NMR) is vital for unconventional oil and gas reservoir exploration.
• Low signal-to-noise ratio (SNR) in NMR echo signals hinders accurate T2 spectrum inversion.
• Effective denoising is critical for preprocessing low SNR NMR data.

Purpose of the Study:

• To propose and validate a hybrid empirical mode decomposition-singular value decomposition (EMD-SVD) method for denoising low SNR NMR data.
• To enhance the accuracy of T2 spectrum inversion in unconventional reservoirs.
• To improve the quality of NMR log data processing.

Main Methods:

• Empirical Mode Decomposition (EMD) to decompose NMR echo data into intrinsic mode function (IMF) components and a residual.
• Singular Value Decomposition (SVD) applied to high-frequency IMF components for denoising.
• Reconstruction of the denoised signal by summing low-frequency IMFs, denoised high-frequency IMFs, and the residual.

Main Results:

• The proposed EMD-SVD method effectively denoises low SNR NMR data.
• Numerical simulations, experimental data, and NMR log data processing confirmed the method's effectiveness.
• Inverted NMR spectra using EMD-SVD showed higher quality compared to standalone EMD or SVD methods.

Conclusions:

• The hybrid EMD-SVD method offers a significant improvement for denoising NMR data in unconventional reservoir exploration.
• This technique enhances the reliability and accuracy of T2 spectrum inversion.
• The EMD-SVD approach presents a valuable tool for petroleum geophysics and reservoir characterization.