Aarhus University Seal

Efficient processing of surface NMR data with spectral analysis

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review


Due to its direct sensitivity to groundwater, surface nuclear magnetic resonance (NMR) is an intriguing, non-invasive geophysical technique. However, the surface NMR signal is inherently weak. In turn, the signal-to-noise ratio is often poor, and many signal processing steps must be used to improve it. One of the last processing steps is the extraction of the complex envelope of the NMR signal. This is normally done with synchronous detection, but a different method, spectral analysis was recently suggested, and it has shown promising results. Spectral analysis differs from synchronous detection, as it is a non-causal filtering method with a much narrower detection bandwidth and corresponding improved noise rejection. Until now, spectral analysis has not been applied to data sets acquired with commercially available equipment. We investigate the applicability and performance of spectral analysis on two data sets acquired with the two instruments on the market. The investigation is done with synthetic signals embedded in noise-only data and with field data. Our analysis reveals that spectral analysis is applicable for both instruments and leads to high-quality NMR envelopes. With a synthetic mono-exponential NMR signal embedded in noise-only data, we find that spectral analysis is better able to retrieve all NMR parameters. With field data, we find that the relative uncertainty on gated data is improved by a factor of 2-4 in one data set and a factor of 4-9 in a second data set. We invert both data sets, and show that spectral analysis leads to better-resolved models with lower model uncertainty.

Original languageEnglish
JournalGeophysical Journal International
Pages (from-to)286-298
Number of pages13
Publication statusSubmitted - 16 Dec 2021

    Research areas

  • Hydrogeophysics, Fourier analysis, Time-series analysis, MAGNETIC-RESONANCE DATA, NOISE CANCELLATION, STACKING, SOFTWARE, TIME

See relations at Aarhus University Citationformats

ID: 262074305