Stability of MRS signal and estimation of data quality
J.F. Girard, A. Legchenko and M. Boucher
Issue: Vol 3, No 3, August 2005 pp. 187 - 194
Special Topic: Magnetic Resonance Sounding
Info: Article, PDF ( 3.05Mb )
Based on the principle of nuclear magnetic resonance (NMR), the magnetic resonance sounding (MRS) method is used to measure the magnetic resonance signal from groundwater via a surface loop. The accuracy of the measurements depends on both the stability of the magnetic resonance signal under varying experimental conditions (geomagnetic field, electromagnetic noise, etc.) and the accuracy of the measuring device. The efficiency of the signal processing algorithms adopted by the NUMISPlus device, as well as sensitivity to the signal-to-noise (S/N) ratio (S/N), are numerically investigated. The results are shown to be stable when S/N>5. Numerical simulation of the magnetic resonance signal shows that with a fluctuating geomagnetic field and within a few Hertz of the frequency offset, amplitude remains stable whereas phase follows the geomagnetic field variations. An anti-correlation is observed between phase and frequency shift. The numerical results were validated experimentally using the NumisPlus system at St-Cyr-en-Val near the city of Orléans, France. For filtered data with S/N>4 and a frequency offset within ±0.5 Hz, the relative standard deviation of the measurements was found to lie within ±12% for amplitude and ±10° for phase. The standard deviation of the signal, which is strongly weighted by the S/N ratio, is higher for lower pulses than for higher pulses. Analysis of the stability of the current transmitted by NumisPlus reveals that for weak currents (<5 A), the standard deviation of the pulse amplitude is 3.7%, whereas for stronger currents (>100 A), it is only 1.3%. We thus conclude that for currents greater than 5 A, the instrumental instability of NumisPlus is significantly less than the electromagnetic noise, which can be considered as the major source of error. The S/N ratio can be significantly improved by stacking, which is efficient but time consuming. Note that the accuracy of measurements carried out over a long time (due to stacking) may decrease as measuring conditions can change during any one sounding. The accuracy of the field measurements for S/N>5 enables the complex MRS signal to be used in the inversion process instead of only amplitude. Nevertheless, any variation in the geomagnetic field with time must be taken into consideration during inversion of the MRS signal.