Surface-NMR measurements of the longitudinal relaxation time T1 in a homogeneous sandy aquifer in Skive, Denmark

Efficient groundwater management requires reliable means of characterizing shallow groundwater aquifers. One key parameter in this respect is hydraulic conductivity. Surface nuclear magnetic resonance (NMR) is a geophysical exploration technique that can potentially provide this type of information in a noninvasive, cost-effective way. The technique is based on measuring the precession of nuclear spins of protons in groundwater molecules. It involves large loop antennas deployed on Earth's surface to generate electromagnetic pulses tuned to specifically excite and detect groundwater proton spins. Naturally, the excited state of spins is transitory - once excited, spins relax back to their equilibrium state. This relaxation process is strongly influenced by the spin environment, which, in the case of groundwater, is defined by the aquifer. By employing empirical relations, changes in relaxation behavior can be used to identify changes in aquifer hydraulic conductivity, making the NMR relaxation signal a very important piece of information. Particularly, efforts are made to record the longitudinal relaxation parameter T1, because it is known from laboratory studies that it often reliably correlates with hydraulic conductivity, even in the presence of magnetic species. In surface NMR, T1 data are collected by recording the NMR signal amplitude following two sequential excitation pulses as a function of the delay time τ between the two pulses. In conventional acquisition, the two pulses have a mutual phase shift of π. Based on theoretical arguments it was recently shown that T1 times acquired according to this conventional surface-NMR scheme are systematically biased. It was proposed that the bias can be minimized by cycling the phase of the two pulses between π and zero in subsequent double-pulse experiments, and subtracting the resulting signal amplitudes (phase-cycled pseudosaturation recovery scheme, pcPSR). We present the first surface-NMR T1 data set recorded employing the pcPSR scheme and compare it to conventional T1 data. For our feasibility study we have chosen a site in Skive, Denmark, that features excellent signal/noise conditions, allowing us to collect high quality data in reasonable survey time. In addition, proximate boreholes and TEM measurements suggest a relatively homogeneous aquifer extending from 5 to more than 25m below surface. We may therefore expect roughly constant T1 relaxation times throughout the shallow aquifer, providing us a simple framework for our comparative study. We used a 50x50m surface-NMR loop and employed 16 pulse moments selected to spatially cover the shallow aquifer region. For each pulse moment, we recorded surface-NMR T1 data densely sampled at 14 delay times τ between 250 and 4'000 ms. On this high-quality data set we demonstrate that the pcPSR acquisition approach yields to a good degree homogeneous T1 relaxation times, whereas the conventional approach leads to variations in T1 that could be misinterpreted in terms of changes of aquifer characteristics. Thereby we provide first empirical evidence for the superiority of the pcPSR scheme for surface NMR T1 acquisition.