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

Abstract

Surface nuclear magnetic resonance (surface NMR) is a geophysical technique used in the exploration of shallow aquifers. It is based on measuring the NMR response of water molecules to excitation by electromagnetic pulses. By increasing the moment of applied pulses, successively deeper regions of an aquifer can be probed. The longitudinal relaxation time T1, determined from the NMR experiment, depends on pore size and can be potentially used to estimate hydraulic conductivity. A novel scheme was recently proposed that was shown theoretically to be more reliable for acquiring surface-NMR T1 data than traditional acquisition. In this proof-of-concept study we provide the first empirical evidence for the superiority of the novel scheme. We chose a survey site close to Skive, Denmark, where proximate boreholes indicate a homogeneous sandy aquifer in the top 30 m. The homogeneous composition implies that the distribution of pore sizes does not vary significantly across the formation. Because pore size is reflected by the T1 relaxation time, we therefore assume that the homogeneous aquifer can be characterized by a single T1 independent of the applied pulse moment (i.e., sampled depth region)—this is the benchmark condition we tested with our surface-NMR measurements. We collected surface-NMR T1 data employing the traditional as well as the novel acquisition scheme at various pulse moments. For each pulse moment we infer a T1 relaxation time based on extensively sampled data (14 delay-time data points). The T1 relaxation times obtained using the novel scheme show a constant value of about 820 ms (± 38 ms) for all pulse moments. In contrast, the T1 relaxation times determined using the traditional scheme vary significantly between 530 and 750 ms with pulse moment, which in an inversion would result in a spatial variation of T1 across the aquifer. The results based on the novel scheme are consistent with a homogeneous aquifer, which we expect based on the borehole information, whereas the variation of T1 obtained by the traditional scheme could be misinterpreted as a variation of pore size or hydraulic conductivity.