Utilizing prepolarization SNMR for soil moisture measurements

Many biological, chemical, and hydrological processes in the soil depend on soil moisture. Although there exist numerous methods for determining soil moisture, the majority of them are either intrusive or measure soil moisture only indirectly. The latter case requires a calibration, which can be difficult because of the heterogeneity found in many soils. SNMR is a non-invasive method that can directly detect water content. It is commonly used to characterize subsurface aquifers. Typically, a surface transmitter coil is used to transmit an excitation pulse at the local Larmor frequency and the NMR response from the water in the subsurface is detected by a surface receiver coil. Recently, efforts have been made to apply the SNMR method to soil moisture measurements. To achieve this, we use a compact SNMR layout with a prepolarization coil that applies a prepolarization field before each experiment to amplify the spin magnetization at the footprint of the coil layout. Additionally, it becomes necessary to reduce the duration of the excitation pulses and to increase the pulse amplitude instead. In doing so, the effective dead time is reduced to enable the detection of the expected short relaxation times in soils.

The short pulses of high amplitude and the prepolarization switch-off present new challenges for the modeling of the acquired data. We have enhanced the forward modeling operator by the implementation of a numerical solver for the Bloch-equations. This allows us to account for the so-called Bloch-Siegert effect, which influences the measurements at high pulse amplitudes and can lead to significant errors in the SNMR inversion results if not considered properly. Furthermore, the solver of the Bloch-equations allows us to simulate the macroscopic magnetization during the prepolarization field switch-off and, thereby, account for non-adiabaticity during this time.

The new modeling and measurement system was evaluated using water-filled pallet boxes, and a good agreement between measured and simulated data was achieved. We continued with a case study on a peatland near Gnarrenburg, where we performed measurements on peat and mineral soil to demonstrate the applicability of the PP-SNMR method and the improved modeling. The soil moisture measured with PP-SNMR underestimates the original water content of undisturbed samples that have been taken for ground truth. A complementary NMR study in the laboratory shows that water in the micropores, for which the relaxation time is shorter or equal to the PP-SNMR dead time, cannot yet be captured in the field. Furthermore, the vertical resolution properties of PP-SNMR are not sufficient to identify distinct peat layers with thicknesses of less than 10 cm. However, apart from these issues, the soil water in mesopores and macropores is detected correctly and can be accurately characterized by the measured NMR relaxation properties.

Measuring relaxation times shorter than 6 ms still poses a major challenge, which we intend to overcome with further refinements to the receiving electronics and measurement scheme.