Comparison of prepolarized surface NMR and conventional laboratory NMR in a peatland area

A sensor for non-invasive and in-situ soil moisture detection based on the principle of prepolarized surface-nuclear magnetic resonance (PP-SNMR) was previously developed and first measurements in natural environments were conducted. This study aims to evaluate the accuracy of the novel PP-SNMR results against established laboratory NMR based on data from a peatland area. Analyzing undisturbed soil samples from different positions and depths, we found that the effective dead-time of the PP-SNMR device and the spatial resolution of the PP-SNMR approach are important factors controlling the agreement between the two datasets. Accounting for individual uncertainties, the relaxation time distributions of laboratory NMR and PP-SNMR are in good agreement for relaxation times greater than 6.0 ms. This limit essentially corresponds to the effective dead time of the PP-SNMR device. Correspondingly, the moisture content of soils with very short relaxation times lower than 6.0 ms is significantly underestimated by PP-SNMR. However, even if the relaxation time distributions match in shape, significant discrepancies of the total soil water content estimates from field and laboratory have been observed. We assign these discrepancies to the differences between the sample volume and the sensitive volume of the PP-SNMR, i.e., to resolution properties of the PP-SNMR inversion and to spatial heterogeneity of the subsurface. Overall, we conclude that laboratory NMR and PP-SNMR data are in good agreement when accounting for the given limitations. However, we emphasize the need to further reduce the effective PP-SNMR dead time and to develop novel schemes that improve the resolution capabilities.