A feasibility study on the estimation of water retention parameters from surface nuclear magnetic resonance measurements in the vadose zone

The surface nuclear magnetic resonance (SNMR) method is usually applied for groundwater prospection. Its unique property - distinct from other hydrogeophysical methods - is the direct sensitivity to water content in the subsurface. The inversion of SNMR data yields the subsurface water content distribution without the need of a specific petrophysical model. Recent developments in instrumentation, i.e., decreased instrumental dead times and advanced noise cancellation strategies enable the use of this method for investigating the vadose zone. The first attempt to interpret SNMR measurements with the focus on hydraulic parameters is the inversion approach of Costabel and Yaramanci (2011).Their inversion directly provides WR parameters by parameterizing the capillary fringe (CF) by means of soil-physical water retention (WR) models. We have developed and investigated this inversion approach further to assess its general applicability. A sensitivity study based on both synthetic and real data analyzes the resolution properties, the uncertainties and the covariances of the involved parameters: the saturated and the residual water content, a parameter for the height of the CF, a parameter describing the gradient of the water content increase in the CF, and the water table. We found that it is not meaningful to invert for all parameters at once. At least, an estimate of the CF's height or the water table must be available as a-priori information. Otherwise the CF inversion cannot reliably be applied, even when the noise level is unrealistically low. The water content of the saturated zone is generally estimated with high accuracy, i.e., errors of less than 1%. Depending on the actual noise level, the uncertainties of the other WR parameters are in the range of 10 to 100%. We conclude that, for moderate noise conditions, this kind of inversion provides WR parameters sufficiently accurate to estimate the unsaturated hydraulic conductivity roughly. However, a serious conceptual shortcoming of this approach is the inconsistency between observing a static system on the one hand and estimating dynamic parameters on the other hand. A promising approach to overcome this problem is to do time-lapse measurements. To assess the potential of SNMR for such experiments, we combine hydraulic simulations and SNMR forward modeling calculations to develop and investigate adequate measurement strategies based upon realistic field scenarios. We found that the common SNMR measurement scheme must be modified to allow faster repetitions, e.g., when monitoring infiltration of water with high dynamics. Otherwise it is not possible to realize an appropriate resolution in time. For such modifications one must accept the loss of spatial resolution. However, the direct sensitivity of the SNMR method for dynamic water content changes is an important benefit and we expect that future SNMR inversion approaches will provide hydraulic parameters, at least for the vertical water flux through the vadose zone. References: Costabel, S. and Yaramanci, U. (2011). Relative hydraulic conductivity in the vadose zone from magnetic resonance sounding - Brooks-Corey parameterization of the capillary fringe. Geophysics, 76 (3):B1-B11.