Two-dimensional distribution of relaxation time and water content from surface nuclear magnetic resonance
Raphael Dlugosch, Thomas Günther, Mike Müller-Petke and Ugur Yaramanci
Journal name: Near Surface Geophysics
Issue: Vol 12, No 2, April 2014 pp. 231 - 241
Special topic: Magnetic Resonance in the Subsurface
Info: Article, PDF ( 1.75Mb )
Price: € 30
Development in instrumentation and data analysis of surface nuclear magnetic resonance has recently moved on from one-dimensional (1D) soundings to two-dimensional (2D) surveys, opening the method to a larger field of hydrological applications. Current analysis of 2D data sets, however, does not incorporate relaxation times and is therefore restricted to the water content distribution in the subsurface. We present a robust 2D inversion scheme, based on the qt approach, which jointly inverts for water content and relaxation time by taking the complete data set into account. The spatial distribution of relaxation time yields structural information of the subsurface and allows for additional petrophysical characterization. The presented scheme handles separated loop configurations for increased lateral resolution. Assuming a mono-exponential relaxation in each model cell, using irregular meshes, and gate-integrating the signal, the size of the inverse problem is significantly reduced and can be handled on a standard personal computer. A synthetic study shows that contrasts in both the quantities – water content and relaxation time – can be imaged. Inversion of a field data set outlines a buried glacial valley and allows the distinguishing of two aquifers with different grain sizes, which can be concisely interpreted together with a resistivity profile. The impact of the anisotropic weighting factor and subsurface resistivity on the inversion result are shown and discussed. A comparison of the results obtained by the previously used initial value and time-step inversion approaches illustrates the improved stability and resolution capabilities of the 2D qt inversion scheme.