Estimation of Hydraulic Conductivity in Unconsolidated Near-Surface Aquifers Using NMR Geophysics

In this research, we investigated the use of surface and borehole nuclear magnetic resonance (NMR) geophysical measurements to estimate hydraulic conductivity (K) at sub-meter to several-meter resolution in unconsolidated near-surface aquifers. Co-located surface NMR, direct push (DP) NMR, and direct push permeameter (DPP) K measurements were performed over three geologically-distinct unconsolidated aquifers with varying degrees of magnetic mineralization. At each location, multiple co-located NMR and DPP K measurements were performed over lateral distances ranging up to 1000 meters. K estimates from DP NMR measurements were directly compared to DPP K measurements at 0.5-meter vertical intervals at all locations. Optimized NMR-K calibration coefficients computed at different sites within each geographical location varied by less than a factor of 3, and the root mean square deviation between DPP and DP-NMR K was approximately 0.64 orders of magnitude. A significant and unexpected finding was that a single set of fixed NMR calibration coefficients performed nearly as well (at matching the DPP-K measurements) as locally optimized NMR-K coefficients. The surface NMR measurements provided adequate resolution to identify the high-K and low-K sections of each aquifer. Optimized NMR-K estimates for the surface NMR Carr-Purcell Meiboom-Gill CPMG sequence closely matched the optimized DP NMR-K coefficients across all data sets. In contrast, the optimized NMR-K coefficients for the surface NMR free induction decay measurements varied by more than an order of magnitude over the three geographic regions, due to differences in magnetic mineralization among the three aquifer materials. Overall, the results indicate that existing NMR-K transforms, originally developed and used to estimate K in consolidate limestone and sandstone oil reservoirs, can be employed to estimate K in near-surface unconsolidated reservoirs, but the optimized NMR-K transform coefficients can be up to two orders of magnitude different from those used in oilfield NMR logging.