ESTIMATION OF HYDRAULIC CONDUCTIVITY IN UNCONSOLIDATED NEAR-SURFACE AQUIFERS USING NMR GEOPHYSICS

Abstract

In this research, we investigated the capabilities and limitations of using surface and borehole nuclear magnetic resonance (NMR) geophysical measurements to estimate hydraulic condictivity (K) at sub-meter to several-meter resolution in unconsolidated near surface aquifers. Co-located surface NMR, direct push NMR, borehole NMR and direct flow-based K measurements were performed over three geologically-distinct unconsolidated aquifers with varying degrees of magnetic mineralization. At each geographic location, multiple colocated NMR and flow-based K measurements were performed over lateral distances ranging up to 1000 meters. Optimal NMR-K calibration coefficients were computed to minimize the mean squared difference in log-K between NMR-derived K and available high resolution direct K measurements at three scales: at each well site, over each of the three geographic locations, and over the entire data set. K estimates from direct push NMR measurements were directly compared to direct push permeameter (DPP) K measurements at 0.5 meter vertical intervals at all locations. Optimal NMR-K calibration coefficients computed at different sites within each geographical location varied by less than a factor of 3, and the rms deviation between DPP-K and DP-NMR K at 0.5 meter resolution was approximately 0.64 orders of magnitude. A significant and unexpected finding was that a single fixed calibration coefficent performed nearly as well (at matching the DPP-K measurements) as the locally derived NMR-K coefficients. The surface NMR measurements provided adequate resolution to identify the predominant high-K and low-K sections of each aquifer. The optimized NMR-K estimates for the surface NMR CPMG sequence closely matched the optimized NMR-K estimates of the direct push NMR-K coeffients across all data sets, and varied little among the three geographic areas. CPMG surface-NMR-estimated mean K for each aquifer closely matched the mean DPP-derived-K and mean DP-NMR-K for each aquifer. In contrast, the optimized NMR-K coefficients for the surface NMR free induction decay (FID) measurements varied by more than an order of magnitude over the three geographic regions, due to the large differences in magnetic mineral content among the three aquifer source materials.