Ground Penetrating Radar And Electromagnetic Profiling Of Bedrock Topography At The Portsmouth Gaseous Diffusion Plant, Piketon, Ohio

Abstract

Low-frequency ground penetrating radar (GPR) and electromagnetic profiling (EM-3 1 and EM-34) surveys were carried out at the Portsmouth Gaseous Diffusion Plant (PGDP) in Pike County, Ohio. The primary objective of the surveys was to profile sedimentary bedrock topography beneath unconsolidated fluvial and lacustrine deposits. The EM surveys show high conductivity (8.5-39 mS/m) values for the unconsolidated sediments and underlying bedrock, presenting a hostile environment for GPR. The low-frequency GPR, however, penetrated to bedrock depths of up to 10 m, which exceeds the predictions of forward modelling. The data show that low-frequency GPR is effective in this lossy-dielectric environment, and is not accurately modelled using typical low-loss assumptions. The site occupies a buried channel of the late-Tertiary Portsmouth River (Figure 1). The unconsolidated deposits include the clay-rich Minford formation, which extends from the surface to approximately 4 m, and the gravel and clay bearing sand of the Gallia formation. The Gallia deposits are part of interbraided fluvial channels which vary in thickness, and are laterally heterogeneous. It is the bedrock beneath the Gallia that is the primary target of these surveys. The bedrock is generally the Sunbury Shale or, where the channel is more deeply eroded, the Berea Sandstone. Electromagnetic survey data were recorded at 5 m intervals using the EM-3 1 (4 m coil separation) and the EM34 (10 m and 20 m coil separations) in an open field within the X-749 area of PGDP. Substantial well control exists at the site, and shows that the depth to bedrock (Sunbury) is fairly constant at 5.6m. The Gallia thickness ranges between l-2 m, underlying 4-5 m of the Minford formation. Using the well data as constraints, an inversion of the EM data for apparent conductivity produced values shown in Figure 2 for a two-layer model over a halfspace. The model shows bedrock depths in the expected range, with some outliers, and a greater than expected Gallia thickness. Electrical conductivities for the Minford unit (layer 1) are generally constant at 26 mS/m. The Gallia layer (layer 2) conductivity is generally lower, with values in the range of 8.5-15.5 mS/m, except at 85-100 m, where the conductivity is close to that of the shale. The bedrock shale (underlying half-space) shows conductivities consistently near 39 mS/m. The low-frequency ground penetrating radar surveys were conducted at several sites within and adjacent to PGDP using a Sensors and Software pulseEkko IV radar unit with 25 and 50 MHz bistatic antennas. Figure 3 shows a combined 25 MHz and 50 MHz profile within the X-749 area, along with the lithology information derived from nearby wells. The shallow 50 MHz data (cl70 ns) were merged with the deeper 25 MHz data (>170 ns) to show the best resolution with depth. The profile is located near and parallel to the EM surveys shown in Figure 2, but not collinear with them. The 25 MHz data contain high-amplitude reflections originating from the interface between the Sunbury bedrock and Gallia deposits, and deeper reflections related to the Sunbury/Berea interface. The depths to the bedrock reflections correlate with bedrock depths derived from the EM surveys and available well log data. The 50 MHz data reveal the Minford/Gallia contact within the unconsolidated deposits, and show additional reflections and diffractions due to surface layer heterogeneities. Figure 4 shows a 25 MHz profile that was collected along a west-to-east traverse on the Wilber property south of PGDP. The profile clearly shows the eroded bedrock surface, and the downcutting of the Cuyahoga Shale. A bedrock channel is evident beneath a surface wash feature. Despite the highly conductive clay surface layer, the GPR successfully penetrated to bedrock depths of up to 10 m. This penetration depth exceeds the performance predicted by GPR forward modelling, using conductivity values obtained from the EM surveys and depth information from the well logs. The modelling assumes that the GPR is operating in a low-loss environment, where the effects of displacement currents dominate the effects of conduction currents. When this is the case, the loss tangent [described by Ward and Hohmann (1988)] is much less than 1: