We have used geostatistical analysis of radar reflections to quantify the correlation structures found in 2-D ground‐penetrating radar (GPR) images. We find that the experimental semivariogram, the product of the geostatistical analysis of the GPR data, is well‐defined and can be modeled using standard geostatistical models to obtain an estimate of the range or correlation length, and the maximum correlation direction, in the 2-D GPR image. When we compare the results from geostatistical analysis of GPR data from selected deltaic and barrier‐spit depositional environments we find different correlation structures in GPR images from different depositional environments. GPR images from braid deltas have near‐horizontal correlation directions and correlation lengths on the order of a few meters. In contrast, the GPR image of a fan‐foreset delta has a very long (>24 m) correlation length and a maximum correlation direction plunging 20°. In the GPR images from barrier spits, we find maximum correlation directions that are horizontal or plunging a few degrees. The correlation lengths range from 7 to 43 m, depending on the orientation of the GPR image relative to spit end growth, and on the specific radar facies that is analyzed.

No abstract

Abstract. The dielectric response of geologic materials is related to the amount of waterwetted solid surface present. This dependence raises the possibility that dielectric measurements may be used to detect changes in the state of the solid-fluid interface such as those caused by the sorption of a chemical species. In this study, complex impedance data are collected for systems of clean and oil-bearing sand and kaolinite vacuumsaturated with a 0.001 M NaC1 brine in the frequency range 100 kHz to 10 MHz. The effective dielectric constant K and electrical conductivity rr of the samples are calculated from these measurements assuming quasi-static conditions. We define the components in our system as brine, air, and a "wetted" solid phase. In this way, the contribution of the solid-fluid interface to the bulk dielectric response is included in the last component. We use an inclusion-based effective medium theory to extract the dielectric response of the wetted solid from the experimental data. The presence of sorbed oil is found to have little effect on the dielectric response of the wetted sand matrix, presumably because of the 2 very low surface area (0.2 rn/g) of the solid-fluid interface. In contrast, the kaolinite (5-12 m2/g) data indicate a decrease of the order of 50% in the dielectric constant of the wetted matrix phase as the amount of water-wetted surface in the sample decreases. The results from this experimental study show that the presence of sorbed oil leads to a detectable change in the dielectric properties of high surface area geologic materials.

No abstract

[1] Geostatistical analysis of a ground penetrating radar reflection image can be used to quantify the maximum correlation direction and the range of horizontal and subhorizontal radar reflections. A review of previous work, and an analysis of a photograph of layered sediments, suggest that the vertical resolution of a radar image strongly affects its lateral correlation structure. Numerical modeling was used to generate synthetic radar sections and investigate the effect of the vertical resolution of the radar measurement on the link between the correlation structure of the radar reflections and the true correlation structure of subsurface water content. The horizontal range of the radar reflections decreased as the vertical resolution improved, closely matching that of the water content when the vertical resolution was approximately equal to the vertical range of the water content.