Near-Surface Geophysics General Contributions I
Presiding: R Knight, Stanford University; K Singha, Stanford University
NS11A-01 08:30h
Investigating the Relationship Between Sedimentology and GPR Data at the Borden Aquifer
Ground penetrating radar (GPR) is becoming an increasingly popular subsurface characterization tool. While GPR data can provide important insights on subsurface heterogeneity, it is not always clear how the reflections in a radar image are related to sedimentary structures in the earth. Interpreting GPR reflection data to obtain a model of the subsurface therefore requires insight into how variations in geologic properties are recorded in the radar signal. We designed a field study to investigate this problem in an unsaturated region of the glacio-fluvial sediment that makes up the Borden aquifer (Canadian Forces Base Borden, Ontario). To conduct our experiment, we first collected 200, 450, and 900 MHz GPR data along a 20 m line transecting an undisturbed portion of a sand quarry. Next we employed a backhoe to expose the section of the subsurface that was imaged by the radar to a depth of about 3 m. This allowed us to map the sedimentary section and collect core samples that could later be used for laboratory measurements. The sediment in the area we studied was texturally diverse, ranging in size from clays to cobbles. Distinctive depositional and erosional features, including large displaced sediment blocks several meters long, made the site a good candidate for GPR imaging. However, upon comparing the radar images to our sediment map we found a poor direct correspondence between the two. Using the position of marker beds in the radar image and the sedimentary section, we have been able to identify subsurface velocity variations as one of the causes of complexity in the radar image. The magnitude of the variations (8-14 m/ns) is consistent with the range of velocities expected at the site based on water content measurements made on core samples (3-40% vol./vol.). We note that this large range in velocity reflects a contrast in the water retention properties of different sediments at the site and would not be as significant in studies conducted under saturated conditions. In addition, using finite difference simulations of GPR wave propagation through a simplified model of the site, we have found that data at late times are complicated by wave interference (i.e., the concurrent arrival of reflected energy from different parts of the subsurface at the receiver antenna). This interference makes it difficult to identify sedimentary features in the radar data, like the large sediment blocks, even after constant velocity migration. We have found that the use of velocity analysis and numerical modeling were critical tools that aided in our interpretation of the radar data in this complex geologic environment.
NS11A-02 08:45h
Separate and Joint Inversion of TE and TM Georadar Data for Properties of Thin Surface Waveguides
Pronounced dispersion of georadar waves is observed at locations distinguished by thin surface layers of high permittivity material (e.g. water-saturated soil). The dispersion characteristics depend on the permittivity and thickness of the effective surface waveguide and the permittivity of the material below it. We introduce a scheme for estimating the values of these parameters from dispersed transverse-electric (TE) and transverse-magnetic (TM) georadar data that is analogous to recently developed methods for analysing dispersed Rayleigh waves recorded on multichannel seismic data. Our scheme involves calculating phase-velocity spectra, picking dispersion curves from the spectra, and inverting the dispersion curves for the subsurface material properties using a combined local and global minimization procedure. Application of this new scheme to synthetic and field data demonstrate its efficacy in providing the required physical property information. At locations where the surface layer is relatively distinct and uniform, inversions of the resultant high quality dispersed TE georadar data are sufficient. In more heterogeneous environments, joint inversions of the TE and TM data, which usually include information in overlapping but somewhat different frequency ranges, may be required.
NS11A-03 09:00h
Reflectivity Modeling of a Ground Penetrating Radar Profile of a Saturated Alluvial Formation
We analyze a GPR reflection profile with synthetic models generated from VRPs and borehole logs at an unconfined fluvial aquifer near Boise, Idaho. The 1D synthetic traces are developed from profiles of dielectric constant and conductivity values. Reflections at the water table/capillary fringe, the base of a sand-filled channel, and the base of two sand-rich lenses within a cobble-dominated unit are recognizable in the surface reflection profile and in all reflectivity models. Less-prominent reflections within stratigraphic units occur in both the surface profile and reflectivity models. Although such minor reflections are not easily correlated, general similarity in presence and location indicate that they are useful for recognizing internal facies structure or character in some cases.
NS11A-04 09:15h
Seismic Velocity, Q, Lithology and Structural Imaging at a Ground Water Contamination Site
A high resolution seismic velocity model has been determined by waveform tomography applied to two vertical seismic profiles (VSPs) and a 2D surface seismic dataset from a ground water contamination site at Hill Air Force Base (HAFB), Utah. The dataset has useful frequency content between ~ 15Hz and ~ 115Hz for waveform tomography, and 100Hz and 220Hz for depth migration, but significant energy goes up to ~ 350Hz. The target dimension measures 21.4m wide and 15m deep. Features as small as ~ 1.5m are recovered in the model. The structural details in the model correlate well with a post-stack depth-migrated image, using the 2D data recorded at the surface between the two VSP boreholes (Figure 1). Using the final waveform tomography velocity model as an initial model, we determine Q value in the target area using further iterations of waveform tomography. Q values vary from 10 to 50. Generally larger velocity features have larger Q values, with the largest Q value (50) identified near the northern borehole. The good correlation (Figure 1) between the model, the lithologic logs available, and the depth migration makes it possible to geologically interpret the details in the model. This study further shows it is possible to lithologically characterize the material in the model by utilizing a physical relationship between effective seismic velocity and physical rock properties of mineral grains such as bulk/shear modulus, density, Poisson's ratio and porosity. A 1D velocity model is averaged from the 2D model. By inverting the 1D velocity model, 1D profiles of porosity and degree of consolidation are determined, given knowledge of other parameters from published lab experiments and on-site surveys. The porosity ranges from 0.1 to 0.3. The ratio between the local radius and grain radius, which is a measurement of consolidation, varies between 0.1 and 0.55 in the estimated profile.
NS11A-05 09:30h
Natural Gas Hydrates Estimation Using Seismic Inversion and Rock Physics
Gas hydrate drilling worldwide indicates that the formation of gas hydrates in shallow sediments tends to increase P- and S-wave velocities of the hosting rocks. Rock physics models of gas hydrates provide the links between velocity anomalies and gas hydrate concentration. In this abstract, we evaluate the numerical predictions of some of the major rock physics models of gas hydrates and validate those with well log data from the Mallik and Blake Ridge wells. We find that a model in which the gas hydrate is a part of the rock framework produces results that are consistent with well log data. To enhance the accuracy of seismic estimation, we adopt a five-step, integrated workflow that enables us to identify and quantify gas hydrates in the deepwater Gulf of Mexico (GOM). It includes: 1) Reprocessing conventional 3D seismic data at high resolution using an amplitude-preserving flow with prestack time migration, 2) A detailed stratigraphic evaluation to identify potential hydrate zones, 3) Seismic attribute analysis to further delineate anomalous zones, 4) Full waveform prestack inversion to characterize acoustic properties of gas hydrates in 1D (Mallick, 1995; Mallick, 1999) and map in 3D using hybrid inversion techniques (Dutta, 2002; Mallick and Dutta, 2002), and 5) Quantitative estimation of gas hydrate saturation using rock property models. We illustrate the procedure using 3D seismic data, and estimate gas hydrate saturation in the study area in the GOM.
NS11A-06 09:45h
Current Development of Estimation of Near-Surface Elastic Moduli by Analysis of Rayleigh Waves
The shear (S)-wave velocities of near-surface materials (such as soils) are of fundamental interest in many environmental and engineering studies. There are three basic ways to perform shear tests in-situ in soil mechanics: in-situ shear box, shear vane, and penetration. During these tests, either a soil sample must be carefully cut, remolded, or invasive penetrations needed to be performed. Analysis of surface waves offers a non-invasive and cost-effective alternative to rapidly evaluate shear strength. Multichannel Analysis of Surface Waves-MASW, a method that estimates near-surface S-wave velocity from high-frequency (> 2 Hz) Rayleigh waves, has been applied to more and more near-surface problems. The differences between MASW results and direct borehole measurements are 15% or less and random. Several studies have shown that the accuracy and resolution of estimated S-wave velocity can be increased by inverting the Rayleigh wave fundamental mode with higher modes simultaneously. A feasibility study in determining near-surface quality factors had promising results. Acquiring high quality surface-wave data and increasing acquisition efficiency have attracted the attention of the near-surface geophysical community. The MASW method was combined with the standard roll-along acquisition format to generate pseudo-2D S-wave velocity sections. Understanding the resolving power of MASW techniques and improving the resolution of S-wave velocity results were studied. The previous studies mentioned above were all focused on a 1D layered-earth model. We have completed two projects recently: estimating S-wave velocities from Rayleigh waves in a non-layered earth model and modeling high-frequency Rayleigh waves in a 2D earth model. A compressible Gibson half-space is a model of the shear modulus variation linearly with depth. In a half-space of sedimentary granular soil under the geostatic state of initial stress, the density and Poisson's ratio do not vary considerably with depth. In such an earth body, the dynamic shear modulus is the parameter that mainly affects dispersion of Rayleigh waves. An analytical dispersion law of Rayleigh waves in this half-space is in an algebraic form, which makes our inversion processing extremely simple and fast. The main advantage of using this model is that only three Rayleigh wave phase velocities are required in defining this half-space. We developed a scheme using the finite-difference (FD) method to model high-frequency Rayleigh-waves in near-surface mediums. Although many FD programs existed for earthquake research and oil exploration, none have been developed to model high-frequency Rayleigh waves in near-surface elastic mediums with a source and all receivers on the free surface. The scheme used a decoupled system of first-order differential equations. Combined elastic and acoustic free-surface conditions and a combination of one-way sponge filtering and anisotropic filtering methods were implemented to minimize edge effects. Modeling results were proved satisfactory using dispersion analysis of Rayleigh waves in a homogenous or layered half-space. A synthetic seismogram for a 2D corner-edge model matched favorably with dispersion analysis. This method is simple, stable, fast, and accurate and provides a practical tool for improving confidence of 2D interpretation and a basis of 2D inversion. Our future studies will mainly focus on Rayleigh-wave inversion of a 2D subsurface model, determination of dispersion curves with arbitrary geophone settings, and Love-wave inversion.