NS33A-01 INVITED
The Role of Uncertainty and Scale in Rock Physics Transforms on Sequential and Integrated Data Fusion Methods
Geophysical data are increasingly being used to address complex problems across disciplinary boundaries. Rock physics provides a bridge across these boundaries by relating the properties that control geophysical measurements, e.g., electrical conductivity, to properties that are of fundamental interest to a discipline specialist, e.g., solute concentration. However, as geophysical applications become increasingly sophisticated so do the issues that complicate data fusion, such as discrepancies between scales of interest and uncertainty due to geologic heterogeneity. Therefore, understanding the impact of uncertainty and scale on rock physics transforms is a critical problem. Recent advances have been made to account for the upscaling of rock physics relationships in heterogeneous environments, e.g., Full Inverse Statistical Calibration (FISt). In the FISt approach it is possible to cast rock physics relationships within a stochastic framework, thereby allowing one to investigate how different factors, such as prior geologic models, measurement non-linearity, and process uncertainty, ultimately impact this transform. By breaking down a rock physics relationship into these different components it is also possible to make informed decisions regarding the appropriate transform to be used in alternate approaches to geophysical data integration, such as sequential versus integrated approaches to data fusion. In the sequential approach geophysical data are collected, inverted to a geophysical image, and then transformed to, for example, hydrologic properties that can subsequently be used as constraints in hydrologic estimation problems. In this case, scale discrepancies related to model resolution are a critical issue in selecting the appropriate rock physics relationship. In contrast, the integrated approach to data fusion avoids the geophysical imaging step by linking the geophysical data directly to hydrologic properties through process-based models. Therefore, the scale of concern is that of the measurement resolution rather than model resolution. Ultimately, rock physics relationships are critical to the design of any data fusion algorithm and should be carefully selected to account for the issues specific to the particular estimation problem under consideration.
NS33A-02
Accuracy and Precision of GPR Velocity Models Obtained From Semblance Analysis of CMP Gathers
Interest is growing in the use of ground penetrating radar (GPR) methods for quantifying subsurface properties (e.g. porosity, water content) derived from EM wave interval velocity, vINT. This velocity is usually calculated from Dix's Equation with velocity and time picks obtained from semblance analysis of reflection moveout times in common midpoint (CMP) data. However, this process leads to imprecision and inaccuracy in velocity estimates (and resulting estimates of subsurface properties) for 3 reasons. (1) The CMP geometry (through the range of offsets and thus moveout times) controls observational error and hence imprecision in `root-mean-square velocity', vRMS. (2) Dix's Equation only delivers true vINT when input velocities are vRMS. The latter are not recovered from actual, non-hyperbolic, moveout times; instead they define `stacking velocities', vST, which in turn yield a systematically inaccurate `interval stacking velocity', vIS. (3) Peak semblance response is to the maximum amplitude of the GPR wavelet, typically the second or third half-cycle, rather than first-break travel-times. This delay from first-break results in a further systematic bias of vST to slower values. GPR velocity analyses rarely recognise these issues; this paper evaluates the severity of the problem, and suggests field procedures and remedial measures to minimise its consequences. Analyses of synthetic data show that vST uncertainty is <1% if a reflection exhibits a ratio of [moveout/wavelet period] of >7.5. A CMP acquisition must contain traces to sufficiently far offset to fulfil this criterion. Such a vST is then used to predict the vST that would be recovered from near- offset traces, which in turn more accurately matches vRMS. Finally, in order to simulate first-break times, synthetic travel-times generated from near-offset vST are subtracted from the (assumed constant) delay between the wavelet first-break and its maximum amplitude. For 50 MHz mixed-phase energy (delay = 9.4 ns) reflected at t0 = 200 ns in an overburden velocity of 0.1 m/ns, simulation of first-break travel-times reduces the discrepancy between vST and vRMS from -16.0% to +0.5%. The initial error in vRMS reduces as the delay to maximum amplitude becomes a smaller fraction of t0 (e.g. shorter wavelet period and/or longer t0). For a glaciological field example, this approach modifies a water content estimate from 1.71% to 1.18% (-0.38, +0.15), due to improved accuracy and precision of vINT.
NS33A-03
An airborne hydrogeophysical study of San Christobal, Galapagos Islands
Airborne electromagnetic methods are practical tools for large scale hydrogeophysical mappings. In this
presentation we show initial results from a study on the volcanic island of San Christobal, one of the islands in
the Galapagos archipelago. The survey is an essential part of a project aiming on characterizing the entire
hydrological system on this island and the island of Santa Cruz. The island has a serious growing lack of quality
drinking water. A total number of about 800 km of time domain electromagnetic data were measured using the
helicopter borne system, SkyTEM. The SkyTEM system is specially developed for high resolution surveys, the first
time gate is measured at 15 micros and the maximum magnetic moment is approx. 105 kAm2. Data does not
require levelling. The data were processed using state of the art filtering techniques and inverted using a newly
developed 3D sharp boundary inversion algorithm with locally 1D derivatives. We discuss pros and cons of the
applied methodology and present results from the survey. The results clearly reveal the saltwater boundary and
outlines distinct structural differences between the western and central zone in relation with their different
hydrological characteristics.
http:www.hgg.au.dk
NS33A-04
GPR Profiling Of Lacustrine Stratigraphy With Embedded Targets: A Case History of UXO Detection
We used ground-penetrating radar, mainly at 135 MHz, to profile the subbottom sedimentation of a small lake in south central New Hampshire, and to detect and characterize likely unexploded ordnance lying upon and embedded within it. The lake itself had been a bombing range for a variety of munitions. Maximum water depth was 8.5 m. The low water conductivity, flat surface and the near perfect antenna contact with the water provided ideal electromagnetic survey conditions. Small percentages of expandable clays were probably responsible for up to 8 m of subbottom penetration into deltaic deposits and to bedrock beneath deeper basin deposits. Penetration into bedrock itself is also evident. Comparative surveys on non hazardous Squam Lake in New Hampshire verified the geological nature of the strata and of false targets. Likely UXO were detected by the unique phase signature of their diffractions and exceptional amplitudes. Controlled experiments verified phase consistency of the diffractions from all angles and polarizations. We used wave migration to identify target type where target density was too great to exploit wavelet phase. We interpret perforated horizons underlain by diffractions having appropriate phases to be evidence of craters and UXO. Most false targets appear to be trees, most likely drowned when the lake was dammed in the mid nineteenth century, and blown down by detonations along the shores. We conclude that lake sediments in resistive water are ideal media for detection of UXO and that target identification could be improved with cross polarization profiles.
NS33A-05
Ground Penetrating Radar Response Over Coal Seam Discontinuities
Coal is one of the most important sources of energy. There are a number of problems associated with the mining of coal; one of them is the detection of unknown structural features encountered in coal measures. The commonly occurring structural features include fault, wedge out, seam splitting, dykes, cavities both water and air filled. The geophysical techniques, which are extensively used by the coal mining industry, are surface reflection seismic, dipole-dipole resistivity and geophysical borehole logging. The resolution of GPR for the shallow subsurface features is higher and the survey is easier to carry out in comparison of other methods. In this paper we map the coal seam discontinuities with the help of GPR data. GPR response over coal seam discontinuities viz., fault, wedge out, seam splitting, dykes, cavities both water and air filled, prevailing in Raniganj and Jharia coalfields has been obtained using GPRMAX (ver 1.5) software. The source pulse is a Ricker wavelet of frequency 50 MHz and sampling interval is 0.23587 nanosec. Here we use the relative permittivity values as 3.0 for dry sandstone, 6.0 for coal, 10.0 for shale and 80.0 for water whereas the relative permeability is assumed to be 1.0 for all the media. The conductivity values are taken as 0.000010 S/m for dry sandstone, 0.001S/m for coal, 0.01 S/m for shale and 0.1 S/m for water. The obtained response is contaminated with diffraction noise. Time migration has been carried out with help of MATLAB tools, to overcome the diffraction noise present in the radargram. The obtained GPR response after time migration helps in identifying these coal seam discontinuities prevalent in Raniganj and Jharia coalfields of India.
NS33A-06
Small-Scale Structure of the Sea Ice Near Barrow, Alaska
A capacitively coupled resistivity survey was conducted on the sea ice near Barrow, Alaska in March of 2006. The purpose of this survey was to determine the effectiveness of a capacitively coupled resistivity array in determining the small-scale structure of the ice. This was new ice, less than 2 months old, and was ~100m offshore from the Naval Arctic Research Laboratory near Barrow. The data was obtained along both a single 300m line as well as in a 7m by 100m grid. With the ice being ~1-2m thick, unusually small n-spacings (dipole-dipole array) were used, the smallest of which was n/4. This allowed 5-6 layers of data to be obtained in the ice. The data inversion using RES2DINV was able to model the ice as shallow as ~20cm. The data from the 7mx100m grid has been inverted and shows the 3-d structure of the ice. Further examination of the inverted data has been performed in order to determine any fractal dimension present for the underside of the ice. The results of this analysis will be reported.
NS33A-07
Mapping Near Surface Heterogeneities for Geotechnical and Environmental Applications Using Seismic Surface-Waves
Surface wave processing and inversion is a non-destructive method often preferred in near surface studies for estimating shear-wave velocities as a function of depth. In processing, changes in phase as a function of frequency are used to compose single or multi-modal dispersion curves for subsequent interpretation. In multi- channel processing, frequency-wave-number (f-k) or frequency-ray-parameter (f-p) transformations followed by interpretation produces these curves. In single channel-processing, a single dispersion curve is generated by cross-correlating data from receiver pairs at several receiver separations. In either case, phase velocities are inverted through iterative minimization to produce shear-wave velocities as a function of depth. A straightforward extension of the analysis for predicting lateral variations of velocity consists in application of 1D inversion at several locations within some area of interest followed by interpolation in multiple dimensions. Of course this procedure would neglect or treat as noise any wave interaction resulting from lateral heterogeneities due to changes in the elastic properties occurring in the near surface. The scales of the heterogeneities would need to be comparable to those of the recorded wavelengths in order for the data to be sensitive to lateral changes. A possible means to detect the heterogeneities would consist in matching synthetic seismograms computed from a detailed subsurface model with recorded seismic data through waveform inversion. Such a hypothetical inversion would require narrowly constrained parameters at every cell point of a discretized target region as well as a forward modeling method capable of reproducing complex wave propagation phenomena. This procedure is not viable if there is inadequate a priori information to constrain the inversion, and it would be too costly in terms of computation time. An alternative is an interpretative approach that relies on mapping phase and amplitudes from surface wave data to a smooth velocity background model superimposed with local perturbations. The work presented here summarizes inversion of surface wave data in 1D and then discusses the benefits of a staged interpretation procedure for mapping lateral near surface heterogeneities. The proposed methodology has important applications for geotechnical and environmental engineering investigations.