H14C-01 INVITED 16:00h
Bridging the Gap Between Gephysical Measurements and Hydrological Modelling
It has long been recognized that the spatial variability of system properties can dominate hydrological fluxes at all scales. Therefore, modellers have attempted to include heterogeneity in their hydrological models. Commonly, the distribution of key system properties has been determined by inverse modelling on a limited number of point measurements (e.g. pressure heads, water content) or a spatially aggregated flux (e.g. stream flow). Generally, these inverse modelling problems are ill-posed due to an inadequate spatial measurement resolution and/or a low information content of the measured signal with respect to the system properties. Ill-posed problems result in non-unique inversion results and considerable uncertainty in the system property distribution and the simulated hydrological fluxes. Geophysical measurement techniques, such as ground penetrating radar, electrical resistivity tomography, and electromagnetic induction among others, have the unique ability to provide measurements with a spatial resolution that is much higher than conventional measurement techniques. Therefore, it is reasonable to expect that the application of geophysical measurement in hydrology (i.e. hydrogeophysics) should allow the construction of hydrological models that better describe the heterogeneity of the system. Despite the consolidation of hydrogeophysics as a research field, there have been a surprisingly small number of studies where geophysical measurements are actually combined with hydrological modelling. Instead, system characterization has been the main focus of hydrogeophysics in the past years. In this contribution, we provide an overview of methods that can be used to combine geophysical measurements and hydrological modelling. We also discuss the potentials and limitations of both the current generation of hydrological models and selected geophysical measurement techniques for their combined use. Finally, we will attempt to point out future research needs.
H14C-02 INVITED 16:15h
Using Data Collected At Different Scales To Constrain Model Parameters
Models of heterogeneous natural systems are intrinsically scale dependent: as the model resolution changes the model parameters or the entire model structure changes. This presents a challenge for characterization and model parameterization when data from tools or instruments with different resolutions are employed at a site. We present three methods to relate quantities measured on different scales to each other. Typically, these methods are based upon some kind of averaging recipe that describes how smaller-scale information is to be processed into larger-scale data. These relationships can be derived if the mechanics of the measurement process are known. We illustrate these approaches for the case of the slug test and pumping test. The mechanics of many geophysical measurements are, however, not simple and often depend upon factors that are not easily quantified in a mechanistic model. Relationships between different measurements can then be determined empirically through experiments that are best treated in a probabilistic context. Finally, we discuss the relative worth of data on different scales and how these relationships can be used to constrain model parameters in formal inverse procedures, or be used in forward modeling.
H14C-03 16:30h
Cross-Borehole GPR Resolution Analysis for Vadose Zone Imaging Using Coupled Unsaturated Flow and Electromagnetic Modeling
Although Cross-Borehole (XB) Ground Penetrating Radar (GPR) is used extensively in shallow groundwater investigations, few studies have been undertaken to analyze XBGPR resolution and accuracy in a dynamic flow environment. This is an important step as it allows the user to asses the images for artifacts that may result due to data collection or processing procedures. To analyze XBGPR resolution, two-dimensional unsaturated flow modeling was conducted to simulate a water infiltration experiment that was conducted at the Sandia/Tech Vadose Zone (STVZ) site. Using a petrophysical model determined from controlled experiments and well log measurements, the spatial variation in water content provided by flow modeling was converted to dielectric constant and electrical conductivity. These parameters were used in the Finite Difference Time Domain (FDTD) Electromagnetic (EM) forward modeling to simulate the response of cross borehole radar signals. The synthetic data were processed in the same manner as the real data collected in the STVZ site, and the inverted tomographic images were then compared with both the vadose zone model and the results of the unsaturated flow modeling to analyze the XBGPR resolution. The results from the resolution analysis suggest that the configurations for XBGPR tomography in the STVZ site may underestimate the dielectric constant for clay layers thinner than 1 m. Moreover, attenuation within the clay layer is overestimated near the boreholes but underestimated between boreholes in the inverted images synthesized by FDTD modeling. Compared to the attenuation images, inverted water content images of the FDTD modeling are more representative of the input model. The inverted attenuation also shows that artifacts appear in the time-lapse images during fresh-water infiltration. These artifacts are caused by the use of straight-ray inversion to approximate curved-ray, near-field propagation of the EM waves. Thus, it is necessary to properly incorporate physics into the inversion algorithms in order to correctly invert for the attenuation in a non-steady-state environment.
H14C-04 16:45h
A fractional calculus approach to interpreting transient electromagnetic field behavior in near--surface hydrogeophysical investigations
Among the various geophysical technologies that have found a niche in shallow subsurface characterization and monitoring is the electromagnetic induction (EMI) method---a method that maps the spatial variability in ground conductivity arising from lithologic changes and the presence of pore fluids. Interestingly, a growing body of evidence suggests that the electrical structure in some geological settings is inherently hierarchical, and presumed to arise from the dynamical systems that generate and alter the formation and its overlying soil. In these limited cases, the observations challenge the applicability of the standard modeling paradigm which rests squarely on the assumption of a spatially smooth (or at least piecewise smooth) distribution of physical parameters within the subsurface. As an alternative, and drawing upon additional near--surface EMI data recently collected from geologically distinct sites throughout New Mexico and Texas, we investigate the concepts of fractal signals and random walks through spatially-correlated heterogeneous media as a means to describe the observed variations in shallow subsurface EMI response. The fundamental hypothesis examined in this phase of the research is that low-frequency electromagnetic fields penetrate into a multiscale, hierarchical geological medium according to a fractional--order diffusion equation. Such an equation is fully compatible with Maxwell's equations and naturally reduces to the classical limit of integer order derivatives (which seem to be adequate for the vast majority of geological settings) but has the advantage of naturally accommodating generalized constitutive laws and the effects of multi--scale heterogeneity when they arise. Sections of this work were performed at Sandia National Laboratories. Sandia is a multi--program laboratory operated by the Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE--AC04--94AL85000.
H14C-05 17:00h
Processing Ultra-Shallow Reflection Data in Areas with Laterally Varying Velocities
Ultra-shallow seismic reflection methods have proven effective as non-invasive subsurface characterization and monitoring tools, particularly in areas of high electrical conductivity where ground-penetrating radar is ineffective. Hydrologists often rely on geophysical methods to sense the properties of the near surface, at times in shallow layers that exhibit rapid lateral variation caused by saturated and unsaturated sand lenses, clay lenses, and unevenly weathered bedrock formations. Under these conditions, methods traditionally used to collect and process seismic reflection data must be examined for validity prior to application. In this study we present 2-D common-midpoint (CMP) ultra-shallow seismic reflection data collected in a flood plain near a tributary stream channel outside of Lawrence, Kansas. The 20-m long survey line intersected a visually identified abandoned stream channel roughly paralleling the current stream. Near-source first-arrival analysis shows that the P-wave velocities near the surface double over a horizontal span of 15 m, and a $\sim$5 m deep reflection that dominates shot gathers at one end of the survey is not evident in shot gathers at the other end. Such heterogeneity negates the equivalency of pseudo-walkaway and walkaway data, and makes normal-moveout (NMO) based CMP processing difficult. Walkaway surveys are often used in conjunction with CMP surveys to determine seismic velocity. We show that a drastic difference can be encountered when a pseudo-walkaway is substituted for a walkaway in the presence of significant lateral velocity variations. We also compare the final time-to-depth converted stacked sections created when a variety of source-to-receiver offset ranges are NMO corrected and stacked using several different velocity functions.
H14C-06 17:15h
Mapping hydrogeophysical structures with time--domain electromagnetic methods: Resolving small-scale details with large loops and three--component measurements
One of the outstanding problems in managing water resources in geologically complex aquifers is to develop improved techniques for mapping compartmentalization due to faulting. And although the role of faults in aquifer dynamics can vary considerably, knowledge of their location is key to understanding aquifer recharge and developing a sensible model for predicting aquifer response due to anthropogenic loads. We have explored the application of time--domain electromagnetic methods for mapping shallow aquifer faults on the western flanks of the Estancia Basin, central New Mexico. The field site is underlain by massive Pennsylvanian limestones (Madera Group) subsequently faulted by Laramide tectonics of the Ancestral Rockies and Neogene extension of the Rio Grande Rift. Two experimental configurations were deployed: a large $50 \times 40$ m transmitter loop with receiver stations located on a 5 m grid over the loop's interior; and an azimuthal survey consisting of a smaller fixed transmitter with receiver stations at $\sim$2 m intervals along a 30 m radius circle centered on the transmitter. Three--component transients of magnetic field due to a fast linear ramp--off in the transmitter were recorded at each station. As a rapid reconnaisance tool, the azimuthal experiment is well--suited for identification of subsurface fault planes since symmetry constraints require a vanishing azimuthal $\hat\phi$ component of magnetic field when the electrical strike, or fault plane, lies in the $\hat\phi$ direction. However, each of the experimental configurations revealed that the site's electrical structure is far more three--dimensional than previously believed and is not dominated by the response of a previously identified fault plane. Instead, we have observed spatially coherent transient signals which may indicate compartmentalization over length scales as small as a few tens of meters. Sections of this work were performed at Sandia National Laboratories. Sandia is a multi--program laboratory operated by the Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE--AC04--94AL85000.
H14C-07 17:30h
Comparison of Three Geophysical Methods for Characterizing a Gravel Aquifer
Seismic refraction, ground-penetrating radar (GPR) and electrical resistivity tomography (ERT) were used to characterize a gravel aquifer. The aquifer is part of a terrace that is bounded on the North by the Bow River and on the south by the valley wall that consists of lakebed silts. The gravel is underlain by shaley sandstone of the Paskapoo Fm. and has a maxium saturated thickness of 1.5 m. Depth to bedrock is between 6 and 8 m. Borehole logs provided a method for validating the interpretations. Inversion of refraction data yielded a gravel velocity of approximately 800 m/s and the velocity of the underlying bedrock varied between 2400 and 4000 m/s. Refraction inversion provided accurate depth to the gravel-bedrock interface, but it is unclear how to interpret the valley margin. Paskapoo Fm. resistivity was in the range of 17 ohm-m whereas the gravel resistivity was greater than 2000 ohm-m. The gravel-bedrock interface could be interpreted from the GPR data as a strong reflection or as the beginning of the transparent portion of the record due to the strong signal attenuation in the relatively conductive bedrock. However, the interpretation of the gravel-bedrock interface was ambiguous. GPR also imaged structure within the gravel layer that was not visible with the other methods. The ERT results were extremely good due to the strong resistivity contrast between the gravel and the bedrock. In addition, variations from gravel to sandier or siltier regions were well imaged by the ERT. The ERT derived interpretation of the exact depth to bedrock was not as accurate as the interpretations of the GPR or refraction due to the lower resolution of the method. Using all three methods provided an interpretation with detail and reliability not possible using any single method.
H14C-08 17:45h
Borehole Time Domain Reflectometry in Layered Sandstone: Impact of Measurement Technique on Vadose Zone Process Identification
An investigation is reported into the hydraulic behaviour of the vadose zone of a layered sandstone aquifer using borehole-based Time Domain Reflectometry (TDR). TDR has been widely applied to shallow soils but has seen limited application at greater depth and in cemented lithologies due to the difficulty of installing conventional TDR probes in rock and from boreholes. Here, flat TDR probes that are simply in contact with, rather than inserted within the medium under investigation, have been developed and applied in a field study. Both a commercially available portable packer TDR system (TRIME-B3L Borehole Packer Probe) and specially designed TDR probes, permanently installed in boreholes on grouted-in packers were used to monitor seasonal fluctuations in moisture content in the vadose zone of a layered sandstone over one year under natural rainfall loading. The data show that the vadose zone contains seasonal perched water tables that form when downward percolating moisture reaches layers of fine grained sandstone and siltstone and causes local saturation. The formation of perched water tables is likely to lead to lateral flow bypassing the less permeable, finer layers. This contrasts with behaviour inferred from previous studies of the same aquifer that used borehole radar and resistivity, which suggested its vadose zone behaviour was characterized by uniform downwards migration of wetting fronts. To investigate the impact of measurement technique on observed response, the TDR data reported here were used to produce simulated zero offset profile (ZOP) borehole radar responses. This simulation confirmed the limited ability of ZOP borehole radar to detect key vadose zone processes, because the phenomenon of critical refraction minimizes the sensitivity of the results to high moisture content layers. The study illustrates that inappropriate technique selection results in hydrological process mis-identification, with serious consequences for the usefulness of data in hydrological modeling.