H42B-01 INVITED
Pore-scale Spectral Induced Polarization (SIP) signatures associated with FeS biomineral transformations
We measured Spectral Induced Polarization (SIP) signatures in sand columns during (1) FeS biomineralization produced by sulfate reducing bacteria (D. vulgaris) under anaerobic conditions, and (2) subsequent biomineral dissolution upon return to an aerobic state. The low-frequency (0.1-10 Hz peak) relaxations produced during biomineralization can be modeled with a Cole-Cole formulation, from which the evolution of the polarization magnitude and relaxation length scale can be estimated. We convert the polarization magnitude to an equivalent FeS surface area per unit pore volume using a previously published empirical relation. The modeled time constant is converted to an equivalent polarizable sphere diameter using a theoretical relation and an assumed value of the surface diffusion coefficient. We find that the modeled time constant is consistent with the polarizable elements being biomineral encrusted pores rather than the biominerals themselves. The temporal SIP signatures suggest FeS surface area increases and pore-size reduction during biomineral growth, with subsequent FeS surface area decreases and pore expansion upon FeS dissolution that occurs during the return of the system to an aerobic state. The geoelectrical interpretation is consistent with changes in aqueous chemistry during the experiment, and solid phase analysis conducted at the termination of the experiment. We conclude that SIP signatures are diagnostic of pore-scale geometrical changes associated with FeS biomineralization by sulfate reducing bacteria. The study highlights the potential for using geoelectrical methods to monitor microbial driven changes in hydraulic conductivity resulting from biomineralization.
H42B-02 INVITED
Estimating hydrological parameters using surface nuclear magnetic resonance (SNMR) tomography
Surface nuclear magnetic resonance (SNMR) takes advantage of precession phenomena that result from exciting groundwater hydrogen protons in the Earth's magnetic field by secondary magnetic fields generated by large surface loops. This method, which allows subsurface water content to be determined from inversions of surface measurements, is commonly employed in a 1-D mode. Conventionally, the same loop is used to generate the secondary magnetic fields and measure the tertiary magnetic fields that result from the excited hydrogen protons precessing back to their equilibrium state. Recent developments provide greater sensitivity in the shallow subsurface and extend the method to true 2-D investigations in which surface topography is correctly accounted for: i) separated transmitter and receiver loops with variable offsets supply significantly improved subsurface coverage and sensitivity, with emphasis on shallow structures, ii) state-of-the-art finite-element-based modelling and inversion techniques allow complex 2-D structures underlying complicated topography to be resolved. In laboratory and borehole NMR applications, it is not only possible to estimate water content, but also hydraulic conductivity. Unfortunately, these well-established small-scale techniques cannot be employed in surface applications. We are in the process of developing and testing new measurement procedures that might eventually allow hydraulic conductivity to be estimated from surface NMR observations.
H42B-03
Estimating hydrogeological parameters in the vadose zone using tomographic GPR first- arrival traveltime data – applications of the eikonal solver within an MCMC-Bayesian inversion framework
Tomographic ground-penetrating radar (GPR) techniques have been widely used for shallow subsurface characterization. One of the common practices using tomographic GPR is to use GPR first-arrival traveltimes for hydrological parameter estimation through hydrogeophysical inversion. Its applicability depends on the accuracy and efficiency of the numerical method which is used to calculate GPR traveltimes, as well as on the robustness and efficiency of the inversion approach. In this study, we adopt a finite-difference method for computing the first- arrival traveltimes by solving the eikonal equation in the celerity domain. This algorithm computes the head and diffraction wave paths and incorporates a fast sweeping method to obtain accurate first-arrival times in complex velocity models. We also introduce a Bayesian inversion framework based on Markov Chain Monte Carlo (MCMC) sampling methods. We test intensively the Bayesian inversion framework, together with the eikonal solver, through infiltration experiments under various flow conditions with different field geometry, soil heterogeneity, and flow initial and boundary conditions. We also evaluate the method by combining GPR first-arrival travletime data with neutron probe data collected at different times during the infiltration experiments. Results show that the inversion approach can effectively estimate the hydraulic parameters of interest and demonstrates the great benefits of coupling different yet complementary information for hydrogeological site characterization.
H42B-04
Multiple-Point Geostatistics and Near-Surface Geophysics for Modeling Heterogeneity in a Coastal Aquifer
In order to effectively manage groundwater resources, water agencies have begun to incorporate precipitation, temperature, stream-gauge, land-cover and groundwater level data into their aquifer models. For Western States in particular, stored groundwater is an important provider for agriculture and human consumption. But the estimates of groundwater quantity are arguably the most uncertain in the water balance equation. Current practice in constructing subsurface models relies on substandard and incomplete data due due, in large part, to budgetary constraints. Once a final model has been developed, the possible inaccuracies in the geological scenarios are rarely examined or investigated. How wrong can the subsurface model be while still giving accurate prediction results? How sensitive is the model response to perturbations in the subsurface parameters and long-term irrigation, precipitation and recharge conditions? This study examines these questions through a sensitivity analysis. The "working" aquifers of California's agricultural central coast were used as analog systems for the construction of this sensitivity study. The fluvial geologic interpretations of these coastal aquifer systems were used in Boolean (object-oriented) and multiple-point geostatistical algorithms to create many alternative permeability fields, reflecting the uncertainty in the spatial distribution and geological scenario of the subsurface permeability field. Two sets of models were created using SNESIM, a multiple-point geostatistical algorithm. SNESIM is able to generate a stochastic facies realization using a training image (TI -- a conceptual idea of geologic system) with rotation and affinity maps. The first set of models are higher entropy, representing less continuous clay layers. These were created from a TI of clay ellipses (which was created using GSLIB Ellipsim program). The second set of models are more heterogeneous by using a fluvial TI within SNESIM. All the realizations were conditioned to the same synthetic drillers' logs. Drillers' log were the most commonly used data in creating the flow models used by water managers. This illustrates the wide range of different subsurface scenarios that can be fit to the incomplete and interpreted lithology information in drillers' logs. The second phase of the study investigates how different temporal and spatial pumping schemes for different permeability fields result in varying seawater intrusion rates and the efficiency of artificial recharge. Transient, multiphase flow simulations were performed in Eclipse on all subsurface realizations. These results emphasize the importance of understanding the heterogeneity at the regional scale. Furthermore, simulation results are shown for different aquitard/aquifer geometries at the coastline, which heavily influence seawater intrusion rates. The third phase of the study uses geophysical forward models to determine if near-surface geophysical techniques are able to differentiate between the different geometries of heterogeneity. This lays the foundation for a cost-benefit analysis of the acquisition of additional subsurface information. Considering the sensitivity study results, what is the benefit of acquiring either well-log information or surface geophysical data? Would this information help in prediction or planning for water-shortage (or water-management) for the different scenarios?
H42B-05
Combining Electrical Techniques to map a Till Aquitard for Quantifying Lateral Flows and Improved Recharge Estimation
Where low permeability layers are present in the unsaturated zone, groundwater recharge can be significantly modified by lateral flows. To improve estimates of the magnitude and spatial distribution of lateral flows, a well defined model of the unsaturated zone hydraulic properties is required. Electromagnetic (EM) surveys, using Geonics EM31 and EM34, along with Electrical Resistivity Tomography (ERT) have been used in the Tern Catchment, Shropshire, UK to determine the distribution of Quaternary glacial deposits above the Triassic sandstone aquifer. The deposits are generally less than 10m thick and comprise low permeability lodgement till and high permeability outwash. Modelling studies have shown the depth and slope of the till surface to be key parameters controlling the magnitude of lateral flows with recharge focussed at the till edge. The distribution of permeability within the till is of secondary importance. The spatial extent of the till is well constrained by EM data and is shown to be continuous. ERT profiles provide data on the depth to the till surface in detailed 2D sections. Combining the two data sets has enabled the depth estimates from the ERT surveys to be extrapolated across a 2D map area. Recharge estimates based on the depth maps take into account lateral flows across the top of the till and show that these flows can contribute significantly to catchment recharge.
H42B-06
Regularized Inversion of Tomographic Pumping Tests Incorporating Geophysical Data Through Smoothing Splines
Although simultaneous analysis of multiple hydraulic tests, or hydraulic tomography, can help reduce the nonuniqueness in aquifer parameter estimates relative to conventional single-test analysis, this inverse problem still needs to be regularized in some sensible fashion to yield plausible parameter estimates. A number of investigators have presented techniques for incorporating geophysical data into hydraulic inverse problems, using the geophysical data to help define a parameter zonation or exploiting quantitative correlations between hydrogeological and geophysical parameter fields. Both of these approaches are complicated by the potentially complex and site-specific relationships between geophysical and hydrogeological properties and the relative scarcity of hydrogeological parameter data from which to estimate these relationships. I present a simple approach to incorporating geophysical information into the analysis of tomographic pumping tests for characterization of the hydraulic conductivity (K) field in an aquifer. The fundamental idea is to augment the objective function for the hydraulic inversion with a regularization term that encourages the estimated K field to reflect the geophysical parameter field(s) in some smooth fashion, but without specifying the exact nature of that relationship in advance of the inversion. In this study, I explore the incorporation of ground penetrating radar data into the inversion of tomographic pumping tests in an alluvial aquifer through a regularization term relating the estimated K distribution to the radar property distributions through a smoothing spline model. The geophysical data considered are vertical profiles of radar velocity and attenuation from zero-offset radar surveys between the two pumping wells used in the hydraulic tomography experiments and a velocity tomogram between these two wells.
H42B-07
Laboratory Study of Quaternary Sediment Resistivity Related to Groundwater Contamination at Mae-Hia Landfill, Mueang District, Chiang Mai Province
This study was aimed to understand the nature of the resistivity value of the sediment when it is contaminated, in order to use the information solving the obscure interpretation in the field. The pilot laboratory experiments were designed to simulate various degree of contamination and degree of saturation then observe the resulting changes in resistivity. The study was expected to get a better understanding of how various physical parameters effect the resistivity values in term of mathematic function. And also expected to apply those obtained function to a practical quantitatively interpretation. The sediment underlying the Mae-Hia Landfill consists of clay-rich material, with interfingerings of colluvium and sandy alluvium. A systematic study identified four kinds of sediment, sand, clayey sand, sandy clay, and clay. Representative sediment and leachate samples were taken from the field and returned to the laboratory. Both the physical and chemical properties of the sediments and leachate were analyzed to delineate the necessary parameters that could be used in Archie's equation. Sediment samples were mixed with various concentration of leachate solutions. Then the resistivity values were measured at various controlled steps in the saturation degree in a well- calibrated six-electrode model resistivity box. The measured resistivity values for sand, clayey sand, sandy clay when fully and partly saturated were collected, then plotted and fitted to Archie's equation, to obtain a mathematical relationship between bulk resistivity, porosity, saturation degree and resistivity of pore fluid. The results fit well to Archie's equation, and it was possible to determine all the unknown parameters representative of the sediment samples. For sand, clayey sand, sandy clay, and clay, the formation resistivity factors (F) are 2.90, 5.77, 7.85, and 7.85 with the products of cementation factor (m) and the pore geometry factors (a) (in term of -am) are 1.49, -1.63, -1.92, -2.24 and the saturation exponents (n) are 2.06, 2.58, 3.52, and 2.46, respectively. These results were used to reinterpret the existing resistivity data of this area. The result of this reinterpretation is a map showing the quantitative distribution of the contaminant plume in the vicinity of the Mae-Hia Landfill.
H42B-08
Determination of Aquifer and Fault Permeabilities by Studying the Poroelastic Response of Rocks Using InSAR in the Upper Coachella Valley Area, CA, USA
Coachella Valley in California is an area enclosed between Little Bernardino, San Jacinto, and San Bernardino Mountains. The San Andreas Fault system runs through the valley. Satellite interferometry (InSAR) has revealed a differential uplift of the land surface over a period of time across two basin-cutting faults, the Banning Fault (BF) and the Garnet Hill Fault (GHF), in the upper Coachella valley region. This uplift is due to a combined effect of both elastic response of the aquifer due to groundwater recharge at the White Water Spreading Facility and tectonic stresses in the region (Wisely et. al., AGU 2006). In this region, faults appear to act as semi-permeable barriers to groundwater flow (Wisely et. al., AGU 2005). As a result, water levels are not restored uniformly throughout the entire aquifer leading to differential uplift of the land surface. Hence, by developing a numerical model of groundwater flow in this aquifer system, that includes hydraulic representations of the existing faults, a better water management strategy can be achieved which aims at restoring the groundwater levels throughout the aquifer. In addition, the proposed model would better constrain fault permeabilities with implications for research related to earthquake dynamics and assessment of potential slip along segments of the San Andreas Fault system. A finite element code is being used to implement the groundwater flow model. The model simulates the poroelastic response of the faulted aquifer to changing groundwater levels. InSAR data and well data collected in this region will be used to constrain the model yielding parameters such as the hydraulic conductivity field of the aquifer and fault system, storage co-efficients, skeletal compressibilities, and the grounwater influx, and outflux, of the basin.