Optimal Estimation of Two-Dimensional Contaminant Transport

The state-space estimation technique presented herein provides a method for obtaining optimal estimates of concentrations for two-dimensional plumes in ground water. The concentration of a plume was defined as the state variable. The technique uses the Kalman filter and involves combining two independent estimates of plume concentrations. One estimate is called the process modeling data and the other is called the “measurement” data. The process modeling data is obtained from a numerical model. The “measurement” data is obtained from field measurements; however, for illustration in this paper it was generated by a different numerical model than the one used to obtain the process modeling data. The state-space technique produces a distribution of contaminant concentrations that is more accurate than either of the distributions generated by the process modeling or the “measurement” data. An example is presented to show that the technique produces significant improvements in the prediction of plume concentration distributions.     

Effects of River Stage and Waste Water Discharges on the Unconfined Aquifer, Hanford, Washington

This study examined the effects of river stage and waste water discharge on the unconfined aquifer near the N nuclear reactor on the U.S. Department of Energy-operated Hanford site in Washington State. River levels were statistically correlated with water-level data from 12 wells.
During the course of this study, water table elevations declined in the study area primarily as a result of a significant decrease in discharge to waste water disposal facilities, A minor contributing factor was the regional decline of the water table caused by decreasing waste water discharges upgradient of the study area.
High-frequency river-level fluctuations (e.g., short-term daily fluctuations) had good correlation with water-level variations in a well approximately 750 feet inland. Low-frequency river-level fluctuations (e.g., long-term seasonal fluctuations) had good correlation with water-level variations in a well approximately 1000 feet from the river shore. Time lags and attenuation generally increased with distance from the river as expected, with the exception of two northern wells. These two wells were relatively more responsive to river-level fluctuations at a greater distance inland from the river. This suggests that hydraulic properties (e.g., hydraulic conductivity) are a control on the aquifer reponses.
During peak river stage in June, the river level rose above water table elevations in several wells implying a temporary reversal in ground water flow direction near the river.     

Cost Comparisons of Aquifer Heterogeneity Characterization Methods

The characterization of heterogeneity in hydraulic conductivity (K) is a major challenge for subsurface remediation projects. There are a number of field studies that compare the K estimates obtained using various techniques, but to our knowledge, no field‐based studies exists that compare the performance of estimated K heterogeneity fields or the associated characterization costs. In this paper, we compare the costs of characterizing the three‐dimensional K heterogeneity and its uncertainty estimates of a glaciofluvial aquifer‐aquitard sequence at a 15 m × 15 m × 18 m field site situated on the University of Waterloo campus. We compare geostatistical analysis of high resolution permeameter K data obtained from repacked core samples in five boreholes and hydraulic tomography analysis of four pumping tests consisting of up to 41 monitoring points per test. Aside from the comparison of costs, we also assess the performance of each method by predicting several pumping tests. Our analysis reveals that hydraulic tomography is somewhat more costly than the geostatistical analysis of high resolution permeameter K data due to the higher capital costs associated with the method. However, the equipment may be reused at other sites; hence these costs may be recovered over the life of the equipment. More significantly, hydraulic tomography is able to capture the most important features of the aquifer‐aquitard sequence leading to more accurate predictions of independent pumping tests. This suggests that more robust remediation systems may be designed if site characterization is performed with hydraulic tomography.     

Implications of Subsurface Heterogeneity at a Potential Monitored Natural Attenuation Site

Direct-push sampling was conducted at a site previously characterized with conventional monitoring wells. Hydrogeological and chemical heterogeneities not represented in the original site conceptual model were detected by direct-push sampling. These heterogeneities were important in terms of their impact on the assessment of monitored natural attenuation at the site. This research suggests that additional sampling efforts could be attempted at some sites to test the accuracy of site conceptual models. The leading edge of a contaminant plume must be examined closely, because conceptual errors in this area could easily allow impacts to receptors to remain undetected.     

Simulation of Tidal Effects on Contaminant Transport in Porous Media

A one-dimensional numerical model is developed with oscillating velocities and dispersions to simulate the migration process of a contaminant plume within tidally influenced aquifers. Model simulations demonstrate that a major effect the tidal fluctuation has on the migration process of a contaminant plume is the exit concentration discharging to the tidal estuary. Tidal fluctuation causes the exit concentration levels to be significantly diluted by the surface-water body of the estuary. Sensitivity analyses demonstrate that tidal fluctuation hastens the rate of plume migration near the bank of the estuary because of the relatively high advective and dispersive fluxes induced by tides. However, tides affect the migration process only over a short distance from the tidal-water interface (about 40 ft for the parameters used in this study). If the contaminant plume is located far beyond the interface, tidal fluctuations will not affect the rate of plume migration until an existing regional ground-water flow velocity brings the plume to the tidally active zone. With or without tides, the rate of contaminant migration increases with higher regional hydraulic gradient. Furthermore, the effects of tidal fluctuations on the transport process become insignificant with higher regional hydraulic gradients.     

Techniques and Equipment Used in Contaminant Detection at Hoe Creek Underground Coal Gasification Experimental Site

Data from an existing network of ground water monitoring wells at the U.S. Department of Energy (DOE) Hoe Creek Underground Coal Gasification (UCG) Experimental Site indicated that organic contaminants, particularly phenols produced during gasification experiments, were threatening neighboring ground water resources. The existing monitoring well network was sparse and further definition of the extent and direction of contaminant migration was needed. Additionally, water level data, important in determining flow directions, was incomplete. A field program was designed and implemented to locate and define the organic contamination and expand the existing ground water monitoring program. The program utilized field analysis of phenol for contaminant detection and well location, followed by completion using gas-drive ground water samplers/piezometers. Geophysical logging was used to permit optimum placement of the samplers. The geologic aspects of the site posed some interesting problems to the installation of the samplers. The contaminant plume edge was defined in the east, west and south directions during the field program. Further work is needed in the north direction.     

A comprehensive and systematic approach to developing and documenting conceptual models of contaminant release and migration at the Hanford Site

The U. S. Department of Energys Richland Operations Office has initiated efforts to adapt and implement the features, events, and processes (FEP) methodology used in scenario development for nuclear waste disposal programs to the environmental management and remediation problems facing the Hanford site. These efforts have shown that modification of the FEPs methodology to incorporate the use of process relationship diagrams (PRD) is effective in facilitating the development of conceptual models and selection of potentially relevant factors (i.e., FEPs) to be incorporated into a specific environmental assessment. In adopting this methodology for Hanford, a master PRD was created to provide a structure to identify these factors and to illustrate the relationships among them. The organizational framework of the master PRD was developed to match the organization of current Hanford site-wide environmental assessment activities and to facilitate screening of the FEPs relevant to the specific assessments needed for the site.     

Vertical Contaminant Profiling of Volatile Organic* in a Deep Fractured Basalt Aquifer

Volatile organic compounds delected in ground water from wells at Test Area North (TAN) at the Idaho National Engineering Laboratory (INEL) prompted RCRA facility investigations in 1989 and 1990 and a CERCLA-driven RI/FS in 1992. In order to address ground water treatment feasibility, one of the main objectives, of the 1992 remedial investigation was to determine the vertical extent of ground water contamination, where the principle contaminant, of concern is trichloroethylene (TCE). It was hypothesized that a sedimentary interbed at depth in the fractured basalt aquifer could be inhibiting vertical migration of contaminants to lower aquifers. Due to the high cost of drilling and installation of ground water monitoring wells at this facility (greater than $100,000 per well), a real time method was proposed for obtaining and analyzing ground water samples during drilling to allow accurate placement of well screens in zones of predicted VOC contamination. This method utilized an inflatable pump packer pressure transducer system interfaced with a datalogger and PC at land surface. This arrangement allowed for real lime monitoring of hydraulic head above and below the packer to detect leakage around the packer during pumping and enabled collection of head data during pumping for estimating hydrologic properties. Analytical results were obtained in about an hour from an on-site mobile laboratory equipped with a gas chromalograplvmass spectrometer (GC/MS). With the hydrologic and analytical results in hand, a decision was made to either complete the well or continue drilling to the next test zone. In almost every case, analytical results of ground water samples taken from the newly installed wells closely replicated the water quality of ground water samples obtained through the pump packer system.     

Multidimensional Investigation of Bedrock Heterogeneity/Unconformities at a DNAPL‐Impacted Site

Organic solvent (i.e., dense nonaqueous phase liquid, DNAPL) migration in the subsurface is known to be extremely sensitive to geologic heterogeneity. There is often a focus on heterogeneity that results from changing depositional conditions over short spatial scales. Similar or even more extreme spatial heterogeneity can result postdeposition due to erosional processes. This study applies a synergistic approach based on a combination of high‐resolution lithologic logs of continuous cores, borehole geophysical logs, surface electrical resistivity, and seismic refraction tomography models to assess spatial heterogeneity in a shallow bedrock sequence subject to multiple unconformities and contaminated with a mixture of organic chemicals. The persistence of DNAPL in the source zone and an associated dissolved‐phase plume led to variable impacts on formation resistivity across the study site. Seismic refraction in combination with electrical resistivity tomography improved interpretation of highly irregular erosional boundaries by delineating sharp lateral transitions in lithologic composition near the source zone and across the dissolved‐phase plume. Electrical resistivity was effective at differentiating between clean and mud‐rich sandstones and their unconformable contact with an underlying dolostone. Geophysical measurements revealed eroded dolostone mounds encased by a network of younger mud‐rich sandstones channelized by clean semi‐lithified sand, all of which was buried beneath variable glacial drift. Our synergistic multidimensional approach resulted in the development of a detailed three‐dimensional shallow bedrock geospatial model, which has led to an improved understanding of DNAPL migration and contaminant plume heterogeneity.     

Persistence of a Groundwater Contaminant Plume after Hydraulic Source Containment at a Chlorinated‐Solvent Contaminated Site

The objective of this study was to characterize the behavior of a groundwater contaminant (trichloroethene, TCE) plume after implementation of a source‐containment operation at a site in Arizona. The plume resides in a quasi‐three‐layer system comprising a sand/gravel unit bounded on the top and bottom by relatively thick silty clayey layers. The system was monitored for 60 months beginning at start‐up in 2007 to measure the change in contaminant concentrations within the plume, the change in plume area, the mass of the contaminant removed, and the integrated contaminant mass discharge (CMD). The concentrations of TCE in groundwater pumped from the plume extraction wells have declined significantly over the course of operation, as have concentrations for groundwater sampled from 40 monitoring wells located within the plume. The total CMD associated with operation of the plume extraction wells peaked at 0.23 kg/d, decreased significantly within 1 year, and thereafter began an asymptotic decline to a current value of approximately 0.03 kg/d. Despite an 87% reduction in contaminant mass and a comparable 87% reduction in CMD for the plume, the spatial area encompassed by the plume has decreased by only approximately 50%. This is much less than would be anticipated based on ideal flushing and mass‐removal behavior. Simulations produced with a simplified three‐dimensional (3D) numerical model matched reasonably well to the measured data. The results of the study suggest that permeability heterogeneity, back diffusion, hydraulic factors associated with the specific well field system, and residual discharge from the source zone are all contributing to the observed persistence of the plume, as well as the asymptotic behavior currently observed for mass removal and for the reduction in CMD.