Sampling of an emplaced creosote source installed below the water table at CFB Borden was conducted over a period of ten years, with over nine thousand samples taken from approximately 250 multilevel samplers. This extensive dataset was used in several attempts to model the multi-chemical plumes emanating from this emplaced source, and to further understand the chemical and biological processes affecting these plumes and their natural attenuation. An aerobic microcosm study of naphthalene, 1-methylnaphthalene and acenaphthene was conducted in order to determine the possibility of interactions between these three chemicals. All three chemicals degraded within the eight days of the study, and the degradation of naphthalene and 1-methylnaphthalene was not affected by the presence of any of the three chemicals studied. Acenaphthene degraded more quickly when naphthalene was present in the microcosm. The programs Visual MODFLOW and RT3D were used to model the transport and degradation of naphthalene at CFB Borden. Both a first order rate reaction module and a multiple electron acceptor reaction module were used, and contaminant mass was introduced to the model through a fence of observed concentrations. Good results were found at early time with the multiple electron acceptor reaction package, however at late time the model did not match to observations. The program BIONAPL/3D was used in a similar attempt to model the transport and degradation of naphthalene. Naphthalene mass was introduced to the model through a fence of observed concentrations, and multiple electron acceptors were used to degrade this chemical. Results were good at early time, but at late time the model did not match observations. BIONAPL was then used to simulate the dissolution of the original source NAPL. Several chemicals of interest were examined: naphthalene, m-xylene, 1-methylnaphthalene and acenaphthene. Naphthalene and m-xylene dissolved from the source at rates similar to observations, however the dissolution of 1-methylnaphthalene and acenaphthene was not as well modeled. As with the Visual MODFLOW model, the BIONAPL model which best matched observations generally worked well at early times, but did not at late times. The models were not able to successfully simulate many processes that occur in the field, such as chemical and biological interactions and NAPL source dissolution. Mismatches between the models and observations are likely due to these reasons. It may be that we do not fully understand these processes, so we are unable to model them.

La modelación numérica del flujo de agua subterránea y del transporte de masa se está convirtiendo en un criterio de referencia en la actualidad para la evaluación de recursos hídricos y la protección del medio ambiente. Para que las predicciones de los modelos sean fiables, estos deben de estar lo más próximo a la realidad que sea posible. Esta proximidad se adquiere con los métodos inversos, que persiguen la integración de los parámetros medidos y de los estados del sistema observados en la caracterización del acuífero. Se han propuesto varios métodos para resolver el problema inverso en las últimas décadas que se discuten en la tesis. El punto principal de esta tesis es proponer dos métodos inversos estocásticos para la estimación de los parámetros del modelo, cuando estos no se puede describir con una distribución gausiana, por ejemplo, las conductividades hidráulicas mediante la integración de observaciones del estado del sistema, que, en general, tendrán una relación no lineal con los parámetros, por ejemplo, las alturas piezométricas. El primer método es el filtro de Kalman de conjuntos con transformación normal (NS-EnKF) construido sobre la base del filtro de Kalman de conjuntos estándar (EnKF). El EnKF es muy utilizado como una técnica de asimilación de datos en tiempo real debido a sus ventajas, como son la eficiencia y la capacidad de cómputo para evaluar la incertidumbre del modelo. Sin embargo, se sabe que este filtro sólo trabaja de manera óptima cuándo los parámetros del modelo y las variables de estado siguen distribuciones multigausianas. Para ampliar la aplicación del EnKF a vectores de estado no gausianos, tales como los de los acuíferos en formaciones fluvio-deltaicas, el NSEnKF propone aplicar una transformación gausiana univariada. El vector de estado aumentado formado por los parámetros del modelo y las variables de estado se transforman en variables con una distribución marginal gausiana. / Zhou ., H. (2011). Stochastic Inverse Methods to Identify non-Gaussian Model Parameters in Heterogeneous Aquifers [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/12267 / Palancia

Marx Creek, near Hyder in southeast Alaska, is a groundwater-fed, artificial salmon-spawning stream that was constructed to enhance the habitat of the atypically large chum salmon. The success of the upper Marx Creek has been limited primarily by the infiltration of silty water from the Salmon River through its flood-control dike, which results in a turbid stream environment that is not conducive to salmon spawning. The purpose of this project was to determine whether baseflow from the groundwater system is sufficient to support a proposed 1,000-foot extension of Marx Creek. The extension would be constructed approximately 500 feet east of the existing channel, and would connect with the existing Marx Creek at a point downstream of the sediment-settling stream cell. The location of the new channel would prevent the turbid water from reaching the new channel, as it would flow into and settle out in the existing Marx Creek. In order to accomplish this purpose, 20 monitoring wells were installed. Water level data were collected in each of the monitor wells. Slug and pumping tests were performed to determine the hydraulic conductivity at each of the well locations. Discharge measurements were also collected in July 2006 and July 2007. These data were used to create a three-dimensional groundwater flow model using Visual MODFLOW. The model was calibrated to hydraulic head measurements and Marx Creek discharge. After achieving model calibration, three predictive simulations were run. In the first simulation the proposed extension was added to the calibrated model. The result was that baseflow to the extension significantly exceeded baseflow to Marx Creek, and that the addition of the proposed extension reduced baseflow to Marx Creek by 17%. In the second simulation, Marx Creek was removed from the model, while the proposed extension remained. The result was that discharge in the extension increased by 5%. In the third simulation a 1.06-foot drop in the model's hydraulic head was simulated, and the result was that discharge in the extension decreased by 18%. Based on these results, it is likely that baseflow to the proposed extension would be sufficient to provide habitat conducive to salmon spawning.

This thesis describes a new method to create three-dimensional finite element meshes using the horizons to mesh algorithm. The algorithm uses available geologic data and user-defined inputs to guide the mesh generation process. This new approach allows for material layer pinch outs and many different layer refinement options to create well-formed elements that better represent hydrogeologic formations. Two case studies are presented that demonstrate the application of the algorithm's options and capabilities. A graphical interface for the algorithm was developed in the Groundwater Modeling System.

Fluvial flooding has long been recognized as one of the most frequently occurring natural disasters worldwide, with consequences as large economic losses from damages on infrastructure and agriculture, as well as severe impacts on human health. A less known and explored type of flooding is groundwater flooding. A flood type that for instance can arise in river-connected floodplains when groundwater levels rise to the ground surface due to increased river stages in the watercourse. Although groundwater flooding in general is a poorly understood phenomenon, it has become more recognized since its inclusion in the European Floods Directive (2007/60/EC) in 2007. Sweden has however excluded pure groundwater flooding as a separate flood type in its interpretation of the directive, but recognizes groundwater as a component which together with soil water and river water can influence the appearance of a flood event. One of the difficulties regarding groundwater floods that occur in connection to a river is that they typically are hard to differentiate from inundations of fluvial or pluvial origin. It is however important to address the role of groundwater in the inundation of these settings, since traditional flood protection strategies like levees might be circumvented by flows through the subsurface. The aim of this study has been to investigate the role of groundwater in the flooding of a river-connected floodplain by setting up a groundwater model in the integrated hydrological modeling tool MIKE SHE and couple it to an existing MIKE 11 river model, developed by DHI. The study area is a floodplain located along the river Silverån, a tributary to the river Emån, located in the south eastern part of Sweden. By running the model using four different sub-scenarios, regarding initial groundwater level and amount of precipitation, flood extent and contribution of groundwater to the inundation, in relation to other flood sources, has been investigated for different river discharges. A scenario with artificial levees constructed along parts of the river was also examined as levees have been found to have little effect on groundwater floods. As the model provides a simplified and generalized representation of reality it possesses several uncertainties, and so does the results. In summary, the results are in line with what is stated in the Swedish interpretation of the European Floods directive. It has not been possible to demonstrate pure groundwater flooding, but the results suggest that an elevated groundwater level in the beginning of a flood event will increase the extent of the inundation and result in a larger contribution of groundwater to the total amount of flood water. This suggests that there, in some cases, might be a value in integrating groundwater processes in flood risk mapping. Something that is not included in the conventional hydraulic 1D and 2D models, which traditionally are used in flood mapping. As could be expected, the results indicate that groundwater only accounts for a minor part of the flood water added to the total floodplain, while the major sources are river water and surface runoff. A delimited floodplain section that was investigated more in detail, as an increased flow from groundwater to overland water was detected along it, did however show larger contributions from groundwater. This river reach was less vulnerable to fluvial flooding, which in total resulted in a less severe flood, but also enabled a larger amount of groundwater to seep up to the floodplain surface. These conditions did also result in that the river section experienced a worsened inundation at the sub-scenario of high precipitation and high initial groundwater level, as levees were constructed along the river. Most likely because a lot of surface runoff, otherwise able to drain to the river along this section, got trapped outside the levees since it was unable to drain both to the river and to the saturated ground. These results support the theory that levees have little impact on groundwater flooding and stresses the importance 0f surveying and understanding the governing processes in the inundation of a floodplain when planning which type of flood protection scheme to use. / Översvämning utmed vattendrag, så kallad fluvial översvämning, har länge varit känd som en av de vanligast förekommande naturkatastroftyperna världen över, med konsekvenser i form av stora ekonomiska förluster, skador på infrastruktur och jordbruk samt allvarlig påverkan på människors hälsa. En mindre känd och utforskad översvämningstyp är grundvattenöversvämning. En typ av översvämning som kan uppstå i svämplanet längs ett vattendrag då grundvattennivån går upp i markytan till följd av förhöjda nivåer i vattendraget. Trots att grundvattenöversvämning generellt sett är ett outforskat fenomen har det blivit mer uppmärksammat sedan det inkluderades i det europeiska översvämningsdirektivet (2007/60/EG) som antogs 2007. I Sverige har man dock valt att exkludera renodlade grundvattenöversvämningar ur sin tolkning av direktivet och sagt att sådana inte förekommer i Sverige. Istället ser man grundvattnet som en av delarna i ett samverkande system, där det tillsammans med markvatten och ytvatten kan ha påverkan då ett vattendrag översvämmas. En svårighet med grundvattenöversvämningar som inträffar i anslutning till vattendrag är att de kan vara svåra att skilja från översvämningar med fluvialt eller pluvialt ursprung. Det är dock viktigt att uppmärksamma grundvattnets roll i den här typen av översvämningar då traditionella åtgärder som sätts in mot översvämningar, såsom invallningar, kan kringgås av flöden genom marken. Syftet med den här studien har varit att undersöka grundvattnets roll vid en översvämning utmed ett vattendrag genom att konstruera en grundvattenmodell i det integrerade hydrologiska modellverktyget MIKE SHE och koppla denna till en befintlig MIKE 11 vattendragsmodell, utvecklad av DHI. Modellområdet som studerats är beläget längs Silverån, ett av biflödena till Emån i sydöstra Sverige. Genom att undersöka fyra olika delscenarion, avseende initial grundvattenyta och nederbördsmängd, har översvämningsutbredning samt grundvattnets bidrag till översvämningen utvärderats för olika vattenflöden. Ett scenario där invallningar konstruerats längs delar av vattendraget har också undersökts, eftersom invallningar visat sig ha begränsad effekt på grundvattenöversvämningar. Eftersom modellen utgör en förenklad och generaliserad representation av verkligheten har den flertalet osäkerheter, något som även gäller för modellresultaten. Sammanfattningsvis kan sägas att resultaten är i linje med den svenska tolkningen av det europeiska översvämningsdirektivet. Det har inte varit möjligt att påvisa renodlade grundvattenöversvämningar. Däremot pekar resultaten på att en förhöjd grundvattennivå under inledningen av ett översvämningstillfälle kommer att bidra till en ökad översvämningsutbredning, samt ett större bidrag av grundvatten till den totala mängden översvämningsvatten. Detta indikerar att det i vissa fall skulle kunna finnas en mening i att inkludera grundvattenprocesser vid översvämningskartering. Något som inte finns med i de konventionella hydrauliska 1D- och 2D-modeller som traditionellt används vid översvämningskarteringen. Som väntat visar resultaten på att grundvattnet står för en mycket liten del av det vatten som totalt översvämmar det undersökta svämplanet, och att de främsta källorna är vatten från vattendraget tillsammans med ytavrinning. Längs en avgränsad sträcka av svämplanet som undersöktes mer i detalj, då ett ökat flöde från grundvatten till vatten på markytan påträffades längs denna, återfanns dock ett större bidrag från grundvattnet. Denna del av svämplanet var mindre känsligt för fluvial översvämning, något som på det hela taget resulterade i en mindre allvarlig översvämning, men också tillät en större mängd grundvatten att tränga upp på markytan. Dessa förhållanden ledda också till att den aktuella delen av svämplanet kom att få en förvärrad översvämning då vallar konstruerades för delscenariot med hög nederbörd och initialt hög grundvattenyta. Detta till följd av att en stor mängd ytavrinning, som tidigare kunnat dräneras till den här delen av vattendraget, fastnade utanför vallarna istället för att avledas till vattendraget eller infiltrera den mättade marken. Dessa resultat kan sägas stödja teorin kring att invallningar har liten påverkan på grundvattenöversvämningar och visar på vikten av att undersöka och förstå styrande processer kring översvämningen av ett svämplan då åtgärder mot översvämning planeras.

Passive site remediation is a new concept proposed for non-disruptive mitigation of liquefaction risk at developed sites susceptible to liquefaction. It is based on the concept of slow injection of stabilizing materials at the edge of a site and delivery of the stabilizer to the target location using the natural groundwater flow. The purpose of this research was to establish the feasibility of passive site remediation through identification of stabilizing materials, a study of how to design or adapt groundwater flow patterns to deliver the stabilizers to the right place at the right time, and an evaluation of potential time requirements and costs. Stabilizer candidates need to have long, controllable gel times and low viscosities so they can flow into a liquefiable formation slowly over a long period of time. Colloidal silica is a potential stabilizer for passive site remediation because at low concentrations it has a low viscosity and a wide range of controllable gel times of up to about 100 days. Loose Monterey No. 0/30 sand samples (Dr = 22%) treated with colloidal silica grout were tested under cyclic triaxial loading to investigate the influence of colloidal silica grout on the deformation properties. Distinctly different deformation properties were observed between grouted and ungrouted samples. Untreated samples developed very little axial strain after only a few cycles and prior to the onset of liquefaction. Once liquefaction was triggered, large strains occurred rapidly and the samples collapsed within a few additional cycles. In contrast, grouted sand samples experienced very little strain during cyclic loading. What strain accumulated did so uniformly throughout loading and the samples remained intact after cyclic loading. In general, samples stabilized with 20 weight percent colloidal silica experienced very little (less than two percent) strain during cyclic loading. Sands stabilized with 10 weight percent colloidal silica tolerated cyclic loading well, but experienced slightly more (up to eight percent) strain. Treatment with colloidal silica grout significantly increased the deformation resistance of loose sand to cyclic loading. Groundwater and solute transport modeling were done using the codes MODFLOW, MODPATH, and MT3DMS. A "numerical experiment" was done to determine the ranges of hydraulic conductivity and hydraulic gradient where passive site remediation might be feasible. For a treatment are of 200 feet by 200 feet, a stabilizer travel time of 100 days, and a single line of low-head (less than three feet) injection wells, it was found that passive site remediation could be feasible in formations with hydraulic conductivity values of 0.05 cm/s or more and hydraulic gradients of 0.005 and above. Extraction wells will increase the speed of delivery and help control the down gradient extent of stabilizer movement. The results of solute transport modeling indicate that dispersion will play a large role in determining the concentration of stabilizer that will be required to deliver an adequate concentration at the down gradient edge. Consequently, thorough characterization of the hydraulic conductivity throughout the formation will be necessary for successful design and implementation of passive site remediation. The cost of passive site remediation is expected to be competitive with other methods of chemical grouting, i.e. in the range of $60 to $180 per cubic meter of treated soil, depending on the concentration of colloidal silica used. / Ph. D.

A numerical model is presented to describe solute transport in groundwater coupled to sorption by plant roots, translocation into plant stems, and finally evapotranspiration. The conceptual model takes into account both Root Concentration Factor, RCF, and Transpiration Stream Concentration Factor, TSCF for chemicals which are a function of Kow. A similar technique used to simulate the solute transport in groundwater to simulate sorption and plant uptake is used. The mathematical equation is solved using finite difference technique to solve for the concentration at any grid cell with respect to time. The new package is integrated into SEAM3D to create a new SEAM3D Plant Uptake Package, or PUP. The model is then verified by comparing results for root sorption in one side to the SEAM3D Reaction Package, and results for plant uptake to the SEAM3D Source Sink Mixing Package. The verification results showed an excellent match, which led to using the new package in a series of design application scenarios to evaluate phytoremediation effect. Hypothetical design scenarios included: 1) the effect of a phytoremediation system dimensions, 2) the effect of phytoremediation plant density or maximum ET rate, 3) the effect of out-flux of the phytoremediation with respect to the natural aquifer in-flux, and 4) the effect of using a phytoremediation system when the source of contamination is removed. For all the previous study cases, the results evaluate the effect on: 1) contaminant concentrations downstream the source (expressed in plume length at a concentration 1% of the source concentration), 2) solute mass removal from the aquifer, and 3) mass-flux changes at different cross-sections downstream the contaminant source. The results indicating the followings: 1) the width of the phytoremediation system, WET, has a limited effect on the solute mass-removal; 2) high tree density close to the contaminant source has a greater effect on solute mass removal relative to uniform density of trees planted over the entire plume; 3) the width of the ET area will have only a slight effect on the mass removal if the TSCF value is small; 4) as the value of TSCF gets lower, the efficiency of solute mass uptake is lower, and thus the solute concentration in groundwater is higher regardless of the quantity of water transpired; 5) dynamic steady-state plume dimensions (specially the plume length) are affected by the groundwater in-flux, which will control the dimensions and density of a phyto system; 6) splitting the phyto system into two halves does not have the same outcome of having one piece of area closer to the contamination site; 7) using a phyto system after the contamination source is removed led to increasing the solute concentration in the areas of the trees and decreases the concentration in the areas downstream the trees. The alternative model gives more options for simulation of solute mass uptake by plants by making use of field and lab data between the solute dissolved concentration in groundwater C, and solute mass in tree's core M to select a modeling category of three: Linear (ISO-1), Freundlich (ISO-2), and Langmuir (ISO-3). Each modeling option depends on the designer selection according to the fitted equation parameters between, C and, M. In terms of conservative results, ISO-1, and ISO-2 give less mass removal results than ISO-3 in case of sources with low concentrations. ISO-2, and ISO-3 give less mass removal results than ISO-1 in case of sources with high concentrations. / Ph. D.

Over the last decade, an increase in earthquake occurrence in Oklahoma and Kansas has been linked to oilfield wastewater injection disposal, particularly into the regionally underpressured Arbuckle Group. The Arbuckle is hydraulically connected to Precambrian basement through an extensive fracture system, which transmits pressure perturbations from wastewater injections to seismogenic depths. Previous studies have convincingly attributed induced seismicity to pore pressure diffusion and solid elastic stressing, both resulting from fluid waste injection. Recent work adds to the physical understanding of injection-induced seismicity by demonstrating that the density differential between injection fluids and formation brines may also drive fluid pressure into the seismogenic basement. In this thesis, variable density groundwater flow is modeled in a numerical simulation comprising parts of the Anadarko Basin, the Anadarko Shelf, the Cherokee Platform, and the Nemaha Fault Zone as well as injection data from 2006-2018. Results show buoyancy forces interacting with regional stratigraphic dip to force density-driven pressure transients into the deep Anadarko Basin, aligning with previously unexplained earthquakes in that region. / Master of Science / Increased earthquake activity in Oklahoma and Kansas over the last decade is linked waste disposal related to hydrofracking. Oil and gas produced in the fracking process is often mixed with large amounts of water that is too salty to be used for public or industrial purposes, thus this water is disposed of via injection into deep rock layers in the upper portion of the Earth's interior, or crust. This injection disturbs the crust to trigger earthquakes where none have been historically observed. Previous studies examining this phenomenon assume that the rock layers of the crust lie flat and level; simplify the nature of major faults, or cracks, in the crust; and do not consider differences in water chemistry between injected water and water that already occupies the crust. The study developed in this thesis considers the effect of these three factors with regard to how they influence the extent of the linkage between waste water injection and earthquakes in Oklahoma and Kansas.

The goal of this study was to determine the impact of land use activities on the subsurface flow regime in the Upper Roanoke River Watershed in Virginia to determine the impacts of land use change on the subsurface flow system, and to provide a tool for future management decisions. Land use activities can impact the groundwater system in two ways. The volume of water recharging the groundwater system can be reduced due to an increase in low permeable areas. It is assumed in this investigation that the input recharge values reflect the increase of low permeability zones that may occur due to land use activities. Increased water withdrawal associated with an increase in population can be another impact of land use change. This possible increase in water withdrawal is explicitly simulated in this investigation. MODFLOW, the USGS, three-dimensional, finite-difference, groundwater flow model was used to develop a regional conceptualization of the flow system. The fractured bedrock aquifer system consists of three sloping geohydrologic units: the Ordovician to Mississippian clastics, the Cambrian and Ordovician carbonates, and the Precambrian and Cambrian metamorphics and clastics. The 575 mi² study area was divided into cells with dimensions of 0.25 miles by 0.25 miles and containing four layers. The upper model layer was used to simulate the saturated unconsolidated deposits that lie on top of the fractured bedrock and serve primarily as a recharge reservoir. The second layer simulated shallow flow driven by recharge and the withdrawal of water by pumping wells. The bottom two layers were used to simulate deep regional flow within the system and account for possible vertical flow that may be occurring through deep fractures. Several simplifying assumptions were made during the conceptualization of groundwater flow in the study area: (1) Flow through fractures is approximately equivalent to flow through a porous medium; (2) Darcy's Law is applicable from a regional perspective; (3) Hydraulic properties are homogeneous and isotropic for an area that is represented by a model cell; and (4) Groundwater flow divides correspond to surface-water flow divides. Although these assumptions are probably valid for parts of the study area, the validity of each assumption is mostly unknown. Therefore, the model results are considered to be conceptual and should be interpreted carefully. The groundwater flow model was calibrated using UCODE, a USGS code for universal inverse modeling. Parameter estimation was conducted using UCODE for a total of 18 parameters, including hydraulic conductivities, river bottom conductance values, and recharge rates. The model was calibrated to observed hydraulic head information from 1969-1970. Due to the limited data availability, however, the calibrated values are at best, approximate. Nonetheless, several inferences can be made regarding flow in the province. The calibrated recharge values indicate that approximately 28% of the total precipitation recharges the aquifer system. This is consistent with previous estimates performed in the study area (Rutledge, Mesko, 1996). The Cambrian and Ordovician carbonates were found generally to have the highest hydraulic conductivity in each layer which reflects the notion that due to dissolution, this geohydrologic unit contains more fractures than the other two units. The calibrated values of hydraulic conductivity for the Cambrian and Ordovician carbonates ranged from 0.89m/d in layer 2 to 0.0011m/d in layer 4. The calibrated values of hydraulic conductivity for the Precambrian and Cambrian metamorphics and clastics ranged from 0.013m/d in layer 2 to 0.708E-3m/d in layer 4, and for the Ordovician to Mississippian clastics followed a similar trend in layers 2 and 3, with values of 0.390m/d in layer 2 and 0.242E-4m/d in layer 3. The streambed conductance values reflected both the variation in streambed thickness, which ranges from nonexistent in some areas to several feet thick in others, and streambed material, which ranges from sandy material with relatively high conductivity values to silty material with lower hydraulic conductivity values. The streambed conductance values range from 4.79 m²/d in the upland reaches to 234.13 m²/d in reaches closer to the outlet. Present pumping conditions were simulated with the groundwater flow model to establish a "baseline simulation" to which all future scenarios could be compared. Three future scenarios were developed based on the projected increase in population for Roanoke County through the year 2010. Each scenario represented a distinct settlement pattern within the watershed. Development scenario 1 simulated the impacts of the increased population if settled in the same areas as present development. Development scenario 2 simulated the impacts of the increased population if half settled in areas of present development and the other half in the western half of the watershed. Development scenario 3 simulated the impacts of the increased population if half of the population increase settled in areas of present development and the other half settled in the Tinker Creek sub-watershed. Development scenario 2 resulted in a drastic change in hydraulic head values, and the volume of water discharged from the streams was, on average, reduced by 56%, whereas, for both scenarios 1 and 2, these reductions were less than 1%. Results indicate that flow in the system is predominantly horizontal. There is no deep vertical flow from possible deep fractures. There may be shallow vertical flow occurring that is driven by recharge, however due to the resolution of the model, this flow is not simulated. In general, the simulation of horizontal flow follows the overall trend of the hydraulic gradient from west to east, which also follows the overall topographic trend. Therefore, upland regions in the province are recharging down-gradient areas. However, simulations indicate that the hydraulic head values in the eastern part of the study area are relatively insensitive to this horizontal recharge contribution from the west. The most sensitive areas in the basin to increased water withdrawal are the upland areas in the west side of the study area that are receiving no horizontal flow contribution from other places in the watershed. These areas are only being recharged by precipitation, and are the first to react to regional flow changes. Since the resolution of the model is such that local variations in the flow system are not simulated and the model represents regional trends, inferences can only be made about regional impacts. Therefore, if increased withdrawals are so great as to impact the regional system, the west- side of the study area will be affected before all other areas in the watershed. The study results include estimates of hydraulic properties, direction of regional flow, possible impacts from land use change, and a discussion of the results with respect to gaining a more complete understanding of the subsurface flow system. Perhaps this work will be the first step in learning more about the subsurface flow system of the Upper Roanoke River Watershed, and provide a useful tool to manage and properly plan future land use changes to minimize the impacts on the groundwater resources of the basin. / Master of Science

There are many methods for cleaning up contaminated soil and groundwater. Phytoremediation is an engineered method that utilizes plants and trees to remove or immobilize inorganic and organic contaminants. The plants and trees can contain contaminant plumes, uptake the contaminants, or aid in the degradation of the contaminants through several poorly understood mechanisms. Hybrid poplar trees were planted to contain a creosote contaminant plume at the study-site in Oneida, Tennessee. This research looks at how the trees will affect groundwater flow in the site. This is accomplished with the groundwater modeling program MODFLOW. The trees are simulated using the Evapotranspiration Package within MODFLOW, within the GMS modeling platform, to produce a two-dimensional unconfined aquifer viewpoint groundwater model. Site characterization, setup, and rationale are provided. The modeling methodology including calibration, sensitivity analysis, non-unique solution check, and verification are also provided. The modeling methodology included steady-state model calibration at the study-site to match observed field data; precursory steady-state and subsequent practice transient calibrations at the site; and incorporation of the simulation of evapotranspiration in the final transient model calibrations at the site. The results show that a phytoremediation system consisting of densely-planted hybrid poplar trees can indeed impact groundwater flow, although not to the extent that clearly would contain a creosote contaminant plume. Various input parameters including specific yield, transient recharge, starting heads, evapotranspiration rates, and evapotranspiration extinction depths impacted MODFLOW model sensitivity in transient calibrations. Varying the time steps in post-precipitation stress periods did not significantly impact the model output. The interception trench conductance played a minimal role in the calibration, but trench groundwater collection data was lacking, and the trench was frequently in need of maintenance. Further suggested data requirements include more frequently collected rainfall and piezometer data, as well as the installation of more piezometers outside the model domain contained in this study. Using the Evapotranspiration Package in MODFLOW provided more realistic and authentic results than using the Well Package (used in a previous study of the site by Panhorst in 2000) to simulate evapotranspiration. The Evapotranspiration Package in MODFLOW incorporates transpiration extinction depths that prevent transpiration when the water table drops below a certain depth. Further suggested program development includes incorporating an asymptotic function for transpiration rates and allowing the Evapotranspiration Package to import evapotranspiration rates, extinction depths, and elevations. It may be deduced from this impact of flow that the tree system will aid in containment of a contaminant plume, but at the trees current growth stage, and with the coal layer present at the site, the containment is limited. / Master of Science