Deterministic Modeling of Bromide Tracer Transport During the Tracer/Time-Lapse Radar Imaging Test at the Boise Hydrogeophysical Research Site in August, 2001

Contaminant transport and remediation strategies are based on knowledge of site specific parameters and states (e.g., hydraulic conductivity, porosity, hydraulic gradient, plume density). Often little is known in detail about these parameters or states and their spatial distributions. This uncertainty can introduce inaccuracies and errors in transport predictions. This thesis is a deterministic approach to reducing this uncertainty for transport of bromide, a conservative tracer, in a heterogeneous unconfined fluvial aquifer in a natural environment through a series of numerical simulations of the tracer test conducted at the Boise Hydrogeophysical Research Site (BHRS) in 2001. In particular, the investigation progresses from initial one-dimensional analytical modeling to three-dimensional numerical modeling that examines the influences of boundary conditions, heterogeneity (including consideration of hydraulic conductivity, porosity, and dispersivity), plume density effects, and evapotranspiration on bromide breakthrough behavior that was measured at multiple levels in a sampling well ~4 m down-gradient from the tracer injection well. Analytical modeling provided initial estimates of average hydraulic conductivity and longitudinal dispersivity. Two approaches were used to numerically model the flow during the tracer test: (1) uniform gradient (simple boundaries) approach; and (2) complex boundaries approach. The uniform gradient approach included boundaries that defined a uniform regional hydraulic gradient through the BHRS. The complex boundaries approach included the river as a groundwater divide, and general head and constant head boundaries calculated from a larger flow model of the gravel bar upon which the BHRS lies. Irrespective of which boundary flow approach was used, the same transport approach was used. The software codes used for this modeling are MODFLOW 2000 for groundwater flow, SEAWAT 2000 for including density effects in the flow model, and MT3DMS for solute transport. The goodness of fit for a given set of model conditions was evaluated by comparing visual matches of model and observed head changes in well B3 (injection well) and in the C wells (a ring of six wells ~5 m outside the injection and pumping wells), and matches of model and observed breakthrough summary statistics such as Time to Peak and Peak Concentration. Findings from this modeling progression include: (1) improvements to fitting the breakthrough data resulted from including boundary conditions such as adding leakage through the bed of the Boise River which is given specified heads at a groundwater divide boundary, and also using general head boundaries on other sides of the gravel bar; (2) although the BHRS has a noticeable evapotranspiration signal (~2.5 cm daily fluctuation in the water table), simulations with and without evapotranspiration show that evapotranspiration is not important for transport during this tracer test; (3) plume density effects were expected to be significant for the conditions of this tracer test based on calculations of Barth et al.\u27s (2001) dimensionless α2 parameter for the BHRS and for the tracer test, and this was confirmed by comparison of simulations with and without variable density flow; (4) using homogeneous hydrogeologic properties throughout the BHRS is not sufficient to match observed tracer behavior measured during the tracer test; (5) heterogeneity (e.g. layers; contact geometry; distinct zonation within layering; variable distributions of hydraulic conductivity, porosity, or hydraulic conductivity and porosity) significantly affects simulated transport behavior and as more known heterogeneity is included, simulated breakthrough behavior matches observed behavior more closely; (6) by defining well bores in the simulations, more water flux from a sand channel in the upper part of the system and less water flux from the direction of the tracer plume is observed by later tracer breakthrough in the sampling well; (7) as expected, larger dispersivity results in more spreading of the tracer plume; (8) a uniform increase in effective porosity will cause later breakthrough; and (9) increasing vertical hydraulic conductivity anisotropy does not improve matches between simulated and observed breakthrough. By accounting for more complexity (heterogeneity) in the hydrogeologic properties of the BHRS, improvements are made to matching observed tracer breakthrough with simulated breakthrough. This thesis is a first pass at hydrologic modeling upon which more follow-up or further development will be conducted