VanderKwaak, Joel E.2006-07-282006-07-2819991999http://hdl.handle.net/10012/412Stream water quality is affected by processes occurring in and below the stream channel, by baseflow contributions from groundwater, and by the short-duration, high-volume contributions from precipitation events. Defining the source and pathway of stream water inputs is a prerequisite to understanding the impact of contaminants originating in rainfall, from industry and agriculture, and from urban runoff. Numerical models provide a useful tool in evaluating possible flowpaths and the timing and magnitude of stream inputs from various sources. An integrated numerical model is developed and evaluated in this work. This numerical model considers the flow of water and transport of multiple solutes on the two-dimensional land surface and three-dimensional, dual continua subsurface. Linkage is through first-order, physically based flux relationships or through continuity assumptions. Coupling of flow and transport is achieved by assembling and solving one system of discrete algebraic equations so that water and solute fluxes between continua are determined as part of the solution. Specified boundary conditions can be spatially- and temporally variable, or in the case of state-dependent flux boundaries, can be specified as nonlinear functions of the local flow or transport solution. The numerical model is modular in form, is tailored towards irregular geological, surficial and areal geometries, and utilizes robust and efficient discretization and solution techniques. A new prism-based discretization is introduced and shown to be consistent with two and three-dimensional finite elements, while utilizing less memory and computational effort and generating significantly fewer negative influence coefficients. An adaptive temporal-weighting scheme is presented which partitions the flow and transport equations into active and inactive zones. Solutions for inactive equations are calculated using explicit temporal weighting and are excluded from the flow or transport Jacobians, reducing assembly and solution time. The numerical model is evaluated by simulating two-dimensional laboratory and three-dimensional field experiments of coupled surface-subsurface flow and transport. Observed surface discharge volumes and timings are simulated with reasonable accuracy using published or measured parameter values and minimal calibration. The observed dynamic response is a nonlinear function of multiple parameters, affected by subsurface permeability, surface roughness, topography, and initial conditions. Excess rainfall and groundwater seepage flow overland, generating surface ponding and streamflow along topographic lows. The ponded surface water forms an internal, transient constraint on the porous medium pressure head at the land surface. Solution based on traditional seepage face algorithms do not reflect the effects of this ponded surface water on the distribution of subsurface head gradients. Groundwater seepage is therefore overestimated during rainfall events while infiltration of ponded surface water is neglected at all times. Simulations of the transport of a conservative tracer introduced with rainfall indicate that processes affecting solute concentrations in the surface water are restricted to a relatively thin region adjacent to the land surface. Concentrations in surface water are very sensitive to which equations the rainfall boundary condition (i.e. specified flux) is applied. For rainfall applied to the surface equations, mixing between the surface and subsurface continua is more heavily influenced by the magnitude of diffusive/dispersive exchange coefficient. The magnitude of advective exchange is controlled by hydrodynamics within the porous medium and not by the movement of excess rainfall from the porous medium to the surface continuum. While having little affect on the flow solution, subtleties in rainfall boundary condition assignment in the discrete equations impact predictions of tracer concentrations in discharge water and, therefore, also affect interpretations of water origin. Application of the coupled surface-subsurface model to the transport of conservative tracer in the field-scale experiment re-enforces the conclusion that mixing processes occurring at the land surface interface dominate tracer concentrations in stream discharge. Simulated hydrograph separations (i.e. relative concentration multiplied by stream discharge) replicate separations based on measured values with reasonable accuracy only if rainfall is applied to the surface equations and both advective (i.e. infiltration/seepage) and diffusive exchange processes are considered. Simulated flux-weighted concentrations, however, exceed measured concentrations during hydrograph rise. Simulation of field-scale transport processes is considerably more complicated than at the laboratory-scale, where topography is better defined and extremely fine spatial discretization can be utilized. Successful simulation of coupled surface-subsurface transport depends on the accurate representation of the spatial and temporal variability of water exchange processes (i.e. advection) and diffusive-type processes associated with concentration differences between continua. The field-scale simulations clarify the role of the capillary fringe on streamflow generation in the relatively homogeneous sand underlying CFB Borden. The coupled surface-subsurface flow model is able to reproduce the observed rapid water table response and resulting overland and stream flow. Observed surface discharge volumes and timing are simulated with reasonable accuracy using published or measured parameter values and minimal calibration. The simulated response of the capillary fringe to rainfall is consistent with both theory and observations but suggest that increased subsurface head gradients do not cause significant groundwater seepage. Rather, infiltration rates along the stream axis are reduced, with runoff formed largely by excess rainfall over a dynamic contributing area. The corresponding transport simulations suggest that, despite the rapid, large-scale response of the capillary fringe, rainfall tracer dilution occurs largely by diffusive processes as water flows over the land surface to the stream, over relatively short flow paths, and subsequently down the stream channel. Tracer originating above the initial water table enters the surface water by similar processes, augmenting the small volumes of seepage (advective transport) caused by increased subsurface hydraulic gradients. The simulations suggest that hydrograph separation theory is fundamentally flawed if diffusive modification of tracer concentrations in surface water is prevalent in nature.application/pdf13592771 bytesapplication/pdfenCopyright: 1999, VanderKwaak, Joel E.. All rights reserved.Harvested from Collections CanadaDoctoral Thesis