Head-Dependent Flux Boundary Representation of Linear Hydrologic Features in Embedded Groundwater Flow Models

Linear hydrologic features such as rivers are important external stresses in groundwater flow systems that are implemented numerically as head-dependent flux boundaries. In MODFLOW, the USGS modular finite difference saturated groundwater flow model, information for cells designated as these types of boundaries are contained in input files, where the only spatial reference are the grid coordinates (Layer-Row-Column). Linear hydrologic spatial information is lost when embedded local models are created from regional models due to the coarser discretization of the latter. Two main problems arising from discretization are addressed: (a) the lack of information about the spatial configuration of linear hydrologic features in the regional model prevents their proper representation in an embedded local model, and (b) conductance values, the main calibration parameter for this type of stress, are scale-dependent. By utilizing the Dynamic Segmentation data model in Geographic Information Systems (GIS), the true alignment of the linear features can be used to assign parameters to the correct local cells (those that overlie the feature). An ArcView (ESRI) GIS program, AvHDRD (ArcView Head-Dependent River and Drain) has been developed that uses dynamically segmented information of the hydrologic features to construct MODFLOW RIVER and DRAIN input files. Each linear feature is a "route", a linear representation of the actual physical feature. In a route, a linear referencing system is established where the origin is located at the headwaters and the terminus at the point of discharge, with measures that increase downstream. This allows the specification of information about the route by using only the route's identifier (i.e., River Name) and the position along it (i.e., 2,500 meters from the origin). Tables are constructed containing route identifier, position, and physical parameters that include sediment hydraulic conductivity, sediment thickness, and bottom width. By mapping these properties using the routes as a reference and overlaying the local model grid, it is possible to extract the parameters necessary to compute conductances and to construct MODFLOW RIVER and DRAIN input files. These RIVER and DRAIN files represent the linear features at the scale of the local model, resulting in a more accurate representation of the local system at the desired reduced scale. A proposed enhancement to the Telescopic Mesh Refinement (TMR) process would enable the transfer of the calibrated regional model conductances to the local model cells utilizing the capabilities of dynamic segmentation, possibly reducing calibration efforts. The regional grid can be overlain on the linear hydrologic spatial data to generate length-normalized conductance values. Length-normalized conductance is obtained by dividing the conductance by the length of the segment of the feature within the regional cell. Dynamic segmentation allows length-normalized conductance values to be attributed to the hydrologic features at the midpoint of each reach. A local model grid is subsequently overlain on the dynamically segmented data to interpolate values of length-normalized conductance to the local cells. Finally, local cell conductance is obtained by multiplying its length-normalized conductance by the length of the river segment within it. In this manner, the calibrated regional model conductance values will be apportioned properly to the local model. In addition to being spatially accurate, the resulting MODFLOW RIVER and DRAIN input files inherit the calibrated conductances from the regional model. This approach can improve the consistency between the local and the regional flow model.