Model Limitations

Information on model limitations is presented in Barton and others (2005) and summarized here. The braided-straight and meander-straight reach multidimensional models are designed to simulate steady streamflows less than about 2,130 m³/s. The models can simulate higher streamflows, but with increasing uncertainty in the model simulation results. The accuracy of simulations of the velocity flow field simulated by these models is constrained largely by (1) accuracy and amount of channel geometry data; (2) errors that could arise because channel geometry was measured during streamflows that varied between measurements; (3) uncertainty in the relation between streamflow and bedforms; (4) simplification of the roughness of the riverbed including bars, vegetation, and sediment on the riverbed, and of the river banks; and (5) potential errors incurred by applying a steady-state model to unsteady streamflow conditions. Additionally, the two-dimensional model cannot explicitly account for variations in velocity magnitude and direction in the vertical dimension. Therefore, streamflow dynamics adjacent to anthropogenic or natural vertical obstructions in the flow field would be better described using a fully 3-dimensional model. Similarly, the two-dimensional model treatment of turbulence is simple, and does not thoroughly describe flow patterns for areas in the river where undulating and complex bathymetry exist. Because of the complex computations and the requirement for more data collection, the beneficial return of an unsteady 3-dimensional, turbulence-resolving model probably is limited.

Another potential source of error results from the upstream boundary conditions input by the model user. Velocity magnitude and direction at the upstream boundary of the model grid have a profound effect on the simulated streamflow patterns in the upstream part of the modeling reach. Velocity conditions have been specified at the upstream boundary of the meander-straight reach model based on field measurements taken at the boundary during streamflows of 776 and 1,260 m³/s. However, velocity data were not available and thus velocity conditions were not specified for the upstream boundary of the braided-straight reach model. Velocity boundary conditions likely will vary for streamflow conditions that are greater or less than streamflow during field measurements.

Periodic updates to the braided-straight reach model are required to continue to accurately simulate hydraulic conditions in the braided reach into the future. The basis for this requirement is that bathymetry input into the braided-straight reach model will, over time, misrepresent that reach because of shifting channels, scouring, and deposition. Therefore, the reach must be mapped periodically, the resulting bathymetry data input into the model, and the model recalibrated. Additionally, calibration of this model uses stream discharge measured at the Tribal hatchery gaging station that is computed using an index-velocity ratio. In the future, an adjustment may be applied to the computation that causes a shift in discharge over certain ranges of streamflow. This adjustment would take into account that the river has a mobile bed, and that changes in bed elevation across the channel at this gaging station may occur between low and high streamflow. As additional bed-elevation measurements are made at the gaging station and are evaluated for effect on flow, that information may be included in the discharge computations, and cause a shift in discharge over certain ranges of streamflow. Therefore, the model will need to be recalibrated to reflect this shift in discharge.