Chang (1998) states that the actual sediment rate is obtained by considering sediment
material of all size fractions already in the flow as well as the exchange of sediment load with
the bed using the method by Borah et al. (1982). If the stream carries a load in excess of its
capacity, it will deposit the excess material on the bed. In the case of erosion, any size
fraction available for entrainment at the bed surface will be removed by the flow and added
to the sediment already in transport. During sediment removal, the exchange between the
flow and the bed is assumed to take place in the active layer at the surface. Thickness of
the active layer is based upon the relation defined by Borah, et al. This thickness is a
function of the material size and composition, but also reflects the flow condition. During
degradation, several of these layers may be scoured away, resulting in the coarsening of the
bed material and formation of an armor coat. However, new active layers may be deposited
on the bed in the process of aggradation. Materials eroded from the channel banks,
excluding that portion in the wash load size range, are included in the accounting. Bed
armoring develops if bed shear stress is too low to transport any available size.
Two-Dimensional Sediment Transport Models are also available. Both finite-element and
finite-difference numerical schemes have been applied. Examples of finite-element models
are the SED2D (Letter et al. 1998) and Flo2DH (Froehlich 1996) models. SED2D is an
uncoupled sediment transport model that generally uses RMA-2V (USACE 1997) hydraulic
simulation results to compute sediment transport, scour and fill, then updates the model
geometry and cycles back to RMA-2V to update the flow field. Other sources of hydraulic
computations can be used as input to SED2D. Also under development are Flo2DH Version
3 and Flo1D (Arneson 2001). FESWMS Version 3 will incorporate sediment transport
computations. These models are best suited for short-term sediment transport simulations
although SED2D has features that allow for long-term simulations.
5.6.3 Data Needs
Common to all alluvial river-flow models are requirements for the following input information: (1)
accurate initial conditions, including a cross-sectional profile and bed-material size distribution
at each computational cross section; (2) accurate boundary conditions such as water and
sediment inflows along the boundaries, quantitative expressions of bed-load and
suspended-load discharges, size distributions of boundary-sediment input, and stage
hydrographs at the upstream and downstream boundaries; and (3) bed-roughness
characteristics at each computational point. It is clear that a computer simulation would be
meaningless without the first and second requirements, and the lack of the third requirement
would yield an erroneous estimation of flow characteristics, resulting in erroneous feedback of
flow information to the riverbed. Included in the first set of data are depths of alluvium and
controls such as structures and bedrock outcrops.
The exclusion of even one of these three requirements may lead to serious errors in computer
simulations. However, one can hardly be provided with a complete set of input data in any
prototype numerical application. Therefore, a great number of assumptions often have to be
made to fill the gap in the input data. Even if adequate data are provided for a study river,
there still remains a need to calibrate and verify the model by means of field data. In most
natural rivers, only extremely limited field data are available for high flood stages at which major
riverbed changes occur, and, consequently, adequate calibration or verification of the models
normally cannot be obtained. In this sense, the capability of the alluvial river-flow models can
best be assessed according to how accurately they can predict riverbed changes with limited
sources of input data. A numerical modeler should be aware of which input information is most
important to the final result of predicting riverbed changes. The success of the numeral model
depends on the capabilities of the model, the quality of the input data and the abilities of the
modeler.
5.47
Previous Page
