Introduction - Computational River Dynamics - page 6

(7) Reliability of numerical solutions and applicability of computational models in
different situations.
To insure the quality of the simulation results, a computational model of flow and
sediment transport should be verified and validated before application in solving real-
life problems. Model verification and validation usually follow three steps (Wang and
Wu, 2005):
(1) Verification by Analytic Solutions . The agreement between analytic and numeri-
cal solutions certifies the correctness of the mathematical formulation, numerical
methods, and computer programming. It can also determine errors of numerical
solution quantitatively.
(2) Validation by Laboratory Experiments . Because laboratory experiments con-
ducted in controlled environments can eliminate many unnecessary complications,
the numerical model should be able to reproduce the same physical phenomena
measured in laboratories.
(3) Validation by Field Measurements . One portion of the field data should be used
to calibrate the physical parameters in the model, and the remaining data can
be used to determine whether the computational model can simulate the real-life
problem. Researchers must realize that the numerical results may only approx-
imately agree with the measured data, because the computational model only
represents a simplified version of the physical processes in natural rivers. How-
ever, the realistic trend of spatial and temporal variations should be predicted
correctly.
The application of a computational model to the solution of a real-life problem
involves the following five major tasks:
(1) Data Preparation . Data should be collected and analyzed to understand the phys-
ical processes of study, determine initial and boundary conditions, estimate model
parameters, and calibrate the model. The required data should include, but are
not limited to, geomorphic, hydrological, hydraulic, and sediment information,
largely depending on the model used and the study case. They can be obtained via
in-situ field survey and from historical records.
(2) Estimation of Model Parameters . Model parameters can be classified as numerical
and physical. Numerical parameters, such as time step, grid spacing, number of
size classes, and relaxation coefficient, result from numerical discretization and
solution methods. They should be determined by considering the accuracy the
study problem requires and the stability of the numerical schemes used. Phys-
ical parameters can be subdivided into two groups. One group represents the
physical properties of water and sediment, such as water density, viscosity, sedi-
ment density, particle size, particle shape factor, and bed-material porosity. These
physical properties can be measured. The other group results from the concep-
tualization of physical processes and represents the characteristics of flow and
sediment transport, including channel roughness coefficient, sediment transport
capacity, sediment adaptation length, and mixing layer thickness. These physical