Geoscience Reference
In-Depth Information
8.2 COUPLING OF 1-D, 2-D, AND 3-D MODELS
8.2.1 General considerations
As described in Chapters 6 and 7, the flow and sediment transport in rivers with com-
plex geometries and hydraulic structures should be simulated using 2-D or 3-D models
rather than 1-D models. However, it may not be feasible to use 2-D and 3-D models in
the simulation of the fluvial processes in a long river reach during a long period, because
they require much more computation time than 1-D models. Therefore, it is cost-
effective to couple 1-D, 2-D, and/or 3-D models (McAnally
et al
., 1986; Wu and Li,
1992; Vieira, 1995; Zhang, 1999). The basic idea is to divide the entire study domain
into subdomains (reaches), and apply a 1-D model in less important subdomains with
simple geometries and a 2-D or 3-D model in more important subdomains with com-
plex geometries. For convenience, the subdomains (reaches) handled by 1-D, 2-D, and
3-D models are herein called 1-D, 2-D, and 3-D subdomains (reaches), respectively.
The concept of coupling 1-D, 2-D, and 3-D models in a generic river system is
illustrated in Fig. 8.9. The upstream portion with dams can be simulated using a
1-D model, thus simplifying the boundary conditions. The broad floodplains can be
calculated adequately using a depth-averaged 2-D model. The portion with bridge
crossing should be simulated using a 3-D model. In the estuary, the stratified flow and
salinity intrusion can be simulated using a width-averaged 2-D or 3-D model. In the
estuary entrance, the interaction between river flow and tidal current is very complex
and should be calculated using a 3-D model. Such a coupled modeling can provide a
feasible solution for the entire river system.
Figure 8.9
Concept of coupling 1-D, 2-D, and 3-D models.
Two approaches have been employed in the literature for combining 1-D, 2-D,
and 3-D models. One approach is to use a simpler (1-D or 2-D) model in the entire
solution domain and a more complex (2-D or 3-D) model in confined subdomains.
The simpler model provides boundary conditions for the more complex model, but
they are not coupled. This approach is called the hybrid modeling (McAnally
et al
.,
1986). The other approach is to fully couple 1-D, 2-D, and/or 3-D models (Wu and
Li, 1992; Zhang, 1999) by simultenously solving all component models. Because the
fully coupling approach is more general, it is the main concern here. The discretiza-
tion and solution schemes in each subdomain usually are the same as those used in
