Simulation of coastal meteorology and oceanography is severely hampered by the wide range of scales of forcing.  This multiscale system requires accurate simulation of the synoptic forcing (1000-100 km), the mesoscale forcing (100-10 km), and the extremely important local forcing (10-1 km) due to the land/water boundary.  In addition, the coastal meteorology is dominated by the boundary layer and hence surface physics and boundary layer interactions play an important role.

Typically, models have used nested rectilinear grids in simulation systems of the atmosphere to increasing resolution locally.  This approach, however, ignores two realities: (1) the physical features of the Earth are not oriented North-South / East-West and (2) high resolution is not required everywhere but rather where dictated by the physics.  In addition, nested grids are often used without the two-way passage of information between the scales but only with large scale to small scale transfer of information.

The aerospace community long ago recognized the above problems and developed unstructured triangular grid techniques to improve the generation of grids of complex surfaces and developed adaptive grid techniques, including dynamically adapting grids, to optimize the grid locally where appropriate.  In addition, since a single grid is used, there is always scale interaction in the simulation.

The adaptive unstructured grid technique has been transferred to the atmospheric community over the past decade with the development of the Operational Multiscale Environment model with Grid Adaptivity (OMEGA) (Bacon et al., 2000).  (A more recent application is the nascent development of the Multiscale Ocean Simulation System (MOSS) by the same team.)  OMEGA represents a new type of atmospheric simulation system that is inherently designed to treat non-uniform surface regions by efficiently identifying sub-regions of similar properties and adapting the grid to those features.  In this way, the OMEGA grid generator can automatically provide high resolution over regions of complex terrain, complex land-water boundaries, and disparate land use or other surface properties.  (In addition, OMEGA can dynamically adapt its grid to an evolving weather situation such as a tropical cyclone.

The unstructured grid technique is rather new to the atmospheric science community.  The flexibility of unstructured grids and their ability to adapt to transient physical phenomena are the features that give unstructured grid algorithms for partial differential equations their great power.  Grid adaptivity improves the fidelity of all finite difference or finite element numerical schemes, by increasing resolution in high gradient regions.  The improvement comes from the ability to adapt the grid structure to the flow, and from the local refinement of the grid in the vicinity of rapidly changing horizontal spatial structures in the atmosphere.

Using unstructured grids also eliminates the disadvantages of nested grid technique.  The main advantage of unstructured grids is the ease with which dynamic solution adaptation can be implemented.  There is no longer a need for involved user-expertise/interaction for creating topologies of complicated terrain features; the whole procedure can be fully automated, a feature that is not only highly desirable, but can also be required in operational settings.  Also, since the unstructured grid is a single mesh with a smooth and continuous transition from coarse to fine regions within the whole domain, the model is naturally two-way scale interactive without the interpolation error caused by the transfer of information from one nest to another.