Updated 2016/07/31: This post is now outdated. SWAN has been updated to improve its treatment of the spectral propagation velocities, so these limiters are not needed. Please see this post.
Updated 2012/04/12: This is an old page. It persists on this site for posterity, but the information presented below is no longer up-to-date. When you are done here, then please click forward to this page, which describes how to control refraction errors with limiters on the spectral propagation velocities.
In a previous page on wave refraction, it was shown that mesh resolution plays an important role in how SWAN handles this physical process. If the mesh is resolved coarsely, then SWAN can refract too much energy, resulting in spikes in the wave solution. In our hurricane applications, we have observed spikes in the significant wave heights of 75m or larger, focused at only a few vertices, because the mesh in those regions does not resolve properly a shallow feature.
To address this problem, we added wave refraction as an attribute to the ADCIRC fort.13 file, so that it can be varied spatially. The user can enable refraction in regions with the necessary level of mesh resolution, and disable refraction in regions that are resolved coarsely. We enabled wave refraction along the northern Gulf coastline and within southern Louisiana, and we no longer saw the spikes in the significant wave heights in the regions offshore.
Bathymetry and topography (m) on the SC12 unstructured mesh.
Peak Periods TPS in South Carolina
However, we have learned recently that wave refraction on coarse meshes can be problematic in other ways besides creating spikes in the significant wave heights. For example, in very shallow regions, the dissipation of bottom friction and breaking would limit the significant wave heights to reasonable values, even if wave refraction was a problem. There would not be any spikes in the significant wave heights. But the focusing of wave energy would be apparent in other ways that are more subtle, such as the existence of peak periods with the maximum possible values.
On this page, we examine an instance of this problem with wave refraction on coase meshes. As a specific example, we consider a mesh that was developed for a flood inundation study along the coastline of South Carolina. It should be noted that, although this example mesh does exhibit these problems with wave refraction, the problems are also evident on other meshes and other applications. For example, we have seen similarly problematic peak periods in our hurricane applications in southern Louisiana. As we have noted previously, any wave model would face this problem of how to handle refraction on unstructured meshes.
Continue reading →