grid generation for idealized geometry

[iwaking]

Hi ROMS users,

I am trying to model an idealized estuary with the following bathymetry (exponentially converging zone followed by prismatic channel). Both the length, width and depth of the domain is analytically prescribed, as shown in the attached figure. I think curvilinear grid would be a good option for this geometry. I have tried octant but encountered a lot of problems in compilation. I am also aware of seagrid - but it seems that it is mainly used for realistic bathymetry. Can anyone help me with some recommendations for a good tool to analytically generate an idealized curvilinear grid for such a problem?

Thanks!
Kevin

[kate]

I just want to point out that your two western corners are not right angles and will lead to weirdness for an orthogonal grid.

[stevenmiguelfigueroa]

Hey Kevin,

I am working on the same problem. I tried Seagrid, but it was not easy for me. Let me know if you find a way.

I ended up deciding to make a grid by hand. Bernoulli solved for the curve orthogonal to an exponential. But this would only be orthogonal to one. I don't know if there are analytical solutions of curves which pass through the family of exponential curves. An alternative approach may be to discretize your estuary into prismatic section (orthogonals easy) and the converging section into piecewise lines which converge to points. For a given section of the piecewise lines, the orthogonal curve is also simple (a circle, like polar coordinates). It's elementary and pretty crude, though, honestly.

I settled for now on just keeping my idealized estuary on an equally spaced orthogonal grid (see below a test, including land mask for a straight channel). Hopefully simplicity will be a virtue.

I'm not sure why Kate has said that about western corners, but it's probably right. I'd be interested to know later how you are defining your estuary boundary conditions (estuary mouth and tidal river).

[jcwarner]

For the ROMS distribution we have the classic estuary test grid from our 2005 OM paper, that was rectangular.

In COAWST we have an estuary_test 2 "This application tests the ROMS for a prismatic shaped estuary that is 110 km long with the ocean end being 60 kms wide. The width of the estuary decreases exponentially from 60 km to 2 kms until 80 kms and stays uniform beyond 80 kms until the river end. The vertical layers follow a gaussian shape and consist of 40 vertical levels. "

for the grid have you tried
https://austides.com/downloads/
gridbuilder

[stevenmiguelfigueroa]

I tried Grid Builder. I find it easy to use. I think curvilinear grids for idealized coasts can be generated in it (see attached test). The grid property format should be Free and coordinates Cartesian. With Free, under grid edit, the corners and control points can be customized. I found that the Grid Builder manual assumes the reader is familiar with how control points/SeaGrid works. So, it took a bit to learn how to use control points.

The idealized coastline can be imported as a .mat file. Then the grid fit to it. The software is tailored toward regional scale oceanography, so I'm hoping I can find my small scale idealized estuary by zooming. Also, I'm thinking it's better to generate the grid in Grid Builder and then generate the bathymetry in matlab, instead of importing to Grid Builder the bathymetry. In the former case, the bathymetry is expected to fit the idealized coastline better.

I tried to edit the grid generation section of WikiROMS (last updated 2008) to add Grid Builder. I didn't have permission. If someone who does adds it, it could help others in the future.