Curvilinear grid for ROMS

[cloudy4568]

Hello, everyone!

I am trying to make a curvilinear grid for ROMS, this grid is at equal distance at polar regions so it is curvilinear in longitude and latitude, just like the attached picture. Does anyone have the experience of making such a curvilinear grid for ROMS? Like how to calculate the rotate angels in the curvilinear grid?

Thank you a lot for any of the help!

[kate]

Yes, I can make a grid like that. However, you want your corners to be right angles. Also, a boundary across the pack ice will be hard to make well-behaved. We had a grid with a boundary at Bering Strait and we didn't like how that behaved either. I guess it depends on what your goals are.

How to do it? There are the usual GridBuilder, pygridgen, and other solutions.

[cloudy4568]

Thank you, Kate! I think maybe I should put aside the issue of angles and make the grid first.

[wilkin]

If you make a simple rectangular grid in a conformal map projection, then when you convert back to lon/lat the orthogonality is preserved and you can compute all the grid metrics for the projected lon/lat.

I made a simple test in Matlab using a Lambert Conformal projection for an arbitrary region in the Labrador Sea. I confess it got gnarly when I tested a grid that straddled +/-180 but I'm sure you can figure that out.

fig1.png (86.37 KiB) Viewed 8201 times

[cloudy4568]

Thank you, John! I have got the grid I expected with your method. Just make a rectangle grid in the Lambert Projection and then convert it back to lon/lat. Next I will try this curvilinear grid in ROMS. Thank you again for your help!