Thin Wall Between Grid Cells

[soilinkin]

Hi, all. I have a question hoping someone can help.

Is it possible to put a thin wall between 2 grid cells, so there is no exchange at all between the cells? I have a situation that there is a thin wall in my domain, to separate neighboring cells, but I can't find any information related.

I imagine mask_u and mask_v may be used to separate neighboring wet cells, while mask_rho of the cells still indicate wet cells. But I am not exactly sure whether that works or how to set up things. Do I also need to worry about mask_psi?

[jcwarner]

Typically to provide a permanent wall, then use the mask _u _v _rho variables.
ROMS is solved on a staggered grid, so when you set the mask_rho, then edit_mask (or similar) can determine the appropriate settings for mask_u and mask_v and mask_psi.
However, it seems you want a special case, where 2 adjacent rho points are land mask =1, and you want to force the mask_u (or mask_v) to be 0 inbetween those 2 rho points.
You can do that by directly editing the grid netcdf file, but i am not completely sure what will happen with that. It is worth a try.
IF there is complications of wet/dry, that might just always work out, as the wet dry is typically multiplied by the land/sea mask.

-j

[soilinkin]

Typically to provide a permanent wall, then use the mask _u _v _rho variables.
ROMS is solved on a staggered grid, so when you set the mask_rho, then edit_mask (or similar) can determine the appropriate settings for mask_u and mask_v and mask_psi.
However, it seems you want a special case, where 2 adjacent rho points are land mask =1, and you want to force the mask_u (or mask_v) to be 0 inbetween those 2 rho points.
You can do that by directly editing the grid netcdf file, but i am not completely sure what will happen with that. It is worth a try.
IF there is complications of wet/dry, that might just always work out, as the wet dry is typically multiplied by the land/sea mask.

-j

Thank you, J.
Yes, you are right about my special case. I understand that editing mask_u/mask_v is relatively easy. I just can't find information about how mask_u/mask_v works, so I'm not sure whether this can impose a barrier between 2 cells. At the same time, is mask_psi also need to be edited along with mask_u/mask_v to make it work?

[jcwarner]

just try the adjustment to just mask _u _v for now and see if that works. it may look strange, as there could be dzeta/dx gradients across the rho cells, but no flow. but give it a try.
-j