Some confusion about boundary condition

[zhangtianhao]

Hi everyone,

I have some confusion about my boundary condition. I use Cha for zeta boundary, Fla for 2D-momentum boundary and RADNUD for 3D-momentum boundary.(My region is in South China Sea)

1)I wonder that if I must input the ubar and vbar in the boundary ncfile or in the init ncfile?

2)And if I must input the zeta(not just give zero value)in my boundary ncfile?

3)Because the mask of my database(boundary data) and my grid is not consistent, the mask of some boundary grid is 0 but the database give T and S data(in database it locates in very shallow water). Does it matter if I give data in mask=0 grid? (Of course I have set the data in all of mask=1 grid)

Thank all of you!

[kate]

Yes, ROMS is expecting ubar, vbar in the initial and boundary files. It will ensure consistency between baroclinic and barotropic flows during initialization and after every timestep, but the boundary files ought to be generated in such a way that the u and ubar fields are consistent.

If you are doing an idealized problem, zeta=0 is a reasonable boundary condition. If you are extracting boundaries from some larger model, fetch its zeta along with its velocities and tracers so that everything is consistent.

My boundary extraction tools know about the land mask of the target grid and mask accordingly. I don't know how much it matters. What is more difficult is coming up with sensible values where the target grid is ocean and the source grid is not.

[zhangtianhao]

Yes, ROMS is expecting ubar, vbar in the initial and boundary files. It will ensure consistency between baroclinic and barotropic flows during initialization and after every timestep, but the boundary files ought to be generated in such a way that the u and ubar fields are consistent.

If you are doing an idealized problem, zeta=0 is a reasonable boundary condition. If you are extracting boundaries from some larger model, fetch its zeta along with its velocities and tracers so that everything is consistent.

My boundary extraction tools know about the land mask of the target grid and mask accordingly. I don't know how much it matters. What is more difficult is coming up with sensible values where the target grid is ocean and the source grid is not.

Thank you for your detail reply, Kate! It's really helpful.

I have another question: if my region is partly open, as shown in the picture below, which boundary should I set for closed side (north) and partly closed side(west and east)? I want to set Cha for zeta, Fla for 2d and RadNad for 3d. But does it just suitable for open boundary?

Thanks again!

[kate]

If any part of a boundary is open, you have to set the whole boundary to open. The land masking will take care of the rest. Your north can be closed. If you set it to open, no harm done, but you'll be asking ROMS to do extra work of reading fields (or computing analytic fields).

[zhangtianhao]

If any part of a boundary is open, you have to set the whole boundary to open. The land masking will take care of the rest. Your north can be closed. If you set it to open, no harm done, but you'll be asking ROMS to do extra work of reading fields (or computing analytic fields).

Thanks, Kate.