Rigid Lid

[kthyng]

Hi everyone,

Is there an easy way to do a rigid lid approximation in ROMS? I want to try out the code in some test cases that require it. I searched on the wiki and forum but didn't find any references to it.

Thanks for any help!

[arango]

No, this is not possible in ROMS. The rigid lid approximation was used in the 70's to filter out gravity wave dynamics in ocean models (see GFDL original model formulation by Brian and Cox). This required the solution of a vertically integrated equation in terms of a streamfunction using a Poisson solver at every time-step. The solution of such equation was very expensive in certain applications. The complexity increased in regional applications with open boundary condition. Since then, the GFDL model has evolved to include a free-surface allowing fast-moving gravity wave physics.

ROMS has an split explicit scheme to resolve gravity waves using smaller time-steps when solving the vertically-integrated equations. That is, ROMS kernel is designed to include gravity wave phenomena and cannot be removed. You need to rewrite completely the 2D engine and coupling. This affects ROMS infrastructure completely

[kate]

The Rutgers gang had a rigid lid model before ROMS and moved away from it for several reasons. One is the communications in the solver when running on parallel computers - computers are getting more and more processors, rather than faster processors. It's hard to imagine, but when I started computing on the systems here at ARSC, it was on one of the four processors on a Cray.

[leilanepassos]

Hi,

before I read this topic, I thought that ignoring the barotropic part of the mode splitting (NDTFAST=0). I would be applying the rigid lid approximation, but now I know that's not what I did. What are the physical implications of ignoring the barotropic part of the mode splitting?

Thanks,

Leilane

[kate]

NDTFAST=0? So you never let zeta/ubar/vbar evolve? I'm not sure what ROMS would do - it may be solving the full 3-D problem with the contraint that zeta=ubar=vbar=0 all the time. Is that what you got? This is not at all the same thing as a rigid lid, since a rigid lid can provide pressure gradients at the surface and support depth-integrated flows.