Hi All,
I have a domain along the coastal Gulf of Alaska, including Cook Inlet (second highest tides in North America). I started with a 3-D simulation, not including tides. We ran into the usual sort of boundary troubles and were advised that tides would help flush things in or out.
I did some 2-D simulations with tides. They looked great once I switched to the rdrg2 type of bottom drag (z0 is not active in 2-D problems).
Onward to 3-D with tides. It is blowing up in 4-5 days and I suspect the bottom drag term. In all of these runs, my minimum starting depth is 10 meters. I deepened Cook Inlet and the sneaky devil is blowing up somewhere else instead. Realistically, I should be making it shallower - say 2 m deep with 10 m tides.
I talked to Zygmunt Kowalik, the local tidal expert. He likes the 2-D problem but makes two changes to his model: a limiter on h+zeta and an implicit bottom drag. He increases bottom drag at the open boundaries as a sort of sponge layer, so it has to be implicit. In the 2-D problem, he has a minimum total depth of about 10 cm. It seems to me that you want to limit the advective fluxes, so that you don't advect more water than you actually have to give. You would have to coordinate when it is flowing out of two sides of your box simultaneously. Is doing this in the 2-D part sufficient?
Back to the bottom drag, since it is my immediate problem. Does anyone have good notes on that aspect of the timestepping? The SCRUM manual has nothing to do with what is actually happening in that part of ROMS. Right now, the bottom stress gets added in pre_step3d.F unless BODYFORCE is defined, in which case it is added in rhs3d.F. Is there any reason it can't be moved to step3d_uv.F and made implicit? Has anyone tried this?
Thanks,
Kate