Dear Fellows,
We found a bug in the computation of the V-component of the tides. It is pretty straight foward to see the bug when the equation is written in continuum form but very hard to see it in the fortran code:
omega = 2 * pi * time / period
angle = UVangle - angler (angler is curvilinear angle)
Cphase = omega -UVphase
Elevation: Etide = Etide + amplitude * COS (omega - Ephase)
currents: Utide = Utide + Tmajor * COS(Cangle)*COS(Cphase) - Tminor * SIN(Cangle)*SIN(Cphase)
Vtide = Vtide + Tmajor * SIN(Cangle)*COS(Cphase) + Tminor * COS(Cangle)*SIN(Cphase)
Many thanks to Lyon Lanerolle for finding this bug.
Also seems that a couple of months back, I sent the wrong version of the "rho_eos.F". I sent a newer non-distributed version of the code. Many thanks to Dave Chapman for pointing this out.
I will send the corrected routines: clm_tides.F and rho_eos.F for version ROMS 1.8 via e-mail.
I am still working in the new version of ROMS. I am planning to distribute a new version soon.
I am also working on TOMS. This is a very new model completed based on ROMS but in F90/F95, it has multiple levels of nesting and/or multiple connected domains, both shared-memory (OpenMP) and distributed-memory (MPI) paradigms are supported, a lot of new modules like sediment, generic length scale parameterization (k-kkl, k-epsilon, k-omega, k-tau, and generic m,n). In addition, it will have a tangent linear and adjoint codes.
At that point both ROMS and TOMS will merge into a single code.
So be patient with me,
Best, H