We found a bug in ROMS horizontal viscosity for momentum. This only affect applications with explicit horizontal viscosity. Actually, any value of viscosity chosen does not seem to do anything and the simulation remains inviscid.
This bug has been in ROMS for very long time. We did not noticed it before because by default ROMS uses 3rd-order upstream bias horizontal advection which includes implicit hyper-viscosity in its leading truncation term. In the past, all the horizontal dissipation was mainly specified for the tracer equations.
The problem was that the horizontal stress tensor terms were not included in the residual vertical-integrated terms rufrc and rvfrc used in the split-explicit coupling between baroclinic and barotropic momentum equations. Therefore, the barotropic equations remained horizontally inviscid and the net viscosity was canceled during the barotropic/baroclinic coupling. We corrected this bug and are now getting the correct damping of barotropic Rossby waves and viscous boundary layers.
The next release of ROMS will include the corrections for this bug. In the mean time, get the following tar file:
and substitute the following files with the new ones:
Code: Select all
rhs3d.F uv3dmix2_s.F uv3dmix2_geo.F uv3dmix4_s.F uv3dmix4_geo.F MakeDepend
Many thanks to Genta Mizuta (WHOI) for pointing out this bug.
Hernan G. Arango