For a particular application that I am running, I need to use the volume conservation CPP options in ROMS (I use EAST_VOLCONS, WEST_VOLCONS and SOUTH_VOLCONS).
I test to see whether the volume is indeed being conserved by integrating the (normal) velocities along each of the open boundary faces of the model domain and then summing the contributions up; I do this inside the ROMS itself (in ROMS/Nonlinear/diag.F).
I find that in a SERIAL model configuration, I get near perfect volume conservation (the residual volume ~ 1.0 x 10^-17) but with MPI the volume conservation is destroyed and I get a large net residual volume. The ROMS however does not blow-up but continues to run in a stable fashion.
I have looked at the obc_volcons.F routine (in ROMS/Nonlinear/) but I could not figure out how to fix this MPI-related bug . Could someone please help me out?
Thanks a lot,
Lyon.