updates to BDY module

[davestorkey]

There are two main changes:

1. Implementation of open boundary conditions for LIM2 ice model. Flow Relaxation conditions on leads fraction, ice thickness and snow thickness fields.

2. Bug fixes and updates for use of BDY in Shelf configurations.

[davestorkey]

Main code changes

1. Implementation of FRS boundary conditions for ice fields (ice concentration, ice thickness, snow thickness). Currently only implemented for LIM2. Main code changes:
   1. Extend bdy_oce.F90, bdyini.F90, bdydta.F90 to deal with the ice fields if key_lim2 is set.
   2. An additional file bdyice.F90, to apply the ice boundary conditions. The bdy_ice routine is called from sbc after the call to the LIM2 model.
2. Changes required for Shelf implementation:
   1. We want to be able to have separate files for the baroclinic (FRS) boundary conditions and non-tidal barotropic (Flather) boundary conditions. So set jpbgrd to 6 in bdy_par.F90 and make associated changes to bdyini.F90, bdydta.F90, bdydyn.F90 and bdytides.F90.
   2. Remove the application of Flather boundary conditions on the baroclinic timestep in dynnxt.F90. This was found to destabilise Shelf configurations.
3. Bug fix for zeroing of mass divergence behind open boundaries. The ssha and wn fields are now masked with the bdytmask in wzvssh (similar to what is done for OBC), and the application of bdytmask to hdivn in divcur.F90 is removed. This change removes spurious SSH values along the boundaries in deep ocean configurations (linear SSH + FRS conditions) and prevents errors in the tracer advection at the boundaries in Shelf configurations (nonlinear SSH + FRS + Flather conditions).
4. Add BDY boundary conditions for solver in solmat.F90.
5. Bug correction for time interpolation of boundary data fields in bdydta.F90.

[davestorkey]

Testing

1. The ice boundary conditions have been tested in a one-year hindcast of the FOAM North Atlantic configuration (using NEMO v3.2) taking boundary conditions from the global ORCA025 configuration. There is an open boundary with ice along the north of the model domain (cutting across Baffin Bay and the East Greenland Current), and the ice simulation is found to be satisfactory here.
2. The changes for the Shelf configurations have been tested in hindcasts of the Atlantic Margin model (using NEMO v3.2) with open boundary conditions consisting of tidal harmonics forcing and FRS conditions on T,S,u,v fields taken from the FOAM North Atlantic configuration.
3. The bug fix for the zeroing of the mass divergence behind the open boundary was tested in the Atlantic Margin model - it was found to be necessary for this model to run stably. It has also been tested in the FOAM North Atlantic model and found to remove spurious values of the sea-surface height along the edge of the model domain.
4. The new branch relative to 3.2.1 has been compiled with and without key_bdy. A null test with ORCA2 with key_bdy omitted showed no change in results or resources used between a run with the branch included and a run without the branch included.

[davestorkey]

Hi Rachid,

I've updated the ticket with details of the testing done. Could you review this please?

Thanks,

Dave

[rblod]

Review is located here ​http://forge.ipsl.jussieu.fr/nemo/wiki/718/Review

[nicolasmartin]

Milestone 2010 Stream 0: Coastal ocean ready deleted

[nemo]

Remove 'Development' type