Up to Specific issues in real coupled models
Hi, I tried to send this to the forum discussion, but I need a user id and password. Even though I have registered for download, I did not seem to get a user id. I have been struggling with coupling a high-resolution (10m) non-hydrostatic version of the Bergen Ocean Model (BOM) glacier meltwater plume model with a coarser (250m) hydrostatic fjord-scale version of BOM through boundary coupling. I can run both models simultaneously, but uncoupled, under MPI under Oasis3-MCT. What I am trying to do is a gather on the full 3D fields, then extract a "boundary" Y-Z 2D slice, then send that data as a 1D array to a netCDF file, in the first instance, running just one of the models. I tried using oasis_create_couplcomm to use only a single processor (the master) out of the 12 active processes. Everything seems fine until after the oasis_put call, then, on the next model time step the model results are garbage. Somehow the original (local_comm) model processes get clobbered. In the User Manual, reference is made to sample code being available to illustrate using a subset of processors with oasis_create_couplcomm. Are there sample codes/toy models illustrating oasis_create_couplcomm, particularly in combination with a time-stepping model? Thanks in advance for any help or suggestions you can give. Regards Paul Budgell
Dear Paul, To post a comment on the forum, you do not have to login on the web site. You must go on the web page: https://verc.enes.org/oasis/faq-forum/oasis3-forum Then you have to choose a section and start a new conversation. Indeed, there is a toy, pes_coupling, using the oasis_create_couplcomm routine (see model2). To download it: svn checkout http://oasis3mct.cerfacs.fr/svn/trunk/examples/pes_coupling . You should also try to create a toy representing your real configuration to debug your real configuration and it would be easier for us to help you with such a toy. Best regards, Laure
Thanks Laure. I have downloaded and been able to run the toy pes_coupling code. I will use that as a basis to create my own toy model capturing what I am trying to do in my application. I will get back to you on how that turns out. Regards Paul
