Closed mzapponi closed 10 months ago
Thank you for adding the new method. I'm not sure I want to to be the default for all experiments. On the one hand gsspos is crashing every once in a while. But then again people are already using gsspos for scientific research. I think we would have to go with a model version based coupling method to ensure reproducibility. I could make a version v3.1.3 where the only change is the default to global instead of gsspos.
For the v3.2 evaluation experiments I used glbpos instead of gsspos. While global is the one that is least prone to crashes, it is also the least physical (can e.g. result in negative precipitation or sw rad).
Hi @mzapponi , @JanStreffing , I only added the new coupling method but didn't change the default coupling method for heat_ico or any other coupling fields. See also PR #1094
Hi @mzapponi , your new OASIS coupling method is now merged into release of ESM-Tools.
When dealing with an oasis issue that was causing the crash of the simulations few months ago, we thought that in the remapping of the heat fluxes is safer to use a different coupling method wrt the one we were previously using, i.e. globpos instead of gsspos.
it means that the following lines could be added in the oasis section of the runscript (to avoid modifying directly the awicm3.yaml file):
coupling_target_fields: rstas.nc:
to have 'heat_ico <--gauswgt_glo-- A_Q_ice' instead of 'heat_ico <--gauswgt_gss-- A_Q_ice'
with
I added the new gauswgt_glo coupling method into the awicm3.yaml file in my esm_tools folder but it should maybe be added in the default version so that if someone else would also run the same configuration or a different one of Tco319-DART the runscript could anyhow find the new method in it?
I am adding also @JanStreffing that of course helped in locating and solving the issue.