2.1 awiesm 2.1 with transient tracers #570
Conversation
…sea ice; sea ice delta values now passed to ECHAM atmosphere model; minor code cleanup
…ral minor code improvements
…use of water isotope tracers and other tracers; minor code cleanup
Conflicts: src/ice_oce_coupling.F90
- FESOM2.0 simulations on AWI ollie HPC were not always reproducible. Runs with identical namelists, init-conditions ... could lead to sligthly different results. - Natalja solved that issue by changing the compiler flag -fast-transcendentals with -fimf-use-svml - with this option loops with exp, sin, cos ... (IEEE function) will be always computed through the vector implementation otherwise the compiler is not fixed how IEEE funtions should be treated, which can be either Vector, Scalar or something else. All three can give slightly different results
…SOM/fesom2 into paleodyn/awiesm-2.1-gamma
…with coupled model setups.
…elist variables accordingly.
|
As expected, this branch is quite far from refactoring. I wonder also if some of the things done here are intentional: For some reason the PR tries to remove/revert changes made on the refactoring branch. I'm not sure this is easy to fix with git. This will be manual work I'm afraid. |
|
Most of that seems to come from fb6eab8 |
|
I presume that > 95% of the conflicts are not due to the transient tracer code (which can be identified when searching for "use_transit"). How should I/we proceed? |
|
I had a chat with Patrick today. We had two ideas of how we can proceed:
p.s. From after 2.6. we aim to have a bit different merge strategy, which will not see such long times between PR opening and being merged. I think we placed a bit too much emphasis on maintaining bit-reproducibility since 2.1. |
|
May be superseeded by #572. TBD |
|
Superseded by #576 |
This branch includes the abiotic transient tracers 14C, 39Ar, CFC-12, and SF6.
They are enabled and controlled in namelist.config, section &runctl, where the main switch is "use_transit".
Equilibrium / time slice simulations of 14C and 39Ar can be carried out without additional atmospheric input files.
Transient simulations of 14C, CFC-12, and SF6 require additional additional input data which are available on request and could be copied to a central pool directory.
CFC-12 and SF6 are highly recommended for model validation/tuning because their initial concentrations are zero, their transient atmospheric input history = FESOM integration period is short (less than 100 years), and there are many field observations available. The implementation of similar transient tracers (such as CFC-11, CFC-113, or CCl4) would be straightforward.