add CFC11, CFC12 tracers #4552
add CFC11, CFC12 tracers #4552
Conversation
|
@maltrud, I'm happy to review this. So far, BGC support in compass is very limited and, from what you have told me, significantly out of date or at least out of sync with your workflow. I can easily run some "do-no-harm" tests in compass. But I'm open to suggestions for how I can test this more rigorously. It would be really good to get some up-to-date BGC tests into compass so these parts of the code get exercised on a more regular basis. That would take someone dedicating a good chunk of time to learning compass or me dedicating time to getting spun up on BGC, so not something that will happen without some buy-in from you and others working on BGC. |
|
@xylar i think for now the "do no harm" path is sufficient. it certainly worked with ideal age where i actually did some harm. we can consider more BGC (or more generally, passive tracers) wrt compass in the future. |
|
Okay, sounds good. |
|
With last commit, tested with gnu and intel on grizzly, optimized and debug, with CFCs off. Passes @maltrud can you explain your testing with CFCs on? That can compliment my 'do no harm' testing with CFCs off. |
There was a problem hiding this comment.
I tested compass (without CFCs) using the pr test suite on Chrysalis with intel and impi. All results (including the cosine bell convergence test) were bit-for-bit with the most recent benchmark.
I do not feel qualified to do a more extensive review of the code itself. At a glance, it seems reasonable but I do not know very much about how CFCs are to be treated in MPAS-Ocean.
|
@mark-petersen i tested with CFCs on in several G-cases using the EC30to60r2 mesh. i ran for 60 years (1958-2007) and compared to observations and results from CESM to confirm that the code appears to be working as expected. |
There was a problem hiding this comment.
Thanks @maltrud. Along with tests last Thursday, I merged to today's master and it passes:
SMS_D_Ln9.T62_oQU120_ais20.MPAS_LISIO_TEST.chrysalis_gnu
SMS_D_Ln9.T62_oQU120_ais20.MPAS_LISIO_TEST.chrysalis_intel
PET_Ln3.T62_oEC60to30v3wLI.GMPAS-DIB-IAF-ISMF.chrysalis_gnu
A side note: ERS_Ld3.T62_oQU240.GMPAS-IAF.chrysalis_gnu fails both master and this PR.
Add CFC11, CFC12 tracers This PR adds the capability for the ocean to simulate CFC11 and CFC12 as passive tracers following the OCMIP protocol. This capability is inactive by default and requires a time series of annual average atmospheric mole fractions to be read in as a new stream in streams.ocean. [NML] [BFB]
|
passed e3sm_developer on chrysalis with expected NML DIFFs, merged to next |
|
merged to master and expected NML DIFFs blessed |
This PR adds the capability for the ocean to simulate CFC11 and CFC12 as passive tracers following the OCMIP protocol (http://ocmip5.ipsl.jussieu.fr/OCMIP/phase2/simulations/CFC/HOWTO-CFC.html). This capability is inactive by default and requires a time series of annual average atmospheric mole fractions to be read in as a new stream in streams.ocean.
[NML]
[BFB]