Closed DeniseWorthen closed 3 years ago
Denise, do we commit this PR now or commit with your DATM-MOM6_CICE6 PR?
On Fri, Sep 18, 2020 at 2:53 PM Denise Worthen notifications@github.com wrote:
@DeniseWorthen commented on this pull request.
In mediator/med_phases_prep_ocn_mod.F90 https://github.com/NOAA-EMC/CMEPS/pull/13#discussion_r491133052:
-
- ! local variables
- type(InternalState) :: is_local
- real(R8), pointer :: ocnwgt1(:) ! NEMS_orig_data
- real(R8), pointer :: icewgt1(:) ! NEMS_orig_data
- real(R8), pointer :: wgtp01(:) ! NEMS_orig_data
- real(R8), pointer :: wgtm01(:) ! NEMS_orig_data
- real(R8), pointer :: customwgt(:) ! NEMS_orig_data
- real(R8), pointer :: ifrac(:)
- real(R8), pointer :: ofrac(:)
- integer :: lsize
- integer :: n
- real(R8) , parameter :: const_lhvap = 2.501e6_R8 ! latent heat of evaporation ~ J/kg
- character(len=*), parameter :: subname='(med_phases_prep_ocn_custom_nemsdata)'
- !---------------------------------------
- if (trim(coupling_mode) == 'nems_orig' .or. trim(coupling_mode) == 'nems_frac') then
For the first part of your question, yes these do not need to be calculated in prep-ocn because they're able to be done via the auto-merge in esmFldsExchange: LN 259 https://github.com/DeniseWorthen/CMEPS/blob/af0d961386080ced7f63a14ca194b4c2bbe5a95d/mediator/esmFldsExchange_nems_mod.F90#L259. This is where you're "merging" (really just weighting by ofrac) for example the sensible heat flux calculated by the compmed.
The CMEPS aoflux calculation is the same aoflux calculation done by the NEMS mediator, using the atm import state. When we switched to bulk formulas for the NEMS mediator, it is equivalent to using just ofrac weighting. I was able to reproduce the current DATM-MOM6-CICE5 baseline with a NEMS mediator using only ofrac https://github.com/DeniseWorthen/NEMS/blob/ea1558ad219150465e42c2e6d95982665f9e5b80/src/module_MEDIATOR.F90#L5478. In this case, all the ocean fluxes are coming from the aoflux calculation, weighted by ofrac.
Some fields are sent separately to MOM6 from ice and atm but that is how MOM6 expects them so that it can do it's budgeting.
— You are receiving this because your review was requested. Reply to this email directly, view it on GitHub https://github.com/NOAA-EMC/CMEPS/pull/13#discussion_r491133052, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI7D6TKJS7IS2GDMT4ATH3DSGOUAVANCNFSM4RNMGLTQ .
I think we need to this this PR first because I haven't retained any NEMS mediator files in my cice6-cmeps datm branch. Committing my cmeps-cice6 branch will involve switching CICE and CMEPS at the same time. How much testing do we need to do for that? Do we need to test NEMS mediator with CICE6 first and then compare NEMS/CICE6 to CMEPS/CICE6?
Yes,s please compare NEMS/CICE6 to CMEPS/CICE6, if the results are comparable, we can switch to CMEPS/CICE6.
I realize I mis-spoke. We can't run NEMS+CICE6 (the cice6 cap is mesh based).
We can compare NEMS+CICE5 with CMEPS+CICE5.
That is good too.
Issue #12, #10
Removes special nems_orig_data prep ocean phase which is no longer required if corresponding changes are made in esmFldsExchange_nems to make use of the auto-merge feature.
Modifies fd_nems.yaml to account for differing field names between the nems active and nems data ATM being used for the same ocean import field
Testing on orion and hera for S2S shows all fields are b4b except for the mediator files. However, manual comparison on the mediator files using either nccmp -d or cpnrc -m show no differences. Logs for S2S on orion are here
This branch was tested in DATM-MOM6-CICE5 for the cold cfsr case (run out to 12 hours) compared with a comparable case using the NEMS mediator. The following figures show the fields imported to the ocean at the mid-way point (6 hours) for the NEMS mediator and CMEPS mediator cases.