GFDL generic_tracers with MOM6 NUOPC cap #90

dougiesquire · 2024-01-15T04:19:44Z

This PR is a WIP to enable GFDL generic_tracers to be used with ACCESS-OM3.

The important changes in this PR are in the MOM6 directory. Currently, edits to CMEPS and CDEPS are also included to add oxygen to the jra55do datm datamode, but these were mostly done to familiarize myself with the codebases and will likely be removed. UPDATE: I've now reverted my changes to CMEPS and CDEPS.

Details of the changes to MOM6 can be found in MOM6/generic_tracer_nuopc.md.

For a corresponding ACCESS-OM3 configuration, see here. A few changes to Payu are required to run - see here.

Progress towards #42

micaeljtoliveira

I've had a look at the whole set of changes. Overall all looks reasonable. In the process I noticed a few things and left some comments.

A more general question: are the changes to CDEPS and CMEPS generic enough that they could be accepted upstream?

.gitmodules

build.sh

MOM6/patches/mom_ocean_model_nuopc.F90.patch

MOM6/patches/mom_cap.F90.patch

MOM6/extra_sources/mom_cap_gtracer_flux.F90

dougiesquire · 2024-01-16T03:51:03Z

Thanks for taking a look @micaeljtoliveira!

A more general question: are the changes to CDEPS and CMEPS generic enough that they could be accepted upstream?

They're generic, but I doubt they'd be wanted. But as I mentioned in the PR description, I will probably remove these changes anyway. All they do is add oxygen to the jra55do datm datamode, mostly as a learning exercise for myself and as a test of the coupling I added. I doubt we want to keep this - it's probably easier/cleaner to prescribe oxygen using the data override stuff I added, rather than including a new stream in our datm.

dougiesquire · 2024-01-16T06:34:27Z

@angus-g, you mentioned you might be able to take a look at this. If you have time, any feedback would be much appreciated!

access-hive-bot · 2024-01-17T05:06:20Z

This pull request has been mentioned on ACCESS Hive Community Forum. There might be relevant details there:

https://forum.access-hive.org.au/t/cosima-twg-meeting-minutes-2023/399/17

access-hive-bot · 2024-01-17T05:11:59Z

This pull request has been mentioned on ACCESS Hive Community Forum. There might be relevant details there:

https://forum.access-hive.org.au/t/cosima-twg-meeting-minutes-2024/1734/1

angus-g

All in all, this looks fairly good, in my opinion. Most of the heavy lifting seems to be done by existing code lifted from the FMS coupler, and just a bit of glue in the NUOPC layer to get it working.

My only real concern would be about keeping in sync with the FMS coupler for any bugfixes, if required? Having said that, this is probably still the best way of going about it, rather than pulling in the entire coupler as a dependency!

MOM6/patches/mom_cap.F90.patch

angus-g · 2024-01-23T23:54:03Z

MOM6/patches/mom_cap.F90.patch

+  ! Set NUOPC attribute additional_restart_dir to RESTART/ if not defined
+  additional_restart_dir = "RESTART/"
+  call NUOPC_CompAttributeGet(gcomp, name="additional_restart_dir", value=cvalue, &
+        isPresent=isPresent, isSet=isSet, rc=rc)
+  if (ChkErr(rc,__LINE__,u_FILE_u)) return
+  if (isPresent .and. isSet) then
+    additional_restart_dir = slasher(cvalue)
+  else
+    call ESMF_LogWrite('MOM_cap:additional_restart_dir unset. Defaulting to '//trim(additional_restart_dir), &
+          ESMF_LOGMSG_INFO)
+  endif
+  call NUOPC_CompAttributeSet(gcomp, name="additional_restart_dir", value=additional_restart_dir, rc=rc)
+  if (chkerr(rc,__LINE__,u_FILE_u)) return


The "additional" restart dir is used because you're using FMS to write the tracer restart, rather than NUOPC? Maybe it's not important, but I wonder if it's worth a more explicit name?

Yeah (although NUOPC only really specifies how the files are handled, MOM does the writing). The name is general because it's a general config parameter that in theory it could be used by any model component handled by NUOPC. I'll see what I can do to make this clearer. One option is just to hardcode the directory (it wouldn't be the first).

The main problem is that NUOPC wants the restart input and output directories to be ./ for all components. The restart filenames include a timestep to prevent files from being overwritten and a pointer file keeps track of the latest restart. The name of the additional restart file needed for the generic tracer fluxes is set within each generic_tracer module (i.e. I don't want to change it) and does not include a timestamp, so writing to and reading from MOM's restart_output_dir and restart_input_dir (=./), respectively, causes problems.

MOM6/patches/mom_cap_methods.F90.patch

dougiesquire · 2024-01-24T03:41:40Z

Thanks heaps for the review @angus-g. I share your concern about staying on top of changes to FMScoupler. But I'm not sure of a better approach. It makes me feel slightly better that the routines we're ripping don't seem to change very often: they haven't changed at all since the files were renamed 3 years ago (atmos_ocean_fluxes_calc.F90, atmos_ocean_dep_fluxes_calc.F90)

failing with "The domain associated with the file: does not have an io_domain."

The NUOPC cap appears to have been written from the MCT cap, which does not include changes needed to successfully register restarts for FMS coupler type. See NCAR/MOM6@73304eb

…tart_dir

move new routines into new mom_cap_gtracer_flux.F90 now only mom_cap_gtracer_flux.F90 adds new calls directly into FMS now only adding _USE_GENERIC_TRACER ifdefs to mom_cap.F90

…from FMScoupler commit: eeadda8dc74f605e9cb3b03be8a43ab2ce698496

operate on 2D inputs, rather than 1D add calculation for air_sea_deposition from atmos_ocean_dep_fluxes_calc.F90 multiply fluxes by ice_fraction input, rather than masking based on seawater input use MOM over FMS modules where easy to do so make tsurf input optional, as it is only used by a few implementations use ind_runoff rather than ind_deposition in runoff flux calculation rename gas_fields_ice to gas_fields_ocn add param array into Ice_ocean_boundary%fluxes after spawning

reformat MOM_cap_gtracer_flux.F90

pass optional atm_fields to mom_import as 2D BC type do atm_fields override and flux calcs from mom_cap

spawn 2D coupler_type for atmos fields in InitializeAdvertise and set in ESMF internal state manually override IOB%fluxes since FMS coupler_type routine doesnt set override flag update generic_tracer_nuopc.md

revert changes in build.sh revert changes adding oxygen to datm tidy MOM cap and add links where code was copied from other sources

move allocation of coupler field work array outside of loop

…nknown to allow flux calculation with overridden pcair fields

dougiesquire · 2024-05-13T05:57:12Z

I've done some basic tests to show that these changes don't impact performance when no generic tracers are configured - see here.

@micaeljtoliveira I've made a couple of minor changes since your last review. Would you mind taking a quick look and approving if you're happy. I'd like to get this merged.

I'm not sure whether to squash this or not. There's useful info in many of the individual commits. But I made a bit of a mess of the history with the last few commits because the files being patched changed quite a bit and the conflicts were very difficult to resolve (I basically ended up regenerating the patches...).

dougiesquire marked this pull request as draft January 15, 2024 04:19

dougiesquire requested a review from micaeljtoliveira January 15, 2024 04:20

micaeljtoliveira reviewed Jan 16, 2024

View reviewed changes

dougiesquire requested a review from angus-g January 16, 2024 06:34

angus-g reviewed Jan 24, 2024

View reviewed changes

dougiesquire and others added 20 commits March 4, 2024 09:13

add generic_tracers submodule

6b7f4b3

update build for generic_tracers

4b9de4d

initialise fms

e0b2124

initialise gas exchange structures

7c75971

spawn Ice_ocean_boundary%fluxes

03a5294

register generic tracer coupler restarts

457ee1c

failing with "The domain associated with the file: does not have an io_domain."

replicate FMS cap initialize_ocean_public_type

84a86bf

The NUOPC cap appears to have been written from the MCT cap, which does not include changes needed to successfully register restarts for FMS coupler type. See NCAR/MOM6@73304eb

write generic tracer flux restarts to nuopc-configured additional_res…

a828de7

…tart_dir

couple co2 and o2 concentrations

f018e3c

regorganise

bef47e6

move new routines into new mom_cap_gtracer_flux.F90 now only mom_cap_gtracer_flux.F90 adds new calls directly into FMS now only adding _USE_GENERIC_TRACER ifdefs to mom_cap.F90

add function for getting CMEPS field name from FMS flux name

ff734df

move to COSIMA fork of NOAA-GFDL/ocean_BGC generic tracers repo

15baa23

copy atmos_ocean_fluxes_calc.F90 and atmos_ocean_dep_fluxes_calc.F90 …

e46f873

…from FMScoupler commit: eeadda8dc74f605e9cb3b03be8a43ab2ce698496

move atmos_ocean_fluxes_calc routine into MOM_cap_gtracer_flux module

1789149

reformat MOM_cap_gtracer_flux.F90

add ability to override atm fields and IOB fluxes from data_table

1a7b062

reorganise

3975a08

pass optional atm_fields to mom_import as 2D BC type do atm_fields override and flux calcs from mom_cap

add generic_tracer_nuopc.md

ba30ca0

allow diagnostics for tracer fluxes and related fields

1bb3fcf

spawn 2D coupler_type for atmos fields in InitializeAdvertise and set in ESMF internal state manually override IOB%fluxes since FMS coupler_type routine doesnt set override flag update generic_tracer_nuopc.md

address micaels review

6b8d46a

revert changes in build.sh revert changes adding oxygen to datm tidy MOM cap and add links where code was copied from other sources

address angus review

bcca76e

move allocation of coupler field work array outside of loop

dougiesquire force-pushed the mom-generic-tracers branch from 556c94e to bcca76e Compare March 3, 2024 22:15

set u10 and psurf on atm_fields even when pcair cmeps field name is u…

d2d3023

…nknown to allow flux calculation with overridden pcair fields

dougiesquire mentioned this pull request Mar 27, 2024

1deg_jra55do_ryf_wombatlite COSIMA/MOM6-CICE6#47

Draft

dougiesquire and others added 5 commits March 28, 2024 16:47

add WOMBAT generic tracer module

9e0f63c

zero det_sediment and caco3_sediment tracers below k=1

791e0d9

rename WOMBAT to WOMBATlite

2a8b690

Merge branch 'main' into mom-generic-tracers

5aadebb

Merge branch 'main' into mom-generic-tracers

99f7cf2

dougiesquire self-assigned this May 2, 2024

dougiesquire mentioned this pull request May 2, 2024

Port WOMBAT to MOM6 #42

Open

dougiesquire mentioned this pull request May 13, 2024

Performance testing with generic WOMBAT #155

Open

dougiesquire marked this pull request as ready for review May 13, 2024 05:51

dougiesquire requested a review from micaeljtoliveira May 13, 2024 06:01

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

GFDL generic_tracers with MOM6 NUOPC cap #90

GFDL generic_tracers with MOM6 NUOPC cap #90

dougiesquire commented Jan 15, 2024 •

edited

micaeljtoliveira left a comment

dougiesquire commented Jan 16, 2024

dougiesquire commented Jan 16, 2024

access-hive-bot commented Jan 17, 2024

access-hive-bot commented Jan 17, 2024

angus-g left a comment

angus-g Jan 23, 2024

dougiesquire Jan 24, 2024

dougiesquire Jan 24, 2024 •

edited

dougiesquire commented Jan 24, 2024 •

edited

dougiesquire commented May 13, 2024

GFDL generic_tracers with MOM6 NUOPC cap #90

Are you sure you want to change the base?

GFDL generic_tracers with MOM6 NUOPC cap #90

Conversation

dougiesquire commented Jan 15, 2024 • edited

micaeljtoliveira left a comment

Choose a reason for hiding this comment

dougiesquire commented Jan 16, 2024

dougiesquire commented Jan 16, 2024

access-hive-bot commented Jan 17, 2024

access-hive-bot commented Jan 17, 2024

angus-g left a comment

Choose a reason for hiding this comment

angus-g Jan 23, 2024

Choose a reason for hiding this comment

dougiesquire Jan 24, 2024

Choose a reason for hiding this comment

dougiesquire Jan 24, 2024 • edited

Choose a reason for hiding this comment

dougiesquire commented Jan 24, 2024 • edited

dougiesquire commented May 13, 2024

dougiesquire commented Jan 15, 2024 •

edited

dougiesquire Jan 24, 2024 •

edited

dougiesquire commented Jan 24, 2024 •

edited