Add MARBL to MOM6 #157
Merged
Add MARBL to MOM6 #157
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Copied dye_example.F90, renamed all public routines. This module does not actually tie into MARBL at this time.
Adding USE_MARBL_TRACERS = True to override file turns on MARBL tracers. At this point, we call marbl_instances%init and register all 32 tracers but don't do anything else (so they are initialized to 0 and there is no source term for advecting them yet)
CESM will use MARBL via manage_externals, but other systems may need to bring it in via submodules
I think this is just needed for TravisCI, since it doesn't know to look for code in pkg/
This looks like it was removed from everywhere else when I merged in the latest dev/ncar branch
Created new configure_MARBL_tracers() subroutine to be called from call_tracer_register() [between get_param() calls and register_* calls]
Also uses put_settings() to update the MARBL settings before initialization (tested by setting ciso_on = .true. via user_nl_marbl)
marbl_instance%StatusLog is written after the call to init and in marbl_tracers_end (which is now called from tracer_flow_control_end())
Failed a TravisCI test due to missing documentation
Also added a placeholder for parsing the MARBL timing information
Every call to print_marbl_log() is followed by a call to %erase()
Instead of passively advecting 0s, the MARBL_tracers module now correctly initializes each tracer (but doesn't compute any source-sink terms yet)
Went from 0a806cf to 479f914
call the function from register_MARBL_tracers() rather than MOM_tracer_flow_control.
MARBL will be brought in to CESM via manage_externals, and we will use -DUSE_MARBL_TRACERS to build with access to MARBL.
Now that it comes in from manage_externals, we want to ignore it
Can build MOM without pkg/MARBL, but if USE_MARBL_TRACERS is True in the param file and the build does not include -D_USE_MARBL_TRACERS then the model aborts.
Will allow MARBL diagnostics to be added to history files.
And, from that routine, call marbl_instances%surface_flux_compute(). Note that forcings, surface tracers, and surface fluxes are all zeroed out in this commit. I'd like to get diagnostics posted in the next commit, and then I can start updating tracer surface values, saving saved state, and looking into how to read forcing fields.
Looks like column_physics() is the better place for this call
calls post_data (note that created a temporary data structure to hold both the diagnostic id and a buffer to fill column-by-column as MARBL runs)
During configuration, set indices for each of the surface flux forcings so that each forcing can be set to a different value in column_physics(); all are set to zero except u10_sqr (2.5e5), atmpress (1), and xco2 / xco2_alt_co2 (284.7)
With solo_driver, the following runs are all bit-for-bit with non-scaled runs: C_RESCALE_POWER = 10 H_RESCALE_POWER = 10 L_RESCALE_POWER = 10 S_RESCALE_POWER = 10 T_RESCALE_POWER = 10 Z_RESCALE_POWER = 10
Should pass doxygen test again
The function is meant to help copy fields from the ice_ocean_boundary_type (which is in physical units in all the caps) to the forcing_type (which wants scaled units). So the solo_driver should NOT scale the dust, black carbon, or NDEP inputs from data_override, and instead that scaling should happen in MARBL_forcing_mod.F90
applyTracerBoundaryFluxesInOut expects in_flux_optional in units of conc H, and we were passing conc m T/s. Since riv_flux_loc is now conc H, I also added a debug-gated hchksum on it.
The dimensional scaling tests fail if the MARBL tracer concentrations are very very small (O(1e-300)); this can be avoided by setting the minimum tracer value to be 1e-100 instead of 0. We don't want to do this for production runs, though, so the default for this parameter is still 0.
Sa_co2prog and Sa_co2diag should not be area corrected (they are states) but Faoo_fco2_ocn should be (it's a flux)
When calling marbl_instance%init(), we should tell MARBL that MOM6 doesn't have the global operators that MARBL expects (global sums / running means) so we get the appropriate error message when trying to run with ladjust_bury_coeff = True
Updated ice_ocn_bnd_type_chksum() in the NUOPC cap, though I don't think this function is ever called
I had created CESM_INPUTDATA as a parameter to point to my work directory, but it is no longer necessary because INPUTDATA points to the CESM input data repository and I've moved necessary files there
|
@marshallward -- we (@alperaltuntas, @gustavo-marques, @klindsay28, and I) are starting the review process of this PR, and had a couple of questions that I hope you can answer:
I'd welcome general feedback here as well, though we're still going through the detailed code review so it might make sense to wait until I've had a chance to fix the issues raised from that (or wait until we try to get this to |
If its purpose is to provide some fine-grain control over In any case, I agree it looks very general purpose and potentially useful to others, and it would be good to move outside of
|
-- cleaned up a lot of comments and whitespace -- used source argument in more allocate statements, and deallocated more arrays -- 3D diags now have zl:mean in cell_methods attribute -- marbl_instances%domain%kmt is set once (during initialization)
|
21529b9 addressed most of the concerns from code review. Still remaining
I'll note that I ignored two suggestions from the review:
Next steps:
|
If MARBL is not configured to provide the base biotic tracers, then it will not be able to provide chlorophyll. In that case, if CHL_FROM_FILE=False, MOM6 needs to get chlorophyll from the generic tracers.
To make these subroutines more accessible, they were moved out of src/tracer/ and made available through MOM_interpolate
|
As of 97c0917 I've addressed all the concerns raised in code review, and only see expected test failures (MacOS-related). I'll leave this in draft mode while addressing the companion PR at |
|
(I may also need to merge in the latest |
If variable was described in POP comment, I copied the comment over. Otherwise I came up with a description on my own.
time_interp_external() does not update halo regions, so running CESM with DEBUG=TRUE was triggering some overflows from uninitialized memory. Intead of copying the entire array, we now loop through (is:ie,js:je) when accessing an array returned from time_interp_external()
Most use cases don't include restoring for MARBL tracers, but when that feature is enabled and the time scale is read from a file the user can specify what variable to read from the netCDF file (default is I_TAU to match naming convention in MOM6, but some test cases are based on POP files and will need to read RTAU)
gustavo-marques
changed the title
gustavo-marques
approved these changes
Aug 2, 2024
| call ESMF_FieldGet(lfield, farrayPtr=fldptr2d, rc=lrc) | ||
| if (ChkErr(rc,__LINE__,u_FILE_u)) return | ||
|
|
||
| if (.not. associated(fldptr)) allocate(fldptr(size(fldptr2d,2))) |
There was a problem hiding this comment.
This line causes a memory leak. Any fix suggestions @mnlevy1981 ?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I'm working on adding a new tracer module that calls MARBL, and I've reached a point where it might be useful for other people (e.g. @ashao) to look at the mods I'm making. There will also be a PR to ESCOMP/MOM_Interface necessary to get this running through CESM, and I'll link to that PR once it has been made.