add integrated physics#84
Conversation
JosephMouallem
requested review from
gaokun227,
lharris4,
linjiongzhou and
vithikashah001
|
@JosephMouallem this is really great---thanks for getting this together so quickly. Does this code and associated PRs compile correctly, run in coupled mode, and also execute the integrated physics correctly? The latter may need a runscript from @gaokun227 demonstrating this capability. |
lharris4
left a comment
lharris4
left a comment
There was a problem hiding this comment.
I have one minor cleanup request, and do make sure that the model compiles correctly in all appropriate compilation modes.
One interesting question is how to handle the implicit coupling with the integrated physics' PBL. I presume that we can simply set the atmospheric diffusion coefficients to 0 above the surface layer and run the implicit coupling as we ordinarily would?
| blksz(:) = Init_parm%blksz(:) | ||
|
|
||
| !--- set control properties (including namelist read) | ||
| call Model%init (Init_parm%nlunit, Init_parm%fn_nml, & |
There was a problem hiding this comment.
The level of abstraction here is a bit challenging. This calls control_initialize() in GFS_typedefs.F90, correct? Also, is it possible to pass the whole Init_par structure instead of each individual component? This would make it easier to add flags and other data to this structure without needing to modify calls to control_initialize().
There was a problem hiding this comment.
Note that this cleanup is only a "nice" thing to have and isn't a requisite for making the code work correctly.
There was a problem hiding this comment.
I agree, this will be a good improvement and sets things up more cleanly for the future
|
At the moment, this code compiles without issues, we still need to run tests to make sure this is giving the intended behavior. I haven't looked implicit coupling here but the same refactoring logic should be applied here as well. |
|
A suitable testing strategy would be to compare two versions of the uncoupled SHiELD executable: one built from Linjiong Zhou’s latest shield2025_inline branches in his GitHub repositories, and the other incorporating the code changes in this PR. If the implementation here is correct, the two executables running with the inline physics enabled should produce identical results, including physics diagnostics such as surface fluxes, PBL & Conv scheme tendencies. |
3 participants
Add integrated physics from the original PR #77
Credit to @linjiongzhou
Dependencies:
NOAA-GFDL/GFDL_atmos_cubed_sphere#413
NOAA-GFDL/atmos_drivers#72