+*More consistent non-Boussinesq mixed layer thicknesses #610
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
Hallberg-NOAA
added
enhancement
Hallberg-NOAA
force-pushed
the
NonBous_ML_thickness
branch
from
b97034e to
9031a83
Compare
marshallward
approved these changes
May 30, 2024
|
@Hallberg-NOAA There is one minor conflict that needs to be fixed. |
Added the routine convert_MLD_to_ML_thickness to consistently convert the mixed layer depths (in units of [Z ~> m]) back to the mixed layer thicknesses (in [H ~> m or kg m-2]), with proper error checking and handling of the non-Boussinesq case. The body of this routine was taken from duplicated blocks of code in MOM_mixedlayer_restrat.F90. This is not tested directly in this PR, but it was tested via revisions to MOM_mixedlayer_restrat.F90 that are included in a subsequent commit. All answers are bitwise identical, but there is a new publicly visible interface.
Pass mixed layer thickness, rather than mixed layer depth, to ideal_age_tracer_column_physics, and determine the number of layers within the mixed layer in count_BL_layers in thickness units rather than depth units. To accommodate these changes there is now a call to convert_MLD_to_ML_thickness inside of call_tracer_column_fns. The thermo_var_ptrs type argument (tv) to ideal_age_tracer_column_physics and count_BL_layers are no longer used and have been removed. All answers are bitwise identical in Boussinesq mode, but in non-Boussinesq mode there are changes in the passive ideal age tracers at the level of roundoff. There are also changes in the units of arguments and the number of arguments to a public interface.
Provide the mixed layer thickness, as well as the mixed layer depth as arguments to mixedlayer_restrat. The code that had previously been used to convert between the two has now been removed to the new external function convert_MLD_to_ML_thickness. To accommodate these changes, a new element, h_ML, was added to the vertvisc type, and a call to convert_MLD_to_ML_thickness was temporarily added in step_MOM_dynamics just before the call to mixedlayer_restrat. All answers are bitwise identical, but there are changes to a public interface.
Moved the call to convert mixed layer depths into mixed layer thicknesses from right before the call to mixedlayer_restrat into the various diabatic routines just after they are calculated. This code rearrangement changes answers in non-Boussinesq mode because the layer specific volumes will have evolved between these two calls, but it is consistent with the mixed layer thicknesses as used in the boundary layer parameterizations. A new argument was added to bulkmixedlayer to return the mixed layer thickness that was already being calculated. The previous argument Hml was renamed to BLD and changed from a pointer into a simple array. Both are intent(inout) rather than intent(out) to preserve values in halos. This commit also revises the logic around the allocation of visc%h_ML and its registration as a restart variable, including handling cases where an older restart file is being read that includes MLD but not h_ML. Because the mixed layer depth is used in almost cases, CS%Hml in the control structure for MOM.F90 was changed from a pointer to an allocatable, with values of 0 in those cases (e.g., when ADIABATIC is true) where it is not used. In addition, the argument Hml to the various diabatic routines was changed to MLD to more clearly reflect that it is a depth and not a thickness. Outside of diabatic, BML is only used to provide the tracer point boundary layer depths to the calling routines, so it does not need a halo update. Instead, all of the halo updates for the various elements of visc that do need halo updates after the diabatic calls are collected into one place for efficiency and more accurate timings. The scale arguments of 34 checksum calls were revised from GV%H_to_m to GV%H_to_MKS for more accurate checksums in non-Boussinesq configurations by avoiding multiplication by an reference specific volume, and instead only rescaling by an integer power of 2. All answers are bitwise identical in Boussinesq mode, but in non-Boussinesq mode there are changes in answers in cases that use the boundary layer thicknesses obtained from the boundary layer parameterizations in subsequent calculations, such as mixed layer restratification with some options. There are also changes to the arguments of several publicly visible routines.
Pass visc%h_ML to applyBoundaryFluxesInOut, which avoids the need for the thickness conversions inside of that routine. Answers are bitwise identical for Boussinesq cases, but they will change in non-Boussinesq cases because the layer specific volumes have changed between the time the boundary layer thicknesses have been calculated and when they are being used in the brine plume parameterization. Also use visc%h_ML in place of calls to KPP_get_BLD or energetic_PBL_get_MLD in step_MOM_dyn_split_RK2, hor_bnd_diffusion and neutral_diffusion_calc_coeffs. This change requires a new vertvisc_type argument to tracer_hordiff, hor_bnd_diffusion and neutral_diffusion_calc_coeffs. Boussinesq answers are bitwise identical, but non-Boussinesq answers change because they now use the actual model specific volumes rather than some prescribed constant value to to the relevant unit conversions. The logic in set_visc_register_restarts was also updated so that visc%MLD and visc%h_ML are allocated and registered for restarts in the cases where they will be used. All answers are bitwise identical in Boussinesq mode, but in non-Boussinesq mode there are changes in answers in cases that use neutral tracer diffusion that excludes the boundary layer, horizontal tracer diffusion in the boundary layer or FPMIX. There are also changes to the arguments of several publicly visible routines.
Added the optional argument h_BL to call_tracer_column_fns, and then pass visc%h_ML to this routine as a part of the calls from the diabatic routines. This argument is not provided in adiabatic mode or in offline tracer mode, which is why it has been made optional. All answers are bitwise identical in Boussinesq mode but ideal age tracers can change in non-Boussinesq mode due to the updates to the layer specific volumes between the calculation of the boundary layer properties and the calls to call_tracer_column_fns.
Hallberg-NOAA
force-pushed
the
NonBous_ML_thickness
branch
from
9031a83 to
75b9437
Compare
|
Gaea regression: https://gitlab.gfdl.noaa.gov/ogrp/mom6ci/MOM6/-/pipelines/23666 ✔️ |
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.
This PR includes a series of commits that standardize how the depth of active surface mixing is passed around and how and where that depth is converted into a thickness when in non-Boussinesq mode. This is accomplished via calls to the new routine
convert_MLD_to_ML_thickness()and by accessing the new variablevisc%h_MLin the various places where the mixed layer thickness is used. Because this variable may be used from a previous timestep, it has been added to the restart file in cases where it is in use. These changes make the use of the surface boundary layer thickness much more consistent with how it is calculated when in non-Boussinesq mode, without changing anything in Boussinesq mode.This PR includes a number of changes to public interfaces. There is a new argument to
mixedlayer_restrat(),mixedlayer_restrat_Bodner(),bulkmixedlayer(),tracer_hordiff(),hor_bnd_diffusion()andneutral_diffusion_calc_coeffs(), along with changes to the units of arguments to mixedlayer_restrat_OM4 andapplyBoundaryFluxesInOut(). One widely used argument in diabatic and its children was converted from a pointer to an ordinary 2-d array renamed fromHmltoBLDfor more clarity about its meaning and units. In the ideal age tracer package,ideal_age_tracer_column_physics()has one less argument and changed units to another, while the units of all variables incount_BL_layers()were revised. There is an additional variable the restart files in some cases.This PR will change answers in the fully non-Boussinesq mode, but because this mode is not yet being used in production, there is not an answer-date or bug flag that will recover the previous non-Boussinesq answers. All Boussinesq answers are bitwise identical.
All answers are bitwise identical in Boussinesq mode, but in non-Boussinesq mode there are changes in answers in some cases. There is a new element in the
vertvisc_type, the new publicly visible routineconvert_MLD_to_ML_thickness()and a number of new or modified arguments to existing publicly visible interfaces.The commits in this PR include: