MOM6: +(*)Rescale pressures #1089
Merged
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
Added optional pres_scale arguments to various MOM_EOS.F90 routines to allow pressures to be passed with in various units for flexibility and streamlined dimensional consistency testing. Also added optional rho_scale arguments to the various int_density_dz subroutines. All answers are bitwise identical, but there are new optional arguments to public interfaces.
Rescaled the pressures in Boussinesq pressure force calculations, including changing the units of the densities passed to set_pbce_Bouss and using the new rho_scale and pres_scale arguments to the equation of state routines. All answers are bitwise identical.
Optionally rescale the units of the specific volume integrals. Added new optional SV_scale and pres_scale arguments to int_specific_vol_dp, int_spec_vol_dp_generic, and int_spec_vol_dp_Wright. All answers are bitwise identical, but there are new optional arguments to public interfaces.
Rescaled the pressures and specific volumes in the non-Boussinesq pressure force calculations, including changing the units of the pressures passed to set_pbce_nonBouss and using the new SV_scale and pres_scale arguments to the equation of state routines. All answers are bitwise identical.
Rescaled the pressure used to calculate density integrals in find_eta_3d and find_eta_2d to [R L2 T-2] and used the new pres_scale and SV_scale arguments to int_specific_vol_dp. All answers are bitwise identical.
Rescaled the pressure used in calls to calculate_density_derivs to [R L2 T-2] in calculate_density_derivs. All answers are bitwise identical.
Rescaled pressures to [R L2 T-2] in calculate_vertical_integrals for improved dimensional consistency testing. All answers are bitwise identical.
Added new RL2_T2_to_Pa and W_m2_to_RZ3_T3 elements to the unit_scale_type for code simplification and clarity. Also corrected spelling errors in the get_param descriptions of 5 scaling factors, which will change comments in the MOM_parameter_doc.debugging files. All answers are bitwise identical.
Replaced products of scaling factors (like US%R_to_kg_m3*US%L_T_to_m_s**2) with combined scaling factors (like US%RL2_T2_to_Pa) to simplfy and clarify the code. All answers are bitwise idenical.
Rescaled the units of forces%p_surf, fluxes%p_surf, forces%p_surf_full and fluxes%p_surf_full and related surface pressure variables to [R L2 T-2 ~> Pa] for expanded dimensional consistency testing. All answers are bitwise identical, although there are changes to the rescaled units of elements to two transparent data types.
Revised the dimensional rescaling of ice_shelf_CS%g_Earth to match GV%g_Earth and ice_shelf_dyn_CS%g_Earth to minimize confusion when examining different parts of the code. Also cancelled out pairs of unit conversion factors when setting the ice shelf contributions to fluxes%p_surf and forces%p_surf. All answers are bitwise identical.
Added a new optional pres_scale argument to the calculate_TFreeze interfaces to rescale pressures for dimensional consistency testing. All answers are bitwise identical.
Use the new pres_scale argument to TFreeze and pass rescaled pressures to calculate_TFreeze and several instances of calculate_density. All answers are bitwise identical.
Added variants of calculate_density routines that use a hor_index_type to specify array extents and unit_scale_types for dimensional consistency testing, further overloading existing interfaces. Also replaced the recently added rho_scale and pres_scale arguments to int_density_dz with an optional unit_scale_type argument, and modified calls to use this new argument. All answers are bitwise identical, but there are changes to external interfaces.
Revised numerous calls to calculate_density and calculate_density_derivs to use the new form with domain extents indicated by a hor_index_type argument. Internal density variables were also rescaled in a few cases. All answers are bitwise identical.
Rescaled the units of tv%P_Ref to [R L2 T-2] for expanded dimensional consistency testing. In some cases, other pressure variables were also rescaled and calls to calculate_density are recast into the simpler G%HI forms. All answers are bitwise identical, but the scaled units of an element of a transparent type were rescaled.
Rescaled the reference pressure arguments to 3 set_coord routines, 5 initialization routines, and kappa_shear_column. Also removed the unused pres argument to convert_temp_salt_for_TEOS10 and replaced its ocean_grid_type argument with a hor_index_type. All answers are bitwise identical.
Replaced the remaining pres_scale arguments the various calculate_density and calc_spec_vol routines in MOM_EOS.F90 with new optional unit_scale_type arguments. When the scale and US arguments are present, density is scaled by the product of the indicated scaling factors. Calls to these routines in 11 files were modified accordingly. All answers are bitwise identical.
Corrected pressure unit documentation in comments in 5 files. Also fixed punctuation in comments in MOM_EOS.F90. All answers are bitwise identical.
Rescaled internal pressure, specific volume and energy variables in MOM_diapyc_energy_req.F90. This file is mostly used for testing, and all answers are bitwise identical.
Rescaled pressures in 17 calculate_density or related calls and internal pressure variables in MOM_diabatic_aux. All answers are bitwise identical.
Corrected a rescaling bug in calculate_spec_vol_derivs_H1_1d, but as this code was not yet being exercised, there are no answer changes. Also modified applyBoundaryFluxesInOut to use this fixed routine. All answers are bitwise identical.
Redirected complicated equation of state routines to all work via the same 1-d array versions of the code. This shortens the MOM_EOS.F90 code even as new routines are added under the same overloaded interfaces. All answers are bitwise identical.
Rescaled the units of the pressure, density, and gravitational acceleration arguments to find_depth_of_pressure_in_cell, frac_dp_at_pos, trim_for_ice and cut_off_column_top for dimensional consistency testing and code simplification. One change corrected an omitted scaling factor when the optional z_tol argument is omitted, but this does not impact any solutions as this argument is present in all active cases. Also cleaned up some bizarre error messages starting with 'Blurgh!', which is apparently an artificial expletive invented to circumvent censors, but has now been censored and replaced with a more informative message. All answers are bitwise identical, but the units of some arguments have changed and there are new unit_scale_type arguments to some routines.
Made both arguments to unit_scaling_init optional to enable the use of this routine to initialize unscaled unit_scale_types for certain types of unit testing. All answers are bitwise identical.
Rescaled pressure and density variables in MOM_neutral_diffusion and added numerous comments describing internal variables and their units. Some unused variables were deleted, including unused pressure arguments to find_neutral_pos_linear. In addition unit_scale_type arguments were added to 8 subroutines, including neutral_diffusion_init and tracer_hor_diff_init. All answers are bitwise identical, but there are changes to public interfaces.
Use the US form of calculate_density calls in ISOMIP_initialization. All answers are bitwise identical.
Nullified a pointer used in neutral diffusion unit tests. Without this correction of a problem introduced two commits ago, these unit tests would sometimes fail, but all solutions are bitwise identical.
Rescaled 3 diagnosed densities and the pressures used to calculate them. All answers are bitwise identical.
|
It appears you did not enable the option that "allows maintainers to make edits" so you alone can make the changes. |
Removed the interfaces using G%HI to calculate_density and related routines and added a new variant with a new optional argument 'dom' specifying a two- element integer array with the start and end values of the domain to compute on for 1-d arrays starting at 1. If this array is not present in this new variant, calculations are done over the entire output array extent. All calls to the interfaces from before the rescale_pressure pull request will still work. Also added the new function EOS_domain that sets this two-element domain array from a horiz_index_type. This new interface is used throughout the code where the old, removed form was in use. All answers are bitwise identical.
- As part of the process of disconnected EOS from the unit_scaling_type this adds the necessary unit conversions to the EOS_type. - Initialization is currently donne by passing US to MOM_init() but ultimately it seems passing p_scaling, etc., to MOM_init() would remove all dependency on US. - No APIs other than EOS_init() have been changed yet.
Use the new variant of calculate_density with the 'dom' argument or no array extent argument in calls in 15 files. All answers are bitwise identical.
- Removes US as an argument wherever it was optional since the unit conversion factors are not stored in the EOS type.
- Resolved conflicts
Applied dimensional rescaling to many of the internal calculations in the 4 MOM_CVMix files, although calls to external CVMix routines still use the original MKS units. These changes include rescaling of the input and output variables associated with the calculate_density routines. All answers are bitwise identical.
Rescaled internal density and pressure variables in coord_adapt, as well as some input parameters. These changes include rescaling of the input and output variables associated with the calculate_density routines. One variable that was being reused with different units has been split into two, and there are new arguments to build_grid_adaptive, build_adapt_column, and init_coord_adapt. All answers in the MOM6-examples test suite are bitwise identical.
Changed to the new interfaces for calculate_density and related calls in 22 places. All answers are bitwise identical.
… Adcroft_rescale_pressure Merge branch 'rescale_pressure' of git://github.com/adcroft/MOM6. This merge compiles, but it fails the dimensional consistency testing and will be fixed shortly.
Corrected problems with dimensional consistency testing in MOM_EOS.F90 that had been introduced with a recent merge. All answers are bitwise identical and are once again passing dimesional consistency testing.
Eliminated the US argument to find_depth_of_pressure_in_cell, which was no longer being used. Also stored EOS_domain values in MOM_state_initialization for reduced overhead. All answers are bitwise identical.
Store the return values from EOS_domain for computational efficiency. Also cleaned up unneeded 'dom=' declarations in compute_density calls. All answers are bitwise identical.
Corrected a diagnostic halo extent in a recently added EOS_domain call and added new variables to two openMP directives. All answers are bitwise identical.
Eliminated the use GV%mks_g_Earth throughout the MOM6 code. This variable is being retained and is still set to avoid breaking any user code that might be using it. All answers are bitwise identical.
|
The 22 recent commits added to this PR address all of the concerns raised in reviews of the earlier version of his PR. The additional commits that were added to this PR to address these issues include:
This revised PR is being tested with https://gitlab.gfdl.noaa.gov/ogrp/MOM6/pipelines/10226. |
Corrected compile time errors, related to an incomplete implementation of the pressure rescaling in mom_surface_forcing_mct.F90. With this fix, the changes relative to dev/gfdl are now similar between mom_surface_forcing_mct.F90 and the equivalent files for the other couplers. All answers in the MOM6-examples test cases are bitwise identical.
adcroft
approved these changes
Apr 20, 2020
There was a problem hiding this comment.
This is basically ready to go except @Hallberg-NOAA nad I just discussed a small cleanup in neutral diffusion. Needs to be handed manually due to parameter doc changes.
Removed unused US arguments to routines in MOM_neutral_diffusion.F90. All answers are bitwise identical, but some internal interfaces have one fewer argument, and have been returned to their form in the dev/gfdl version.
adcroft
approved these changes
Apr 20, 2020
There was a problem hiding this comment.
@marshallward Can you take this from here? MOM6-examples will have new doc files (.debugging) so needs a manual merge.
|
Updated Gaea regression test: https://gitlab.gfdl.noaa.gov/ogrp/MOM6/pipelines/10234 Passed with no regressions (up to parameter docs), I will merge this now. |
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 set of changes that introduce rescaling of pressures and
related variables for dimensional consistency testing. These changes include
extensive extensions to the interfaces for the equation of state and the
introduction of new arguments to several routines. The new interfaces are
widely used throughout the MOM6 code. There are no changes to solutions,
diagnostics, or output files, but answers could change in cases that were trying
to use ALE remapping and a RHO coordinate with a reference pressure other than
the previously hard-coded value of 2000 dbar, as the reference pressure will now
be properly set at run-time. The commits in this PR include: