Add horizontal tapering function to mesoscale eddy parameterizations (AMOC PR 2/6) #4864

darincomeau · 2022-03-30T15:38:08Z

This PR adds a new 'RossbyRadius' option to the horizontal tapering of GM and Redi, which is based on Hallberg (2013) - https://doi.org/10.1016/j.ocemod.2013.08.007, which turns the parameterization on if the Rossby Radius is resolved by at least two cells. These flags are:

config_Redi_horizontal_taper = 'ramp' (default), 'none', 'RossbyRadius' (new) 
config_GM_horizontal_taper = 'ramp' (default), 'none', 'RossbyRadius' (new)

This PR replaces the variable gmResolutionTaper, which was used for both GM and Redi, to the separate variables RediHorizontalTaper and gmHorizontalTaper so that Redi and GM tapering can be chosen separately, and it is clear in the code. Logic was added to the GM and Redi schemes so that the there is a base value in the arrays for Redi Kappa and GM Kappa. The base value is then multiplied by the taper array, which has a value between 0 and 1. Arrays that are static in time are computed on init. Under the following conditions, the arrays are recomputed at every time step:

config_Redi_horizontal_taper = 'RossbyRadius' (recompute RediHorizontalTaper and RediKappa)
config_GM_horizontal_taper =  'RossbyRadius' (recompute gmHorizontalTaper)

[NML] adds MPAS-Ocean namelist options
[BFB]

jonbob · 2022-03-30T16:27:13Z

@darincomeau -- I'll run the scripts that automatically update the bld files to reflect Registry changes for this PR and make sure it's all consistent, unless that's what you've already done.

darincomeau · 2022-03-30T16:34:36Z

@jonbob I haven't - the bld file changes were added manually.

jonbob · 2022-03-30T16:36:11Z

Thanks for getting back to me so quickly @darincomeau. I'll run the scripts and update the files if anything turns up

jonbob · 2022-03-31T23:08:29Z

@darincomeau - this PR looks like it also includes the changes from PR #4835? That PR has not been merged to the repo yet...

darincomeau · 2022-03-31T23:43:16Z

@jonbob yes, this PR started from the head of that branch in #4835 , to avoid merge conflicts later. I'll rebase once that gets merged.

darincomeau · 2022-04-04T16:40:40Z

5 year tests with config_gm_enable_horizontal_resolution_function = .true. compared against baselines with option off:

ECwISC30to60E2r1
This branch with flag on
Baseline

SOwISC12to60E2r4
This branch with flag on
Baseline

components/mpas-ocean/src/shared/mpas_ocn_gm.F

mark-petersen · 2022-04-04T20:21:52Z

Passes MPAS-O nightly regression suite with config_gm_enable_horizontal_resolution_function = true with debug using gnu and intel.

vanroekel

changes look consistent with what I implemented previously so approving on visual inspection and final approval from @jonbob on the supporting changes.

mark-petersen

Approved, based on my testing and E3SM simulations by others.

components/mpas-ocean/src/shared/mpas_ocn_gm.F

darincomeau · 2022-04-06T18:37:54Z

Rebased after #4835 was merged to master.

components/mpas-ocean/src/shared/mpas_ocn_gm.F

milenaveneziani · 2022-05-18T21:42:45Z

That works for me.

xylar · 2022-05-18T23:25:02Z

@milenaveneziani, could you make that name change, too? (It needs to happen in various places.)

milenaveneziani · 2022-05-18T23:26:35Z

@xylar: sure. Is that OK if I do it manually though?

xylar · 2022-05-18T23:29:13Z

Yeah, do it manually. @jonbob will check our work in any case.

milenaveneziani · 2022-05-18T23:44:32Z

ok, done. @mark-petersen: you can go ahead and use this branch for the BFB tests. And please do double-check my changes.

mark-petersen · 2022-05-19T02:28:39Z

components/mpas-ocean/src/shared/mpas_ocn_tracer_hmix_redi.F

-                                     *(dcEdge(iEdge) - config_eddying_resolution_ramp_min) &
-                                     /(config_eddying_resolution_ramp_max - config_eddying_resolution_ramp_min)
+                                     *(dcEdge(iEdge) - config_Redi_horizontal_ramp_min) &
+                                     /(config_Redi_horizontal_ramp_max - config_Redi_horizontal_ramp_min)


This fixed the BFB mismatch. I had to put this block of code in the same order as before, with the operations in the same order as before. Now it is BFB in E3SM tests (it was BFB in stand-alone tests before, which is partly what took me so long to find it).

yay, @mark-petersen, that's great!

mark-petersen · 2022-05-19T03:21:30Z

Now passes compare against branch point 4435bfa for:

SMS_Ln9.T62_EC30to60E2r2.GMPAS-IAF.chrysalis_gnu
SMS_Ln9.T62_EC30to60E2r2.GMPAS-IAF.chrysalis_intel
PEM_Ln9.ne30pg2_EC30to60E2r2.WCYCL1850.chrysalis_gnu
PEM_Ln9.ne30pg2_EC30to60E2r2.WCYCL1850.chrysalis_intel

@jonbob please proceed with your tests.

mark-petersen · 2022-05-19T20:05:41Z

I also tested stand-alone and E3SM with the new flag settings for this PR

config_GM_horizontal_taper = 'RossbyRadius'
config_Redi_horizontal_taper = 'RossbyRadius'

in E3SM, passes

PEM_Ln9.ne30pg2_EC30to60E2r2.WCYCL1850.chrysalis_gnu

in stand-alone nightly suite with all three: config_GM_closure = 'N2_dependent', Visbeck, EdenGreatbatch with gnu debug, and all pass.

mark-petersen · 2022-05-19T20:09:39Z

Any other comments on this one, or is it ready for final testing before the merge? @xylar you still have a red on the approval, can you approve if you agree?

milenaveneziani · 2022-05-19T20:14:18Z

@mark-petersen, as I mentioned (somewhere) above, my latest test is behaving as expected. The taper looks consistent with the Rossby Radius size, and everything worked without further problems.
So, yes, this can be merged from my point of view (double green check).

xylar

@milenaveneziani did a ton of work to address my concerns so thanks for the coding and testing!

vanroekel · 2022-05-19T20:23:28Z

I'm also happy with all the changes. Thanks for the great discussion and thorough testing. I'll give you my second green check!

jonbob · 2022-05-23T19:27:23Z

The automated scripts found no discrepancies in the bld files

… next (PR #4864) Add horizontal tapering function to mesoscale eddy parameterizations This PR adds a new 'RossbyRadius' option to the horizontal tapering of GM and Redi, which is based on Hallberg (2013) - https://doi.org/10.1016/j.ocemod.2013.08.007, which turns the parameterization on if the Rossby Radius is resolved by at least two cells. These flags are: * config_Redi_horizontal_taper = 'ramp' (def), 'none', 'RossbyRadius' * config_GM_horizontal_taper = 'ramp' (def), 'none', 'RossbyRadius' This PR replaces the variable `gmResolutionTaper`, which was used for both GM and Redi, to the separate variables `RediHorizontalTaper` and `gmHorizontalTaper` so that Redi and GM tapering can be chosen separately, and it is clear in the code. Logic was added to the GM and Redi schemes so that the there is a base value in the arrays for Redi Kappa and GM Kappa. The base value is then multiplied by the taper array, which has a value between 0 and 1. Arrays that are static in time are computed on init. Under the following conditions, the arrays are recomputed at every time step: * config_Redi_horizontal_taper = 'RossbyRadius' (recompute RediHorizontalTaper and RediKappa) * config_GM_horizontal_taper = 'RossbyRadius' (recompute gmHorizontalTaper) [NML] adds MPAS-Ocean namelist options [BFB]

jonbob · 2022-05-23T21:40:33Z

Passes:

ERS.ne11_oQU240.WCYCL1850NS.chrysalis_intel
SMS_D_Ld1.ne30pg2_r05_EC30to60E2r2.WCYCL1850.chrysalis_intel
PEM_Ln9.ne30pg2_EC30to60E2r2.WCYCL1850.chrysalis_intel
SMS_D_Ld3.T62_oQU120.CMPASO-IAF.chrysalis_intel

with expected NML DIFFs.

Merged to next

xylar · 2022-05-24T01:43:43Z

Thanks @jonbob. Assuming all goes well, I'm sure everyone in E3SM land will be happy to have this thread silenced (at only 155 comments).

jonbob · 2022-05-24T15:50:39Z

@darincomeau - the SMS_PS.northamericax4v1pg2_WC14to60E2r3.WCYCL1850.allactive-wcprodrrm prod test failed on all platforms due to the user_nl_mpaso settings in the testmod definition:

 config_gm_closure = 'constant'
 config_eddying_resolution_taper = 'ramp'
 config_use_redi = .true.

I assume we mostly need to replace "config_eddying_resolution_taper" with "config_Redi_horizontal_taper" and "config_GM_horizontal_taper"?

darincomeau · 2022-05-24T16:11:23Z

@jonbob yes, I agree. The config_eddying_resolution_taper namelist option is removed in this PR, so that should be replaced as you suggested.

…nto next (PR #4864) Re-merge to fix testdefs issue causing failures

jonbob · 2022-05-24T21:04:15Z

re-merged to next with testmods update

jonbob · 2022-05-25T18:04:16Z

merged to master and expected NML DIFFs blessed

darincomeau added enhancement mpas-ocean BFB PR leaves answers BFB NML Do not merge labels Mar 30, 2022

darincomeau requested review from mark-petersen and vanroekel March 30, 2022 15:38

darincomeau assigned jonbob Mar 30, 2022

milenaveneziani reviewed Apr 4, 2022

View reviewed changes

components/mpas-ocean/src/shared/mpas_ocn_gm.F Outdated Show resolved Hide resolved

mark-petersen mentioned this pull request Apr 4, 2022

Add ocean option for Redi kappa to be set equal to GM (AMOC PR 3/6) #4867

Closed

mark-petersen mentioned this pull request Apr 4, 2022

limit Visbeck eddy length to dcEdge (AMOC PR 4/6) #4868

Merged

mark-petersen changed the title ~~Add GM horizontal resolution function~~ Add GM horizontal resolution function (AMOC PR 2/6) Apr 5, 2022

vanroekel approved these changes Apr 5, 2022

View reviewed changes

mark-petersen approved these changes Apr 5, 2022

View reviewed changes

milenaveneziani reviewed Apr 6, 2022

View reviewed changes

components/mpas-ocean/src/shared/mpas_ocn_gm.F Outdated Show resolved Hide resolved

milenaveneziani reviewed Apr 6, 2022

View reviewed changes

components/mpas-ocean/src/shared/mpas_ocn_gm.F Outdated Show resolved Hide resolved

darincomeau force-pushed the darincomeau/ocn/gm-horizontal-resolution-function branch from c744e64 to e7f21ab Compare April 6, 2022 18:36

darincomeau requested a review from milenaveneziani April 6, 2022 18:37

darincomeau added Do not merge and removed Do not merge labels Apr 6, 2022

mark-petersen reviewed Apr 6, 2022

View reviewed changes

components/mpas-ocean/src/shared/mpas_ocn_gm.F Outdated Show resolved Hide resolved

darincomeau force-pushed the darincomeau/ocn/gm-horizontal-resolution-function branch from e7f21ab to c744e64 Compare April 6, 2022 18:58

Fix Rossby Radius taper for small c_mode1 values

7d24582

Update name of Rossby_ramp_min/max config options

63768fc

Change back order of operations for BFB

11b48ab

mark-petersen reviewed May 19, 2022

View reviewed changes

xylar approved these changes May 19, 2022

View reviewed changes

darincomeau removed the Do not merge label May 20, 2022

jonbob added the BFB PR leaves answers BFB label May 23, 2022

Update testmods to new variable names

3e29568

jonbob added a commit that referenced this pull request May 24, 2022

Re-merge branch 'darincomeau/ocn/gm-horizontal-resolution-function' i…

01d5e87

…nto next (PR #4864) Re-merge to fix testdefs issue causing failures

jonbob merged commit ba16635 into master May 25, 2022

jonbob deleted the darincomeau/ocn/gm-horizontal-resolution-function branch May 25, 2022 18:12

xylar mentioned this pull request Jul 5, 2022

Update E3SM-Project to current master MPAS-Dev/compass#416

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add horizontal tapering function to mesoscale eddy parameterizations (AMOC PR 2/6) #4864

Add horizontal tapering function to mesoscale eddy parameterizations (AMOC PR 2/6) #4864

darincomeau commented Mar 30, 2022 •

edited by jonbob

Loading

jonbob commented Mar 30, 2022

darincomeau commented Mar 30, 2022

jonbob commented Mar 30, 2022

jonbob commented Mar 31, 2022

darincomeau commented Mar 31, 2022

darincomeau commented Apr 4, 2022

mark-petersen commented Apr 4, 2022

vanroekel left a comment

mark-petersen left a comment

darincomeau commented Apr 6, 2022 •

edited

Loading

milenaveneziani commented May 18, 2022

xylar commented May 18, 2022

milenaveneziani commented May 18, 2022

xylar commented May 18, 2022

milenaveneziani commented May 18, 2022

mark-petersen May 19, 2022

milenaveneziani May 19, 2022

mark-petersen commented May 19, 2022 •

edited

Loading

mark-petersen commented May 19, 2022

mark-petersen commented May 19, 2022

milenaveneziani commented May 19, 2022 •

edited

Loading

xylar left a comment

vanroekel commented May 19, 2022

jonbob commented May 23, 2022

jonbob commented May 23, 2022

xylar commented May 24, 2022

jonbob commented May 24, 2022

darincomeau commented May 24, 2022

jonbob commented May 24, 2022

jonbob commented May 25, 2022

Add horizontal tapering function to mesoscale eddy parameterizations (AMOC PR 2/6) #4864

Add horizontal tapering function to mesoscale eddy parameterizations (AMOC PR 2/6) #4864

Conversation

darincomeau commented Mar 30, 2022 • edited by jonbob Loading

jonbob commented Mar 30, 2022

darincomeau commented Mar 30, 2022

jonbob commented Mar 30, 2022

jonbob commented Mar 31, 2022

darincomeau commented Mar 31, 2022

darincomeau commented Apr 4, 2022

mark-petersen commented Apr 4, 2022

vanroekel left a comment

Choose a reason for hiding this comment

mark-petersen left a comment

Choose a reason for hiding this comment

darincomeau commented Apr 6, 2022 • edited Loading

milenaveneziani commented May 18, 2022

xylar commented May 18, 2022

milenaveneziani commented May 18, 2022

xylar commented May 18, 2022

milenaveneziani commented May 18, 2022

mark-petersen May 19, 2022

Choose a reason for hiding this comment

milenaveneziani May 19, 2022

Choose a reason for hiding this comment

mark-petersen commented May 19, 2022 • edited Loading

mark-petersen commented May 19, 2022

mark-petersen commented May 19, 2022

milenaveneziani commented May 19, 2022 • edited Loading

xylar left a comment

Choose a reason for hiding this comment

vanroekel commented May 19, 2022

jonbob commented May 23, 2022

jonbob commented May 23, 2022

xylar commented May 24, 2022

jonbob commented May 24, 2022

darincomeau commented May 24, 2022

jonbob commented May 24, 2022

jonbob commented May 25, 2022

darincomeau commented Mar 30, 2022 •

edited by jonbob

Loading

darincomeau commented Apr 6, 2022 •

edited

Loading

mark-petersen commented May 19, 2022 •

edited

Loading

milenaveneziani commented May 19, 2022 •

edited

Loading