Enables and modifies the submesoscale eddy parameterization #5172

vanroekel · 2022-09-02T04:02:19Z

This enables the Fox-Kemper et al 2011 submesoscale eddy parameterization as the default for all configurations.

It also fixes a unit error in the submesoscale eddy code that did not influence the long control run, but was introduced in the first PR (#5099). The variable gradBuoyEddy was not the buoyancy but the density gradient. Here the name is changed to gradDensityEddy for clarity.

[NML]
[CC]

vanroekel · 2022-09-02T04:03:38Z

results are shown in #5099 also there is 800 years of simulation analysis here

https://web.lcrc.anl.gov/public/e3sm/diagnostic_output/ac.vanroekel/E3SMv2/20220715.submeso.piControl.ne30pg2_EC30to60E2r2.chrysalis/

mark-petersen · 2022-09-06T15:21:59Z

This is the metric that shows the major change due to turning on the mesoscale eddy parameterization:

mark-petersen · 2022-09-06T15:26:10Z

Adding reviewers from the AMOC working group, who were involved in these changes. The code is already in. Here we simply turn it on. Your approval here really says that you agree with setting config_submesoscale_enable = .true. on all simulations, based on the 700+ years test above (as well as all the other testing we did along the way).

mark-petersen · 2022-09-06T15:29:23Z

This PR requires that #5167 and #5171 be merged first. Updating the MOC calculation #5170 is needed to produce the same AMOC statistic as above.

xylar · 2022-09-10T04:04:28Z

components/mpas-ocean/cime_config/buildnml

+		lines.append('    <var name="normalMLEvelocity"/>')
+		lines.append('    <var name="vertMLEVelocityTop"/>')


Suggested change

lines.append(' <var name="normalMLEvelocity"/>')

lines.append(' <var name="vertMLEVelocityTop"/>')

lines.append(' <var name="normalMLEvelocity"/>')

lines.append(' <var name="vertMLEVelocityTop"/>')

Thanks for adding theses! They are needed for the offline MOC and are likely to be useful for debugging.

We need to add them to standalone as well. We want to maintain consistency:
https://github.com/E3SM-Project/E3SM/blob/master/components/mpas-ocean/src/analysis_members/Registry_time_series_stats_monthly_mean.xml#L135-L136

It seems like we probably want to enable config_submesoscale_enable = .true. in many compass tests, too, following this PR.

jonbob · 2022-09-21T15:25:09Z

@vanroekel - testing of this PR fails due to the addition of vertMLEVelocityTop to the streams file. Apparently that array is not defined anywhere in the mpas-o source?

vanroekel · 2022-09-21T19:38:10Z

@jonbob -- sorry about that -- this variable should be vertMLEBolusVelocityTop -- do you want me to push a fix?

jonbob · 2022-09-21T20:02:31Z

@vanroekel - I can take care of that, if you're jammed up. I just removed the bad arrays from the streams for testing, so it's not holding anything up

vanroekel · 2022-09-25T04:30:36Z

@jonbob I pushed the fixes I gave you in testing #5171 here, updated the comment, and PR title. Let me know if there are any issues.

components/mpas-ocean/cime_config/buildnml

components/mpas-ocean/src/analysis_members/Registry_time_series_stats_monthly_mean.xml

jonbob · 2022-09-28T18:54:53Z

@vanroekel - do you want to change config_submesoscale_Ce to 0.08 in this PR, to match the value of the long run you did?

vanroekel · 2022-09-28T19:05:16Z

yes definitely. thanks for catching that @jonbob it should be changed now

The computation is for the gradient of density but the array was called gradBuoyEddy. This commit switches it to gradDensityEddy for clarity

mark-petersen · 2022-09-29T21:58:32Z

Coordinated with Luke: rebased on master after #5171 merge. Compiled with gnu and intel, tested with nightly stand-alone suite with gnu debug and config_submesoscale_enable set to both true and false.

jonbob · 2022-09-29T22:03:31Z

@mark-petersen - I had already started merging this to next for testing

jonbob · 2022-09-29T22:26:53Z

@xylar, @lconlon, @katsmith133 - can you please review (or finish your review)?

katsmith133

Looks good to me based upon visual inspection

jonbob · 2022-09-29T22:36:00Z

@vanroekel - can you check and make sure I changed the streams output field to the one you intended?

mark-petersen · 2022-09-29T22:40:58Z

Tested this PR on Chrysalis with

./create_test SMS_D.T62_oQU120_ais20.MPAS_LISIO_TEST.chrysalis_gnu -q acme-small --walltime 30:00

and it died on run start-up with the problem in #5204. I'll see if I can figure that out tomorrow.

jonbob · 2022-09-29T23:25:47Z

@mark-petersen -- acme-small is a queue on anvil, not chrysalis. You also need to add " --project e3sm" on chrysalis. For me, a test merge passes SMS_D_Ld3.T62_oQU120.CMPASO-IAF.chrysalis_intel. I just started SMS_D.T62_oQU120_ais20.MPAS_LISIO_TEST.chrysalis_gnu and it just successfully completed as well.

xylar

Approved based on code inspection and @vanroekel's and @jonbob's testing.

vanroekel · 2022-09-30T04:49:16Z

@jonbob streams changes look good to me. Thanks!

mark-petersen

approved based on testing above.

mark-petersen · 2022-09-30T19:55:29Z

Thanks all. This is a big accomplishment. There were sufficient reviewers, so I'm removing LeAnn and Alice, who are out.

…on' into next (PR #5172) Enables and modifies the submesoscale eddy parameterization This enables the Fox-Kemper et al 2011 submesoscale eddy parameterization as the default for all configurations. It also fixes a unit error in the submesoscale eddy code that did not influence the long control run, but was introduced in the first PR (#5099). The variable gradBuoyEddy was not the buoyancy but the density gradient. Here the name is changed to gradDensityEddy for clarity. [NML] [CC]

jonbob · 2022-09-30T21:07:30Z

test merge passes:

PEM_Ln9.ne30pg2_EC30to60E2r2.WCYCL1850.chrysalis_intel
ERS_Ld5.T62_oQU120.CMPASO-NYF.chrysalis_intel
SMS_D_Ld3.T62_oQU120.CMPASO-IAF.chrysalis_intel

with expected NML and regular DIFFs

merged to next

jonbob · 2022-10-03T15:32:04Z

merged to master and expected DIFFs and NML DIFFs blessed, except:

SMS_PS.northamericax4v1pg2_WC14to60E2r3.WCYCL1850.chrysalis_intel.allactive-wcprodrrm

which did not complete but was running fine

vanroekel added mpas-ocean CC PR is climate changing labels Sep 2, 2022

vanroekel requested a review from mark-petersen September 2, 2022 04:02

vanroekel assigned jonbob Sep 2, 2022

mark-petersen requested review from lconlon, katsmith133 and alicebarthel September 6, 2022 15:26

vanroekel mentioned this pull request Sep 10, 2022

Alters MOC calculation in ocean component #5170

Merged

xylar reviewed Sep 10, 2022

View reviewed changes

jonbob mentioned this pull request Sep 13, 2022

Fixes Redi Surface Taper double counting #5171

Merged

vanroekel changed the title ~~Enables submesoscale eddy parameterization~~ Enables and modifies the submesoscale eddy parameterization Sep 25, 2022

mark-petersen reviewed Sep 26, 2022

View reviewed changes

components/mpas-ocean/cime_config/buildnml Outdated Show resolved Hide resolved

mark-petersen reviewed Sep 26, 2022

View reviewed changes

components/mpas-ocean/src/analysis_members/Registry_time_series_stats_monthly_mean.xml Outdated Show resolved Hide resolved

vanroekel added 7 commits September 29, 2022 14:11

Enables submesoscale eddy parameterization

dd22b39

Adds MLE velocity to default output

58ab068

adds MLE velocity to mpas stand alone

6f010a5

Fixes unit issue in submesoscale code

8b60b73

Changes name of gradBuoyEddy

9503ea8

The computation is for the gradient of density but the array was called gradBuoyEddy. This commit switches it to gradDensityEddy for clarity

fixes one more gradBuoyEddy

7c27a62

Fixes registry variables

f2e7dea

changes submesoscale efficiency default strength

ddd41fd

mark-petersen force-pushed the vanroekel/ocean/enable-submesoscale-eddy-parameterization branch from b1f85d7 to ddd41fd Compare September 29, 2022 21:59

mark-petersen self-requested a review September 29, 2022 22:21

katsmith133 approved these changes Sep 29, 2022

View reviewed changes

Change stream field from normalMLEBolusvelocity to normalMLEvelocity

856b062

xylar approved these changes Sep 30, 2022

View reviewed changes

mark-petersen approved these changes Sep 30, 2022

View reviewed changes

jonbob mentioned this pull request Sep 30, 2022

Test failing with "ERROR: Requested field initialCarbon not available" #5204

Closed

jonbob added the NML label Sep 30, 2022

mark-petersen removed request for alicebarthel and lconlon September 30, 2022 19:55

jonbob merged commit bcf0289 into master Oct 3, 2022

jonbob deleted the vanroekel/ocean/enable-submesoscale-eddy-parameterization branch October 3, 2022 15:17

This was referenced Oct 3, 2022

Fix barotropic thickness on edge in split explicit subcycling #5195

Merged

Fix threading issue in submesoscale code #5215

Closed

xylar mentioned this pull request Nov 19, 2022

Update E3SM-Project submodule MPAS-Dev/compass#461

Merged

32 tasks

xylar mentioned this pull request Nov 30, 2022

Ocean fails stand-alone decomp test, intel optimized #5219

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Enables and modifies the submesoscale eddy parameterization #5172

Enables and modifies the submesoscale eddy parameterization #5172

vanroekel commented Sep 2, 2022 •

edited by jonbob

Loading

vanroekel commented Sep 2, 2022

mark-petersen commented Sep 6, 2022

mark-petersen commented Sep 6, 2022

mark-petersen commented Sep 6, 2022

xylar Sep 10, 2022

xylar Sep 10, 2022

xylar Sep 10, 2022 •

edited

Loading

vanroekel Sep 10, 2022

jonbob commented Sep 21, 2022

vanroekel commented Sep 21, 2022

jonbob commented Sep 21, 2022

vanroekel commented Sep 25, 2022

jonbob commented Sep 28, 2022

vanroekel commented Sep 28, 2022

mark-petersen commented Sep 29, 2022

jonbob commented Sep 29, 2022 •

edited

Loading

jonbob commented Sep 29, 2022

katsmith133 left a comment

jonbob commented Sep 29, 2022

mark-petersen commented Sep 29, 2022

jonbob commented Sep 29, 2022

xylar left a comment

vanroekel commented Sep 30, 2022

mark-petersen left a comment

mark-petersen commented Sep 30, 2022

jonbob commented Sep 30, 2022

jonbob commented Oct 3, 2022

		lines.append(' <var name="normalMLEvelocity"/>')
		lines.append(' <var name="vertMLEVelocityTop"/>')

Enables and modifies the submesoscale eddy parameterization #5172

Enables and modifies the submesoscale eddy parameterization #5172

Conversation

vanroekel commented Sep 2, 2022 • edited by jonbob Loading

vanroekel commented Sep 2, 2022

mark-petersen commented Sep 6, 2022

mark-petersen commented Sep 6, 2022

mark-petersen commented Sep 6, 2022

xylar Sep 10, 2022

Choose a reason for hiding this comment

xylar Sep 10, 2022

Choose a reason for hiding this comment

xylar Sep 10, 2022 • edited Loading

Choose a reason for hiding this comment

vanroekel Sep 10, 2022

Choose a reason for hiding this comment

jonbob commented Sep 21, 2022

vanroekel commented Sep 21, 2022

jonbob commented Sep 21, 2022

vanroekel commented Sep 25, 2022

jonbob commented Sep 28, 2022

vanroekel commented Sep 28, 2022

mark-petersen commented Sep 29, 2022

jonbob commented Sep 29, 2022 • edited Loading

jonbob commented Sep 29, 2022

katsmith133 left a comment

Choose a reason for hiding this comment

jonbob commented Sep 29, 2022

mark-petersen commented Sep 29, 2022

jonbob commented Sep 29, 2022

xylar left a comment

Choose a reason for hiding this comment

vanroekel commented Sep 30, 2022

mark-petersen left a comment

Choose a reason for hiding this comment

mark-petersen commented Sep 30, 2022

jonbob commented Sep 30, 2022

jonbob commented Oct 3, 2022

vanroekel commented Sep 2, 2022 •

edited by jonbob

Loading

xylar Sep 10, 2022 •

edited

Loading

jonbob commented Sep 29, 2022 •

edited

Loading