New params file with changes for Meier roughness, MIMICS, and SNICAR, and changes to leafcn and k*_nonmyc #2258

slevis-lmwg · 2023-11-15T23:47:10Z

Description of changes

Start using existing new params file for Meier roughness:
/glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/paramdata/ctsm51_params.RMz0.c231011.nc
and include bug-fix Meier roughness changes answers on processor count #2219
Update forcing heights per Forcing heights need to be updated when roughness length/displacement height is updated #2071.
Update params file for MIMICS per MIMICS related parameters missing name and units in paramfile #1845.
Make leafcn for pfts 15 and 16 the same per leafcn for pfts 15 and 16 have diverged, should probably be the same #2184.
Switch the values of params kc_nonmyc and kn_nonmyc per Switched parameters for non-mycorrhizal N uptake costs in FUN #2120.
Move SNICAR parameters to params file per Move SNICAR parameter to parameter file #2247.

Specific notes

Contributors other than yourself, if any:
@ekluzek @olyson @wwieder @ecaas @cenlinhe others?

CTSM Issues Fixed (include github issue #):
Fixes #2219
Fixes #2071
Fixes #1845
Fixes #2184
Fixes #2120
Fixes #2247

Are answers expected to change (and if so in what way)? YES.

In step (1) answers change due to the new values for zlnd and zsno in the RMz0 params file.
In step (2) answers will change due to a change in how we calculate forcing heights in the default ZengWang2007 method.
In step (3) answers change within roundoff for MIMICS cases due to changed order of operations.
In step (4) answers will change because we will return leafcn for pft 15 to the same value as pft 16, i.e. 25.
In step (5) answers change because we switch two parameters back to the way that they should have been.
In step (6) answers may not change, unless due to a change in order of operations.

Any User Interface Changes (namelist or namelist defaults changes)?
New params file.

Testing performed, if any:
The answer-changes discussion above pretty much covers the current extent of testing.
After I merge this PR, I will run an Answer-changing simulation to compare with the model before this merge.

slevis-lmwg · 2023-11-15T23:53:01Z

@olyson I'm requesting a review from you because you wrote #2071. Could you check my work in this PR as pertains to #2071?

slevis-lmwg · 2023-11-16T01:01:36Z

I have pushed steps (1) and (2) to the PR.
Locally in my branch I am up to step (5). The cheyenne test-suite returns no fails when issued this command:
./cs.status.fails | grep -v '152: DIFF' | grep -v 'EXPECTED FAILURE' | grep -v NLCOMP

olyson · 2023-11-16T15:45:57Z

src/biogeophys/BareGroundFluxesMod.F90

Should we bother lining up "observational" with the rest of the variable descriptions (off by one space).

Yeah, this is the vertical alignment question. Which I am confused about our current status on. We used to make sure things aligned vertically. But, recently I had understood we were going to go away from that. But, then we have indications that we aren't. So we need to nail this down and all agree on what we are doing.

olyson

One comment, otherwise looks good, thanks.

slevis-lmwg · 2023-11-18T01:49:36Z

Test-suites:

izumi WAITING for now
cheyenne RERUN A FEW TESTS making new baselines for them:
Meier2022_surf_rough
ciso_cwd_hr
OR if changing default z0param_method in this PR, then rerun all tests, making new baselines for all.

slevis-lmwg · 2023-11-18T03:05:44Z

@wwieder @ekluzek @olyson:
Is this the PR when we also change z0param_method from ZengWang2007 to Meier2022?
This far I have not done that.

wwieder · 2023-11-18T12:52:17Z

Yes, I think so, Sam

slevis-lmwg · 2023-11-18T21:54:24Z

I looked through the Meier tests in the cheyenne test-suite and found no exact Clm51 equivalents, so I will keep them without the reference to Meier2022_surf_rough, because the latter is now default for clm51.

One test failed on izumi:
SMS_Ld5_Mmpi-serial.1x1_brazil.IHistClm50BgcQianRs.izumi_gnu.clm-mimics
It passed when I changed it to
SMS_Ld5_Mmpi-serial.1x1_brazil.IHistClm51Bgc.izumi_gnu.clm-mimics
I confirmed with Will that it's fine to assume that MIMICS works with ctsm51 and not clm45 or clm50.
I generated a baseline for this new test.

Four tests failed on cheyenne:

ERP_D_Ld10_P36x2.f10_f10_mg37.IHistClm51BgcCrop.cheyenne_intel.clm-ciso_decStart
ERS_Ly5_P144x1.f10_f10_mg37.IHistClm51BgcCrop.cheyenne_intel.clm-cropMonthOutput
LCISO_Lm13.f10_f10_mg37.IHistClm51BgcCrop.cheyenne_intel.clm-ciso_monthly
SMS_D_Ln6.f09_g17.I1850Clm50BgcNoAnthro.cheyenne_intel.clm-Meier2022_surf_rough

The first three I fixed with with a code change that permits htop < 1e-10 for istcrop.
I generated baselines for these three.

The fourth should have been a clm51 test, but I get
ERROR: Invalid compset name, I1850Clm51BgcNoAnthro so I will delete this test because it is redundant as a clm50 test.

This expected failure should be a clm51 test, so I will rerun and generate a baseline.
ERP_P72x2_Ld30.f45_f45_mg37.I2000Clm50FatesRs.cheyenne_intel.clm-mimicsFatesCold
This test was failing and still fails with
Fire emission can NOT be on when FATES is also on.
This could be fixed by pointing this testmod to /basic instead of /default; however, the change would affect all mimics tests, so we're in a bind at the moment.

I'm rerunning the four clm51 tests that had Meier testmods, now without those testmods.
I'm generating baselines for these four.

I hope I got everything...

ekluzek · 2023-11-18T23:24:28Z

Yes, this is correct the original card in upcoming tags had this in it:

✅ (needs cherry-pick) Turn Meier roughness work on for clm5_1 physics; include bug fix Meier roughness changes answers on processor count #2219 (cherry-pick whatever commit Erik pushed to CESM_dev branch)

The commit in question is:

a75a4ba

To use the commit you need to use cherry-pick so it only brings in that one change, rather than all the changes on CESM3_dev. Or you can copy the changes in by hand. cherry-pick keeps the history, that's it's advantage...

…stmod dirs

ekluzek

I have a question and then an important change to make. Since, it's so simple I'm approving it as it is so you can immediately move forward after the one change.

Nice, to see this coming in.

ekluzek · 2023-11-20T19:25:59Z

cime_config/testdefs/testmods_dirs/clm/Meier2022_surf_rough/user_nl_clm

@@ -1,4 +1,2 @@
 z0param_method = 'Meier2022'


Probably should remove this test mod since the clm4_5 and clm5_0 tests already are doing Meier off, and clm5_1 with Meier on. So both options are covered. So there's no reason for a test mod that does one or the other...

bld/namelist_files/namelist_defaults_ctsm.xml

ekluzek · 2023-11-20T20:42:48Z

@slevis-lmwg since you've run testing and the changes are still relatively modest I think we should go forward with this. But, it does look like there are a mix of changes that could be done as bit-for-bit and others that truly are answer changing. It's good to make the answer changes as small as possible, and do the bit-for-bit parts separate, just in case there's a mistake in the bit-for-bit part. As then it'll show up as an unexpected answer change. That kind of thing is hidden when there's a mix of both.

For example, the move of fresh_snw_rds_max to the param file could be done as a bit-for-bit part. If it changes answers you then know there's a mistake in there. But, it's hidden when it comes in with answer changes.

It does look like you marked clm4_5 and clm5_0 as changing answers. Is that expected and is it possible to make them identical to dev153? If they can be made identical that would ensure that the bit-for-bit parts are bit-for-bit as expected.

… test

slevis-lmwg · 2023-11-21T00:22:53Z

cime_config/testdefs/testlist_clm.xml

@@ -17,15 +17,6 @@
      <option name="wallclock">00:20:00</option>
    </options>
  </test>
-  <test name="SMS_D_Ln6" grid="f09_g17" compset="I1850Clm50BgcNoAnthro" testmods="clm/Meier2022_surf_rough">


Removing the Meier testmods made this test redundant as a clm50 test, so I removed it.
(I did first try changing to a clm51 test and got Invalid compset name, I1850Clm51BgcNoAnthro.)

slevis-lmwg · 2023-11-21T01:13:51Z

Due to updating some tests and then generating baselines for them after running the test-suites, I am rerunning the test-suites and comparing against ctsm5.1.dev154.

@ekluzek when I'm satisfied with the results, I will look into the feasibility of your suggestion of separating the parts of this PR that should be bfb.

slevis-lmwg · 2023-11-21T20:56:42Z

@slevis-lmwg since you've run testing and the changes are still relatively modest I think we should go forward with this. But, it does look like there are a mix of changes that could be done as bit-for-bit and others that truly are answer changing. It's good to make the answer changes as small as possible, and do the bit-for-bit parts separate, just in case there's a mistake in the bit-for-bit part. As then it'll show up as an unexpected answer change. That kind of thing is hidden when there's a mix of both.

For example, the move of fresh_snw_rds_max to the param file could be done as a bit-for-bit part. If it changes answers you then know there's a mistake in there. But, it's hidden when it comes in with answer changes.

It does look like you marked clm4_5 and clm5_0 as changing answers. Is that expected and is it possible to make them identical to dev153? If they can be made identical that would ensure that the bit-for-bit parts are bit-for-bit as expected.

Here are my thoughts in the same order that this PR's commits went in:
f9978db and b8c71fa: These two change answers for Meier2022 (clm51) but the latter commit also changes zlnd and zsno, which changes all three CLMs.
626f520: Takes out if (z0param_method == 'Meier2022'), so changes answers for all three CLMs
319d194: Changes answers for mimics and, therefore clm51 only (order of ops, so roundoff)
2ee6943: Changes clm51 params file, so affects clm51 only (expect more than roundoff)
f185a31: bfb
6dc1966: This git merge escomp/master probably does change answers from previous commit
29ca5ad and 71e174f: Puts snicar params on the params files for all three CLMs; I don't know whether we'd expect bfb or roundoff...

@ekluzek I see in your post that you're suggesting letting this through without additional testing at this point. I do think that additional testing would be convoluted and time consuming but doable. How about I wait for your confirmation (I could also meet if you prefer) by beginning of tomorrow, before moving ahead with the tag?

slevis-lmwg · 2023-11-22T00:26:28Z

From the above, I started a relatively easy test on izumi:
I backed up to 6dc1966 (before the snicar mods), then I put back the changes of the commit right after snicar (71e174f). Comparing to dev154 (the new tag's baseline that I've created), I hope to find at most roundoff diffs.

Update: The above passed bfb against dev154 on izumi (12 gnu, 18 intel, and 32 nag tests).
Other changes are quite confined, so I will not pursue further testing, other than the Answer Changing Tags simulation after making the tag tomorrow.

slevis-lmwg · 2023-11-22T19:48:59Z

doc/ChangeLog

+ New clm5_1 params file with new parameters and with modified values of existing parameters.
+ New clm5_0 and clm4_5 params files with new parameters for SNICAR.
+ ./rimport on the new params files fails with "No space left on device" but the 4 files are safe here:
+ /glade/u/home/slevis/paramfiles/*_params.c231117.nc


Final comment before I merge:
./rimport on the new params files fails with "No space left on device"
The files are safe in my home directory: /glade/u/home/slevis/paramfiles.

Update I have now completed this pending item:

./rimport -file lnd/clm2/paramdata/clm45_params.c231117.nc ./rimport -file lnd/clm2/paramdata/clm50_params.c231117.nc ./rimport -file lnd/clm2/paramdata/ctsm51_params.c231117.nc ./rimport -file lnd/clm2/paramdata/ctsm51_ciso_cwd_hr_params.c231117.nc

New params files for Meier roughness, MIMICS, SNICAR, and with changes to leafcn and k*_nonmyc 1) Start using existing new params file for Meier roughness: /glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/paramdata/ctsm51_params.RMz0.c231011.nc and include bug-fix ESCOMP#2219 2) Update forcing heights per ESCOMP#2071. 3) Update params file for MIMICS per ESCOMP#1845. 4) Make leafcn for pfts 15 and 16 the same per ESCOMP#2184. 5) Switch the values of params kc_nonmyc and kn_nonmyc per ESCOMP#2120. 6) Move SNICAR parameters to params file per ESCOMP#2247. Changes answers. Details in PR ESCOMP#2258 and in the ChangeLog.

ekluzek and others added 3 commits November 15, 2023 15:41

Changes to fix ESCOMP#2219

f9978db

Change namelist_defaults_ctsm.xml to use ctsm51_params.RMz0.c231011.nc

b8c71fa

Update forcing heights per ESCOMP#2071

626f520

slevis-lmwg self-assigned this Nov 15, 2023

slevis-lmwg added priority: high High priority to fix/merge soon, e.g., because it is a problem in important configurations PR status: work in progress next this should get some attention in the next week or two. Normally each Thursday SE meeting. labels Nov 15, 2023

slevis-lmwg requested a review from olyson November 15, 2023 23:51

Update code and params file with MIMICS changes per ESCOMP#1845

319d194

olyson reviewed Nov 16, 2023

View reviewed changes

olyson approved these changes Nov 16, 2023

View reviewed changes

slevis-lmwg added 4 commits November 17, 2023 12:13

Change namelist_defaults_ctsm.xml to use ctsm51_params.c231115b.nc

2ee6943

Revised indentation based on code review

f185a31

Merge remote-tracking branch 'escomp/master' into new_params_file

6dc1966

Move SNICAR params to params files per ESCOMP#2247

29ca5ad

slevis-lmwg added 2 commits November 20, 2023 11:09

Upd. paramfile / z0param_method in namelist_defaults_ctsm and two /te…

71e174f

…stmod dirs

Draft ChangeLog/Sum

13ef0cd

ekluzek approved these changes Nov 20, 2023

View reviewed changes

slevis-lmwg added 2 commits November 20, 2023 13:55

Remove Meier2022 testmods directory; update testlist accordingly

4792efc

Small revision based on code review

6b231bf

slevis-lmwg added 2 commits November 20, 2023 16:07

For Meier IHist...Crop tests to pass, permit htop < 1e-10 for istcrop

939bea7

Update testlists: mimics tests to clm51 and remove obsolete Meier2022…

ae8d808

… test

slevis-lmwg commented Nov 21, 2023

View reviewed changes

Update ChangeLog

5ed438f

slevis-lmwg added 2 commits November 21, 2023 13:11

Update ExpectedTestFails and testlist

fc631f4

Update ChangeLog/ChangeSum

e8aa4eb

Update ChangeLog/ChangeSum

d578587

slevis-lmwg commented Nov 22, 2023

View reviewed changes

slevis-lmwg marked this pull request as ready for review November 22, 2023 19:50

slevis-lmwg merged commit bfbd39f into ESCOMP:master Nov 22, 2023

slevis-lmwg deleted the new_params_file branch November 22, 2023 19:51

slevis-lmwg mentioned this pull request Dec 7, 2023

Test fails with Meier roughness on and with threading and carbon isotopes and transient case #2238

Closed

samsrabin removed the next this should get some attention in the next week or two. Normally each Thursday SE meeting. label Mar 21, 2024

samsrabin added bug something is working incorrectly science Enhancement to or bug impacting science labels Aug 8, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

New params file with changes for Meier roughness, MIMICS, and SNICAR, and changes to leafcn and k*_nonmyc #2258

New params file with changes for Meier roughness, MIMICS, and SNICAR, and changes to leafcn and k*_nonmyc #2258

slevis-lmwg commented Nov 15, 2023 •

edited

Loading

slevis-lmwg commented Nov 15, 2023

slevis-lmwg commented Nov 16, 2023 •

edited

Loading

olyson Nov 16, 2023

ekluzek Nov 20, 2023

olyson left a comment

slevis-lmwg commented Nov 18, 2023 •

edited

Loading

slevis-lmwg commented Nov 18, 2023

wwieder commented Nov 18, 2023

slevis-lmwg commented Nov 18, 2023 •

edited

Loading

ekluzek commented Nov 18, 2023

ekluzek left a comment

ekluzek Nov 20, 2023

slevis-lmwg Nov 20, 2023

ekluzek commented Nov 20, 2023

slevis-lmwg Nov 21, 2023

slevis-lmwg commented Nov 21, 2023

slevis-lmwg commented Nov 21, 2023 •

edited

Loading

slevis-lmwg commented Nov 22, 2023 •

edited

Loading

slevis-lmwg Nov 22, 2023

slevis-lmwg Nov 29, 2023

New params file with changes for Meier roughness, MIMICS, and SNICAR, and changes to leafcn and k*_nonmyc #2258

New params file with changes for Meier roughness, MIMICS, and SNICAR, and changes to leafcn and k*_nonmyc #2258

Conversation

slevis-lmwg commented Nov 15, 2023 • edited Loading

Description of changes

Specific notes

slevis-lmwg commented Nov 15, 2023

slevis-lmwg commented Nov 16, 2023 • edited Loading

olyson Nov 16, 2023

Choose a reason for hiding this comment

ekluzek Nov 20, 2023

Choose a reason for hiding this comment

olyson left a comment

Choose a reason for hiding this comment

slevis-lmwg commented Nov 18, 2023 • edited Loading

slevis-lmwg commented Nov 18, 2023

wwieder commented Nov 18, 2023

slevis-lmwg commented Nov 18, 2023 • edited Loading

ekluzek commented Nov 18, 2023

ekluzek left a comment

Choose a reason for hiding this comment

ekluzek Nov 20, 2023

Choose a reason for hiding this comment

slevis-lmwg Nov 20, 2023

Choose a reason for hiding this comment

ekluzek commented Nov 20, 2023

slevis-lmwg Nov 21, 2023

Choose a reason for hiding this comment

slevis-lmwg commented Nov 21, 2023

slevis-lmwg commented Nov 21, 2023 • edited Loading

slevis-lmwg commented Nov 22, 2023 • edited Loading

slevis-lmwg Nov 22, 2023

Choose a reason for hiding this comment

slevis-lmwg Nov 29, 2023

Choose a reason for hiding this comment

slevis-lmwg commented Nov 15, 2023 •

edited

Loading

slevis-lmwg commented Nov 16, 2023 •

edited

Loading

slevis-lmwg commented Nov 18, 2023 •

edited

Loading

slevis-lmwg commented Nov 18, 2023 •

edited

Loading

slevis-lmwg commented Nov 21, 2023 •

edited

Loading

slevis-lmwg commented Nov 22, 2023 •

edited

Loading