Fixes conservation problem for non-water species #2883

beharrop · 2019-04-25T20:46:59Z

All non-water tracers are kept as dry mixing ratios, but CLUBB and the
gravity wave parameterizations assume wet mixing ratios, and break
conservation when they are called. This fix converts dry mixing ratios
to wet mixing ratios prior to calling the routines that need wet mixing
ratios. The code fixes follow similar fixes from CESM2, with additional
modifications for E3SM-specific components.

Fixes #2704

[Non-BFB] - Non Bit-For-Bit
[CC] - Climate Changing

All non-water tracers are kept as dry mixing ratios, but CLUBB and the gravity wave parameterizations assume wet mixing ratios, and break conservation when they are called. This fix converts dry mixing ratios to wet mixing ratios prior to calling the routines that need wet mixing ratios. The code fixes follow similar fixes from CESM2, with additional modifications for E3SM-specific components. Fixes #2704 [Non-BFB] - Non Bit-For-Bit [CC] - Climate Changing

beharrop · 2019-04-25T20:50:18Z

As noted in the commit message, this bug fix produces changes to the climate. Two 5 year F case simulations were run (with and without the fix) and the RESTOM difference is -0.41 W/m2. The full e3sm_diags comparison of the two runs is here: https://portal.nersc.gov/project/acme/beharrop/e3sm_bugfix_test_1year/viewer/

PeterCaldwell · 2019-04-25T21:00:03Z

I'm surprised the RESTOM difference is so big. My recollection is that this change had less impact on CESM. Do you recall if this is correct? Did you expect to see such large changes? Also, a quick gander of the diags shows that the change is from +0.2 W/m2 in the SWCF and +0.2 W/m2 change in LWclr. Also, I couldn't really see any pattern in terms of which regions had larger or smaller TOA fluxes. Can you articulate any sort of logic for the changes between simulations? Perhaps we need to run for longer than 5 yrs (which is in itself weird).

PeterCaldwell · 2019-04-25T21:00:56Z

Oh, and I forgot to say - great work! Thanks for taking this important task on!

beharrop · 2019-04-25T21:27:04Z

@PeterCaldwell, I am not sure how large the RESTOM difference was for CESM2. I agree that there isn't much of a discernible pattern in the fields (maybe something from dust). I suppose there could be a smaller, more global change that doesn't show up with this colorbar, but I think that's clutching at straws. I don't have a good suggestion for what is driving these changes. I can run out several more years to see if the signals persist.

swrneale · 2019-04-26T14:42:51Z

The difference CESM2 had was not of this magnitude as we decided to move forward with CMIP experiments after fixing the bug.

beharrop · 2019-05-03T22:32:50Z

@PeterCaldwell, the 15 year runs produced a RESTOM difference of -0.009 W/m2 (https://portal.nersc.gov/project/acme/beharrop/e3sm_bugfix_test_15year/viewer/). I went back and looked a bit more at how the RESTOM could be so different for a 5 year run and realized I previously linked a 1 year climo instead. The 5 year RESTOM difference is -0.117 W/m2. I suppose we can strike the "climate changing" tag from the commit message. This also puts our response to the bug fix in line with what CESM2 saw.

PeterCaldwell · 2019-05-03T22:37:05Z

Great! Thanks for extending the run...

rljacob · 2019-05-07T04:29:07Z

So is this ready to merge?

beharrop · 2019-05-14T17:19:36Z

This is ready to merge from my view. We are also going to need these changes to move to maint-1.1. Do I need to issue an additional pull request for that?

Also, we are working on a related issue for the co2 tracers pointed out to us by Keith Lindsay at NCAR (this bug fix does not preserve the near-constant profile of co2). Do you want me to try to combine those changes into this pull request or leave them for a separate PR later?

rljacob · 2019-05-14T17:24:06Z

Separate PRs for each of those.

beharrop · 2019-05-14T17:33:15Z

Sounds good. The only other thing that needs discussion is whether this is really climate-changing or not. The RESTOM difference for 15 years was -0.009 W/m2. Is there a cutoff for what we consider to be a change in climate for this label?

rljacob · 2019-05-30T17:18:25Z

@golaz do you have an opinion on the above question?

beharrop · 2019-06-06T23:09:29Z

@polunma or @PeterCaldwell, do either of you object to calling this change non-BFB instead of CC?

@wlin7, do you have any objections to getting this merged?

wlin7 · 2019-06-11T04:18:40Z

@beharrop , I think it is ready to merge. Hard to say this PR is a CC. @rljacob , please remind me, procedure-wise, any difference in handling of a regular non-BFB vs. a CC?

rljacob · 2019-06-13T20:58:32Z

@wlin7 there's no difference. It should go in by itself since so many tests will have answers change.

wlin7 · 2019-06-13T21:11:05Z

Thanks @rljacob . When is a good time to merge this non-BFB PR?

rljacob · 2019-06-13T21:28:49Z

Nothing's been merged to next today so now would be a good time. Let the other integrators know by sending a message to the mailing list.

wlin7 · 2019-06-13T23:00:09Z

Hi @rljacob , need help on this one.
Here it says on conflict for the PR. But on merging to next, it reports

CONFLICT (file/directory): There is a directory with name components/cam/src/physics/rrtmgp/external in origin/beharrop/atm/tracer_cons_bugfix. Adding components/cam/src/physics/rrtmgp/external as components/cam/src/physics/rrtmgp/external~HEAD

How to handle this? Thanks.

oksanaguba · 2019-06-13T23:03:26Z

If I could comment on this ~HEAD message: it happened to me for one of PRs; after merge I compared folder in next with merged PR with master folder and there were no diffs. then I did git commit to mark resolution of the conflict. I also ran git stat or something like that to make sure I am not committing anything from that folder.

wlin7 · 2019-06-13T23:21:46Z

Thanks for the comments, @oksanaguba. I did similar comparison after the merge. No diffs between the rrtmgp/external in the next after the merge with that in the master. There is no .../rrtmgp/external~HEAD exists as the CONFLICT message indicated. The content under that directory is from an external module and is not part of this PR. I feel it is safe not to do anything about it.

You said "did git commit to mark resolution of the conflict". Do you think in this case I can also just manually issue "git commit" to complete the merge process, essentially ignoring this CONFLICT complaint?

oksanaguba · 2019-06-13T23:25:38Z

I think without git commit your merge is not complete? you can always toss commit and start over. it is my best recollection that I did git commit and, yes, in my case there was no ~HEAD folder. I assume it just means the folder in question was taken from your HEAD, that is, from next. I found my message for slack i merged 2 PRs to master; like yesterday, merge of one of them, from a fork, created a conflict message for one of submodules. it was something about renaming except that there was no renaming in merge. i double checked with git show --stat that i was not going to commit anything into submodules.

rljacob · 2019-06-13T23:28:15Z

When you updated next, did you also do "git submodule update --init" before you did your merge?

wlin7 · 2019-06-13T23:34:22Z

@rljacob , I just switched to next from another branch to do the merge. Didn't do "git submodule update --init" this time. But I don't remember if I had to do it previously when testing certain PR on next.

wlin7 · 2019-06-13T23:45:20Z

It wouldn't let me run "git commit" at this point.

#git commit

U       components/cam/src/physics/rrtmgp/external
error: Committing is not possible because you have unmerged files.
hint: Fix them up in the work tree, and then use 'git add/rm <file>'
hint: as appropriate to mark resolution and make a commit.

Can I just remove the whole components/cam/src/physics/rrtmgp/external and do a git rm?

All non-water tracers are kept as dry mixing ratios, but CLUBB and the gravity wave parameterizations assume wet mixing ratios, and break conservation when they are called. This fix converts dry mixing ratios to wet mixing ratios prior to calling the routines that need wet mixing ratios. The code fixes follow similar fixes from CESM2, with additional modifications for E3SM-specific components. Fixes #2704 [Non-BFB] - Non Bit-For-Bit [CC] - Climate Changing

wlin7 · 2019-06-14T02:46:20Z

Merged to next.

I did the following to resolve the issue mentioned above ( ..external~HEAD directory

1. Abort the merge using "git merge --abort"
2. rm -r components/cam/src/physics/rrtmgp/external (not empty,  from previous git subbmodule update --init)
3. merge the PR as usual

wlin7 · 2019-06-14T14:04:46Z

Requests submitted to bless 36 DIFFs on sandiatoss3 and 7 on melvin.

All non-water tracers are kept as dry mixing ratios, but CLUBB and the gravity wave parameterizations assume wet mixing ratios, and break conservation when they are called. This fix converts dry mixing ratios to wet mixing ratios prior to calling the routines that need wet mixing ratios. The code fixes follow similar fixes from CESM2, with additional modifications for E3SM-specific components. Fixes #2704 [Non-BFB] - Non Bit-For-Bit [CC] - Climate Changing

wlin7 · 2019-06-14T15:18:26Z

Merged to master.

Modifications to preserve the vertical shape of CO2 tracers Follows suggestions by Keith Lindsay for modifying the fix to conserve non-water tracer species (#2883 ) so that the nearly constant profile of CO2 is preserved. Non-BFB when CO2 tracers are active. [BFB] * beharrop/atm/co2_profile_fix: Modifications to preserve the vertical shape of CO2 tracers

Modifications to preserve the vertical shape of CO2 tracers Follows suggestions by Keith Lindsay for modifying the fix to conserve non-water tracer species (#2883 ) so that the nearly constant profile of CO2 is preserved. [NBFB] * beharrop/atm/co2_profile_fix: Modifications to preserve the vertical shape of CO2 tracers

AaronDonahue mentioned this pull request Apr 25, 2019

Conservation problems for non-water species #2704

Closed

rljacob assigned singhbalwinder Apr 26, 2019

rljacob added Atmosphere bug fix PR labels Apr 26, 2019

rljacob assigned wlin7 and unassigned singhbalwinder Apr 26, 2019

rljacob added the CC PR is climate changing label May 9, 2019

wlin7 merged commit ee0852c into master Jun 14, 2019

beharrop mentioned this pull request Jun 18, 2019

Modifications to preserve the vertical shape of CO2 tracers #3003

Merged

kaizhangpnl added the non-BFB PR makes roundoff changes to answers. label Sep 2, 2019

beharrop mentioned this pull request Apr 26, 2021

Make field manager use wet mixing ratios, physics procs use dry mixing ratios as needed E3SM-Project/scream#1008

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixes conservation problem for non-water species #2883

Fixes conservation problem for non-water species #2883

beharrop commented Apr 25, 2019

beharrop commented Apr 25, 2019

PeterCaldwell commented Apr 25, 2019

PeterCaldwell commented Apr 25, 2019

beharrop commented Apr 25, 2019

swrneale commented Apr 26, 2019

beharrop commented May 3, 2019

PeterCaldwell commented May 3, 2019

rljacob commented May 7, 2019

beharrop commented May 14, 2019

rljacob commented May 14, 2019

beharrop commented May 14, 2019

rljacob commented May 30, 2019

beharrop commented Jun 6, 2019

wlin7 commented Jun 11, 2019

rljacob commented Jun 13, 2019

wlin7 commented Jun 13, 2019

rljacob commented Jun 13, 2019

wlin7 commented Jun 13, 2019

oksanaguba commented Jun 13, 2019 •

edited

Loading

wlin7 commented Jun 13, 2019

oksanaguba commented Jun 13, 2019

rljacob commented Jun 13, 2019

wlin7 commented Jun 13, 2019

wlin7 commented Jun 13, 2019

wlin7 commented Jun 14, 2019 •

edited

Loading

wlin7 commented Jun 14, 2019

wlin7 commented Jun 14, 2019

Fixes conservation problem for non-water species #2883

Fixes conservation problem for non-water species #2883

Conversation

beharrop commented Apr 25, 2019

beharrop commented Apr 25, 2019

PeterCaldwell commented Apr 25, 2019

PeterCaldwell commented Apr 25, 2019

beharrop commented Apr 25, 2019

swrneale commented Apr 26, 2019

beharrop commented May 3, 2019

PeterCaldwell commented May 3, 2019

rljacob commented May 7, 2019

beharrop commented May 14, 2019

rljacob commented May 14, 2019

beharrop commented May 14, 2019

rljacob commented May 30, 2019

beharrop commented Jun 6, 2019

wlin7 commented Jun 11, 2019

rljacob commented Jun 13, 2019

wlin7 commented Jun 13, 2019

rljacob commented Jun 13, 2019

wlin7 commented Jun 13, 2019

oksanaguba commented Jun 13, 2019 • edited Loading

wlin7 commented Jun 13, 2019

oksanaguba commented Jun 13, 2019

rljacob commented Jun 13, 2019

wlin7 commented Jun 13, 2019

wlin7 commented Jun 13, 2019

wlin7 commented Jun 14, 2019 • edited Loading

wlin7 commented Jun 14, 2019

wlin7 commented Jun 14, 2019

oksanaguba commented Jun 13, 2019 •

edited

Loading

wlin7 commented Jun 14, 2019 •

edited

Loading