fix area scaling to correct RTM river flux to MOM6

[swensosc]

fix how area is applied to input runoff for rtm

Fixes #50

[slevis-lmwg]

Test-suite to generate baselines on derecho:
./run_sys_tests -s rtm -c rtm1_0_79_ctsm5.2.0 -g rtm1_0_79_ctsm5.2.007
I see NLCOMP failures, which do not seem surprising to me.

I have checked out this branch and submitted the test-suite against the baselines:
./run_sys_tests -s rtm -c rtm1_0_79_ctsm5.2.007 -g rtm1_0_80_ctsm5.2.007

The naming convention is correct, it hasn't changed for the RTM as it has for MOSART.

Also, I tried testing on izumi and found out that we have no rtm tests on izumi.

[slevis-lmwg]

The only failures are DIFF against the baseline, which seems expected, right?

I see diffs in QCHANR, QCHANR_ICE, QCHOCNR, QCHOCNR_ICE
and
lndExp_Flrr_volr, lndExp_Flrr_volrmch, rofImp_Flrr_volr, rofImp_Flrr_volrmch, rofImp_Forr_rofi, rofImp_Forr_rofl, rofExp_Flrl_irrig

Results here:
/glade/work/slevis/git_externals/rtm_fix/tests_0618-182038de

[swensosc]

this will change answers relative to the current tag, but the point of the pr is to get back to the original mct version. Is there a way to test vs a 'correct' mct version using the same version of clm/ctsm?

…

On Tue, Jun 18, 2024 at 6:53 PM Samuel Levis ***@***.***> wrote: The only failures are DIFF against the baseline, which seems expected, right? I see diffs in QCHANR, QCHANR_ICE, QCHOCNR, QCHOCNR_ICE and lndExp_Flrr_volr, lndExp_Flrr_volrmch, rofImp_Flrr_volr, rofImp_Flrr_volrmch, rofImp_Forr_rofi, rofImp_Forr_rofl, rofExp_Flrl_irrig Results here: /glade/work/slevis/git_externals/rtm_fix/tests_0618-182038de — Reply to this email directly, view it on GitHub <#51 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGRN57BHXNYE7W4F7BGOES3ZIDJCNAVCNFSM6AAAAABI3IERQ6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNZXGMYTSOJVG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[slevis-lmwg]

from meeting with @ekluzek

FAIL Try making an mct baseline with ctsm5.2.005 (see posts below about the failure):
./run_sys_tests -s rtm -c rtm1_0_79_ctsm5.2.0 -g rtm1_0_79_mct_ctsm5.2.005 --extra-create-test-args " --driver mct"
and then comparing to it:
./run_sys_tests -s rtm -c rtm1_0_79_mct_ctsm5.2.005 --skip-generate

• @slevis-lmwg compare the code changes against what was in rtm mct (so an older version of rtm)
  In /glade/work/slevis/git_ctsm_tags/ctsm5.1.dev154/components/rtm/src/cpl/mct/rof_import_export.F90
  I see area multiplied by 0.001_r8 (as Sean does in this PR), except when the area divides volr.
• @slevis-lmwg compare what the nuopc cap looks like in mosart (should be doing the same thing)
  In /glade/work/slevis/git_externals/rtm_fix/components/mosart/src/cpl/nuopc/rof_import_export.F90
  I see every instance of area multiplied by 0.001_r8 (as Sean does in this PR).
  I see the same line that Sean now corrected in this PR (line 200).

[slevis-lmwg]

@ekluzek I tried our first TODO with ctsm5.2.005 as well as with ctsm5.2.0 and got this error in all the tests:

SETUP FAILED for test 'SMS_Ld5_Vmct.f10_f10_mg37.I2000Clm50BgcCropRtm.derecho_intel.rtm-rtmColdStart'.
Command: ./case.setup
Output:
ERROR: No variable BUILD_THREADED found in case

[slevis-lmwg]

I think the problem with testing mct is that it was removed from rtm long before these versions. From a sampling of tags that I have in my directories:

• dev154 has mct
• dev173 doesn't

[slevis-lmwg]

My comments appear in a separate post.

[slevis-lmwg]

@ekluzek I will give you a chance to voice further concerns. I will plan on merging tomorrow afternoon.

[slevis-lmwg]

Made new tag:

git tag -a rtm1_0_80
git push escomp rtm1_0_80