add clipping treatment for negative values of non-negative tracers

[AmandaBack-NOAA]

Description

Clips negative values from fields in the tracer file before writing. Fixes #680.

Type of change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

Tested on Hera in a RRFS CONUS workflow with both hybrid_radar_ref and anal_gsi_input tasks.
See run directory at /scratch2/BMC/zrtrr/aback/hydrometeors_test/
For one hour of the control run without this change, and one hour using the change contained in this PR, the INPUT directory was saved for reference/rerunning before each of the tasks: anal_gsi_input_spinup, hybrid_radar_ref_spinup, cldanl_nonvar_spinup, and run_fcst spinup. Those directories are at, for example /scratch2/BMC/zrtrr/aback/hydrometeors_test/INPUT_after_anal_gsi_input_spinup_before_hybrid_radar_ref_spinup ("control") and /scratch2/BMC/zrtrr/aback/hydrometeors_test/clipped_INPUT_after_anal_gsi_input_spinup_before_hybrid_radar_ref_spinup
(using the code in the PR).
Logs are at /scratch2/BMC/zrtrr/aback/hydrometeors_test/v0.7.9/logs/rrfs.20230610/08/, including new logs produced between all the above-listed tasks, showing all the min and max values in fv_tracer.

Checklist

The code is Ting Lei's. It looks stylish and understandable to me (Amanda) but maybe Ting should weigh in on this section.

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• New and existing tests pass with my changes
• Any dependent changes have been merged and published

DUE DATE for merger of this PR into develop is 2/22/2024 (six weeks after PR creation).

[hongli-wang]

The changes look good to me. I approve this PR.

[ShunLiu-NOAA]

@AmandaBack-NOAA Could you run regression test on Hera, Orion and WCOSS2? Thanks.

[hu5970]

@AmandaBack-NOAA @TingLei-NOAA Could you sync with current develop branch? Thanks, Ming

[AmandaBack-NOAA]

@hu5970 synced.

[hu5970]

@AmandaBack-NOAA Thanks.

[hu5970]

The regression test results on Hera:

[Ming.Hu@hfe03 build]$ ctest -j7
Test project /scratch1/BMC/wrfruc/mhu/code/gsi/GSI/build
    Start 1: global_4denvar
    Start 2: rtma
    Start 3: rrfs_3denvar_glbens
    Start 4: netcdf_fv3_regional
    Start 5: hafs_4denvar_glbens
    Start 6: hafs_3denvar_hybens
    Start 7: global_enkf
1/7 Test #4: netcdf_fv3_regional ..............   Passed  610.69 sec
2/7 Test #3: rrfs_3denvar_glbens ..............***Failed  619.89 sec
3/7 Test #7: global_enkf ......................   Passed  1083.49 sec
4/7 Test #2: rtma .............................   Passed  1155.76 sec
5/7 Test #6: hafs_3denvar_hybens ..............***Failed  1347.53 sec
6/7 Test #5: hafs_4denvar_glbens ..............***Failed  1412.45 sec
7/7 Test #1: global_4denvar ...................   Passed  1670.17 sec

57% tests passed, 3 tests failed out of 7

Total Test time (real) = 1670.22 sec

The following tests FAILED:
          3 - rrfs_3denvar_glbens (Failed)
          5 - hafs_4denvar_glbens (Failed)
          6 - hafs_3denvar_hybens (Failed)
Errors while running CTest
Output from these tests are in: /scratch1/BMC/wrfruc/mhu/code/gsi/GSI/build/Testing/Temporary/LastTest.log
Use "--rerun-failed --output-on-failure" to re-run the failed cases verbosely.

The reason of failure for three regional cases is:

The fv3_dynvars are reproducible
The fv3_sfcdata are reproducible
The results between the two runs (rrfs_3denvar_glbens_loproc_updat and rrfs_3denvar_glbens_loproc_contrl) are not reproducible
Thus, the case has Failed siganl of the regression tests.

It is the difference of the fv3_tracer files between control and update runs caused failure. When diff those two tracer files with ncdiff:
ncdiff hafs_4denvar_glbens_loproc_updat/fv3_tracer hafs_4denvar_glbens_loproc_contrl/fv3_tracer diff_tracer.nc
Only sphum fields show positive values. All other fields are identical.

The same diff results are also true for "hafs_3denvar_hybens" and "rrfs_3denvar_glbens".

This PR is to "Clips negative values from fields in the tracer file before writing". This difference in moisture field is expected.
So the regression tests are passed.

[hu5970]

The regression test results from Cactus:

[ming.hu@clogin05 build] ctest -j7
Test project /lfs/h2/emc/ptmp/Ming.Hu/gsi/GSI/build
    Start 1: global_4denvar
    Start 2: rtma
    Start 3: rrfs_3denvar_glbens
    Start 4: netcdf_fv3_regional
    Start 5: hafs_4denvar_glbens
    Start 6: hafs_3denvar_hybens
    Start 7: global_enkf
1/7 Test #4: netcdf_fv3_regional ..............   Passed  483.52 sec
2/7 Test #7: global_enkf ......................   Passed  621.34 sec
3/7 Test #3: rrfs_3denvar_glbens ..............***Failed  665.50 sec
4/7 Test #6: hafs_3denvar_hybens ..............***Failed  1213.37 sec
5/7 Test #5: hafs_4denvar_glbens ..............***Failed  1284.06 sec
6/7 Test #1: global_4denvar ...................   Passed  1567.19 sec

Test project /lfs/h2/emc/ptmp/Ming.Hu/gsi/GSI/build
    Start 2: rtma
1/1 Test #2: rtma .............................   Passed  968.63 sec

100% tests passed, 0 tests failed out of 1

Total Test time (real) = 968.63 sec

The reason of failure for three regional cases are the same as the results from Hera tests:

The fv3_dynvars are reproducible
The fv3_sfcdata are reproducible
The results between the two runs (rrfs_3denvar_glbens_loproc_updat and rrfs_3denvar_glbens_loproc_contrl) are not reproducible
Thus, the case has Failed siganl of the regression tests.

So the regression test on Wcoss2 passed.

[TingLei-NOAA]

Ming Thank you so much for doing the regression test and your clarifiying of the results. Ting

…

______________________________ Ting Lei Physical Scientist, Contractor with Lynker in support of EMC/NCEP/NWS/NOAA 5830 University Research Ct., Cubicle 2765 College Park, MD 20740 ***@***.*** 301-683-3624

On Sun, Jan 21, 2024 at 8:18 PM Ming Hu ***@***.***> wrote: The regression test results from Cactus: ***@***.*** build] ctest -j7 Test project /lfs/h2/emc/ptmp/Ming.Hu/gsi/GSI/build Start 1: global_4denvar Start 2: rtma Start 3: rrfs_3denvar_glbens Start 4: netcdf_fv3_regional Start 5: hafs_4denvar_glbens Start 6: hafs_3denvar_hybens Start 7: global_enkf 1/7 Test #4: netcdf_fv3_regional .............. Passed 483.52 sec 2/7 Test #7: global_enkf ...................... Passed 621.34 sec 3/7 Test #3: rrfs_3denvar_glbens ..............***Failed 665.50 sec 4/7 Test #6: hafs_3denvar_hybens ..............***Failed 1213.37 sec 5/7 Test #5: hafs_4denvar_glbens ..............***Failed 1284.06 sec 6/7 Test #1: global_4denvar ................... Passed 1567.19 sec Test project /lfs/h2/emc/ptmp/Ming.Hu/gsi/GSI/build Start 2: rtma 1/1 Test #2: rtma ............................. Passed 968.63 sec 100% tests passed, 0 tests failed out of 1 Total Test time (real) = 968.63 sec — Reply to this email directly, view it on GitHub <#682 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/APEFS7G7Q23DDRRMXSYH2CDYPW44JAVCNFSM6AAAAABBWXQBYSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMBSHA2TEMJSHE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[RussTreadon-NOAA]

While this PR is closed, I was curious as to why update_guess.f90 was not effective in ensuring a non-negative sphum analysis increment. The rrfs_3denvar_glbens ctest was used to investigate negative sphum values.

Print statements were added to update_guess.f90 and gsi_rfv3io_mod.f90.

• confirm that update_guess ensures sphum analysis is greater or equal to the specified bound qcmin which is set to 1.0e-7 in constants.f90. That is, sphum is positive definite on the GSI analysis grid.
• prints added to gsi_rfv3io_mod.f90 for sphum show that adding workb2 to work_b can result in negative values in work_b. Logic added by this PR detects this and resets negative work_b to 0.0

Is it possible that that negative values of sphum originate as a result of operations performed in fv3_h_to_ll?

[hongli-wang]

That is a good question. Ting and I had discussed about this. fv3_h_to_ll could be one of possible reasons. Ting had other finding and he may add more about it.

[TingLei-NOAA]

I have no more to add to @hongli-wang and yes, we think that fv3_ll_to_h is the reason for negative values in q/sphum. Thanks to @RussTreadon-NOAA 's digging in and verification of its actual behavior.