Ensure sfcsub.F of global_cycle is linked to the version used by the forecast model.

[GeorgeGayno-NOAA]

The sfcsub.F module is used by global_cycle AND the forecast model. Both programs should use the same version. Need a way to ensure they are always linked up.

Forecast model version: https://github.com/NCAR/ccpp-physics/blob/master/physics/sfcsub.F

Will need to coordinate with @junwang-noaa

[junwang-noaa]

@climbfuji @DusanJovic-NOAA do you know if there is a way to point to a single file from other repository?

[climbfuji]

I found this on the web, haven't tried it:

First clone the repo with the -n option, which suppresses the default checkout of all files, and the --depth 1 option, which means it only gets the most recent revision of each file

git clone -n git://path/to/the_repo.git --depth 1
Then check out just the file you want like so:

cd the_repo git checkout HEAD name_of_file

[edwardhartnett]

This will not work in general, because it will break the github distribution of tarballs. In other words, git will be the only way the software can be built, which is unusual and unfair.

Is there some library we can put this code in? Some library that both of these applications can link to?

[climbfuji]

This will not work in general, because it will break the github distribution of tarballs. In other words, git will be the only way the software can be built, which is unusual and unfair.

Is there some library we can put this code in? Some library that both of these applications can link to?

It's a single file ... of course we could put it into a separate repository, possibly together with gcycle.F90, if we think this isn't too much overhead. The benefit of using the same repository is big, but it will create a testing dependency between ccpp-physics and UFS_UTILS. Not sure if this is what we want, especially if UFS_UTILS requires changes that then trigger a whole range of updates in all models using ccpp-physics.

[GeorgeGayno-NOAA]

This will not work in general, because it will break the github distribution of tarballs. In other words, git will be the only way the software can be built, which is unusual and unfair. Is there some library we can put this code in? Some library that both of these applications can link to?

It's a single file ... of course we could put it into a separate repository, possibly together with gcycle.F90, if we think this isn't too much overhead. The benefit of using the same repository is big, but it will create a testing dependency between ccpp-physics and UFS_UTILS. Not sure if this is what we want, especially if UFS_UTILS requires changes that then trigger a whole range of updates in all models using ccpp-physics.

I doubt there will be many changes to sfcsub.F that are initiated from UFS_UTILS. The global_cycle program is simply a stand-alone wrapper that runs sfcsub.F. I am more worried about keeping up with changes to sfcsub.F in CCPP.

[climbfuji]

This will not work in general, because it will break the github distribution of tarballs. In other words, git will be the only way the software can be built, which is unusual and unfair. Is there some library we can put this code in? Some library that both of these applications can link to?

It's a single file ... of course we could put it into a separate repository, possibly together with gcycle.F90, if we think this isn't too much overhead. The benefit of using the same repository is big, but it will create a testing dependency between ccpp-physics and UFS_UTILS. Not sure if this is what we want, especially if UFS_UTILS requires changes that then trigger a whole range of updates in all models using ccpp-physics.

I doubt there will be many changes to sfcsub.F that are initiated from UFS_UTILS. The global_cycle program is simply a stand-alone wrapper that runs sfcsub.F. I am more worried about keeping up with changes to sfcsub.F in CCPP.

This file hasn't seen many changes in CCPP either ... maybe twice a year or so in the last few years?

[GeorgeGayno-NOAA]

@SMoorthi-emc I compiled a version of global_cycle that pointed to the CCPP version of sfcsub.F. I then ran it on a canned C768 case. The sea ice records were different between my control and test. I see you made ice-related changes to the CCPP version about a year ago. Can you look at my case and let me know if the results make sense. There are differences in areal ice coverage at the low end of concentration (15%).

[SMoorthi-emc]

Hi George, The version in CCPP allows for different minimum ice concentration for sea ice and lake ice. See the namelist parameters "min_seaice" and min_lakeice" I also probably made some other changes that I don't remember anymore. If you set "min_lakeice=min_seaice=0.15" then it should just like the original version. Actually, I have an updated version of sfcsub.F as the current one in CCPP/master has some issues that I fixed. However the repositories are changing so rapidly that I don't know when I will get a chance to push my changes. If you can point me to the global_cycle version on Hera, I can try to see what the differences are.

[GeorgeGayno-NOAA]

@SMoorthi-emc Thanks for your help.

I placed the version of global_cycle with the ccpp version of sfcsub.F on Hera: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.ccpp/UFS_UTILS.sfcsub The global_cycle code is under ./sorc/global_cycle.fd. The sfcsub.F routine is under ./ccpp-physics/physics.

The current version of global_cycle with its own copy of sfcsub.F is here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.control/UFS_UTILS

I ran a canned C768 test case using both versions. Only tiles 3 and 6 are different. And the difference is related to the minimum ice value. The control files are here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.control/output. The ccpp files are here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.ccpp/output. Difference files for tile 3 and 6 are also included as is the log file (consistency.log).

You will note differences in the ice coverage at the concentration minimum. (The ccpp version has less ice).

Is there an ice check that was changed from (> min_seaice) or (>= min_seaice)?

[SMoorthi-emc]

Hi George, I am not sure how to visualize your output .nc files. Grads does not open them and dbrowser is very slow. I see that the global_cycle version of sfcsub has been significantly changed from the model version over time. I am trying to look at the code differences. Moorthi

On Thu, Apr 22, 2021 at 11:40 AM GeorgeGayno-NOAA @.***> wrote:

@SMoorthi-emc https://github.com/SMoorthi-emc Thanks for your help.

I placed the version of global_cycle with the ccpp version of sfcsub.F on Hera:

/scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.ccpp/UFS_UTILS.sfcsub The global_cycle code is under ./sorc/global_cycle.fd. The sfcsub.F routine is under ./ccpp-physics/physics.

The current version of global_cycle with its own copy of sfcsub.F is here:

/scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.control/UFS_UTILS

I ran a canned C768 test case using both versions. Only tiles 3 and 6 are different. And the difference is related to the minimum ice value. The control files are here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.control/output. The ccpp files are here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.ccpp/output. Difference files for tile 3 and 6 are also included as is the log file (consistency.log).

You will note differences in the ice coverage at the concentration minimum. (The ccpp version has less ice).

Is there an ice check that was changed from (> min_seaice) or (>= min_seaice)?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NOAA-EMC/UFS_UTILS/issues/424#issuecomment-824951806, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYVC74AUBKTAATDZDT3TKA7NJANCNFSM4ZNGKJIA .

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718

e-mail: @.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[GeorgeGayno-NOAA]

Hi George, I am not sure how to visualize your output .nc files. Grads does not open them and dbrowser is very slow. I see that the global_cycle version of sfcsub has been significantly changed from the model version over time. I am trying to look at the code differences. Moorthi … On Thu, Apr 22, 2021 at 11:40 AM GeorgeGayno-NOAA @.> wrote: @SMoorthi-emc https://github.com/SMoorthi-emc Thanks for your help. I placed the version of global_cycle with the ccpp version of sfcsub.F on Hera: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.ccpp/UFS_UTILS.sfcsub The global_cycle code is under ./sorc/global_cycle.fd. The sfcsub.F routine is under ./ccpp-physics/physics. The current version of global_cycle with its own copy of sfcsub.F is here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.control/UFS_UTILS I ran a canned C768 test case using both versions. Only tiles 3 and 6 are different. And the difference is related to the minimum ice value. The control files are here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.control/output. The ccpp files are here: /scratch2/NCEPDEV/stmp1/George.Gayno/sfcsub.test/global-cycle.ccpp/output. Difference files for tile 3 and 6 are also included as is the log file (consistency.log). You will note differences in the ice coverage at the concentration minimum. (The ccpp version has less ice). Is there an ice check that was changed from (> min_seaice) or (>= min_seaice)? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#424 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYVC74AUBKTAATDZDT3TKA7NJANCNFSM4ZNGKJIA . -- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718 e-mail: @. Phone: (301) 683-3718 Fax: (301) 683-3718

I can try to create some Grads control files.

[GeorgeGayno-NOAA]

@SMoorthi-emc I left a Grads control file - tile3.ctl - in each directory. It does not plot a map background. Just the 2-d data. So do 'set mproj off'.

[SMoorthi-emc]

I was able to see the differences via dbrowser. I am not sure what is causing the differences. I am looking at the code.

On Thu, Apr 22, 2021 at 1:33 PM GeorgeGayno-NOAA @.***> wrote:

@SMoorthi-emc https://github.com/SMoorthi-emc I left a Grads control file - tile3.ctl - in each directory. It does not plot a map background. Just the 2-d data. So do 'set mproj off'.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NOAA-EMC/UFS_UTILS/issues/424#issuecomment-825049486, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYSMX4G5PDF4GVO2HLLTKBMX7ANCNFSM4ZNGKJIA .

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718

e-mail: @.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[SMoorthi-emc]

Hi George, The differences are probably becasue of the following differences in the code: (ccpp) " if (slifcs(i) >= 1.99_kind_io8) then if (sicfcs(i) > crit) then tem1 = 1.0_kind_io8 / sicfcs(i) tsffcs(i) = (sicanl(i)tsffcs(i) & + (sicfcs(i)-sicanl(i))tgice) tem1 sitfcs(i) = (tsffcs(i)-tgice(1.0-sicfcs(i))) * tem1 sicfcs(i) = sicanl(i) else tsffcs(i) = tsfanl(i) ! tsffcs(i) = tgice ! sihfcs(i) = sihnew sihfcs(i) = 0.0_kind_io8 sicfcs(i) = 0.0_kind_io8 slifcs(i) = 0.0_kind_io8 endif endif"

and global_cycle: " if (slifcs(i).ge.2.) then if (sicfcs(i).gt.crit) then tsffcs(i) = (sicanl(i)tsffcs(i) & + (sicfcs(i)-sicanl(i))tgice)/sicfcs(i) sitfcs(i) = (tsffcs(i)-tgice*(1.0-sicfcs(i))) / sicfcs(i) else tsffcs(i) = tsfanl(i) ! tsffcs(i) = tgice sihfcs(i) = sihnew endif endif sicfcs(i) = sicanl(i)"

I think my changes in CCPP may not be completely right. I need to think about it a little more. Thanks Moorthi

[SMoorthi-emc]

Actually, it gets even more messier with fractional grid and flake model for lakes.

[SMoorthi-emc]

You may want to consider postponing this unification as the code is even originally incorrect when ice concentration is < 1.0 (this is the first factional grid that was ignored in the original sfcsub). Depending on just slmsk is incorrect.

[GeorgeGayno-NOAA]

You may want to consider postponing this unification as the code is even originally incorrect when ice concentration is < 1.0 (this is the first factional grid that was ignored in the original sfcsub). Depending on just slmsk is incorrect.

Ok. Thanks for taking a look. I will hold off. Let me know if you want help with testing/debugging.

[arunchawla-NOAA]

There is a new request for fractional grid support in global_cycle under ufs-utils. The question comes back to sharing common set of codes in ufs-utils and ccpp. I think we have similar issues with some land utilities as well. Wondering if it makes more sense for these to be as a library set (instead of trying to create repos out of single files). This is specially true if these codes are not changed too much.

[SMoorthi-emc]

The model is getting too fragmented with too many links to repositories. It is hard to keep up.

On Fri, Jul 23, 2021 at 8:39 AM arun chawla @.***> wrote:

There is a new request for fractional grid support in global_cycle under ufs-utils. The question comes back to sharing common set of codes in ufs-utils and ccpp. I think we have similar issues with some land utilities as well. Wondering if it makes more sense for these to be as a library set (instead of trying to create repos out of single files). This is specially true if these codes are not changed too much.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NOAA-EMC/UFS_UTILS/issues/424#issuecomment-885609359, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYTGOABPPCNWTOZUZYDTZFPIFANCNFSM4ZNGKJIA .

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718

e-mail: @.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[arunchawla-NOAA]

I agree and having yet another repository will not be helpful. Would it make more sense for a common set of surface utilities (I remember land needed some as well) be its own library that can be referenced by UFS-utils and the code base ?

[SMoorthi-emc]

Well, one problem is that if the modelers want to change something, they will have to change a library. More work.

On Fri, Jul 23, 2021 at 9:30 AM arun chawla @.***> wrote:

I agree and having yet another repository will not be helpful. Would it make more sense for a common set of surface utilities (I remember land needed some as well) be its own library that can be referenced by UFS-utils and the code base ?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NOAA-EMC/UFS_UTILS/issues/424#issuecomment-885640244, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYR3NJHOGRWKE5JXXTTTZFVFRANCNFSM4ZNGKJIA .

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718

e-mail: @.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[arunchawla-NOAA]

True, how often do these change ? And can these be tested separately as stand alone routines?

[SMoorthi-emc]

Well, since they are not expected to change that often, maybe the status quo may not be that bad. It makes it easier for the modelers to change if needed without having to depend on changing library. (I am not even sure if the standalone global_cycle is even needed as the model calls it in the first time step).

On Fri, Jul 23, 2021 at 10:02 AM arun chawla @.***> wrote:

True, how often do these change ? And can these be tested separately as stand alone routines?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NOAA-EMC/UFS_UTILS/issues/424#issuecomment-885660756, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYQAYTFDDI4V4G5MJETTZFY6NANCNFSM4ZNGKJIA .

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718

e-mail: @.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[arunchawla-NOAA]

we then have the same code in multiple places

[GeorgeGayno-NOAA]

@SMoorthi-emc - I want to update my branch to use the latest CCPP version of sfcsub.F. How is the "min_ice" array set if the orography files do not contain "lakefrac"? I see this in gcycle:

https://github.com/NCAR/ccpp-physics/blob/3721df1561514a15b6141a6461d26a33472a3fa7/physics/gcycle.F90#L168

Or is the latest version of sfcsub.F not backwards compatible with the current orography files, which do not contain the "lakefrac" record?

[GeorgeGayno-NOAA]

Note to self - should I also link to the ccpp version of 'machine.F'?

[SMoorthi-emc]

Hi George, When you say backward compatible, I assume using orography without lake fraction for the uncoupled model, like the current operational GFSV16. Please note that for the uncoupled model we should set min_seaice=min_lakeice=0.15. Only the coupled model with embedded ice model uses different thresholds for ocean and lakes. Nevertheless, you could define a "lakefraction" for the case without lake fraction in "orog" files by what I added in "FV3GFS_io.F90", which for the full grid case looks like " if (nint(Sfcprop(nb)%slmsk(ix)) == 1 .and. Sfcprop(nb)%stype(ix) > 0 & .and. Sfcprop(nb)%stype(ix) /= 14) then Sfcprop(nb)%landfrac(ix) = one Sfcprop(nb)%lakefrac(ix) = zero Sfcprop(nb)%oceanfrac(ix) = zero else Sfcprop(nb)%slmsk(ix) = zero Sfcprop(nb)%landfrac(ix) = zero if (Sfcprop(nb)%oro_uf(ix) > min_lake_orog) then ! lakes Sfcprop(nb)%lakefrac(ix) = one Sfcprop(nb)%oceanfrac(ix) = zero if (Sfcprop(nb)%fice(ix) > Model%min_lakeice) Sfcprop(nb)%slmsk(ix) = 2.0 else ! ocean Sfcprop(nb)%lakefrac(ix) = zero Sfcprop(nb)%oceanfrac(ix) = one if (Sfcprop(nb)%fice(ix) > Model%min_seaice) Sfcprop(nb)%slmsk(ix) = 2.0 endif" I set min_lake_orog to 200m to avoid coastal ocean treated as lakes. In any case, we don't really use this option in the coupled model. I hope this helps. Moorthi

On Fri, Mar 18, 2022 at 4:46 PM GeorgeGayno-NOAA @.***> wrote:

@SMoorthi-emc https://github.com/SMoorthi-emc - I want to update my branch to use the latest CCPP version of sfcsub.F. How is the "min_ice" array set if the orography files do not contain "lakefrac"? I see this in gcycle:

https://github.com/NCAR/ccpp-physics/blob/3721df1561514a15b6141a6461d26a33472a3fa7/physics/gcycle.F90#L168

Or is the latest version of sfcsub.F not backwards compatible with the current orography files, which do not contain the "lakefrac" record?

— Reply to this email directly, view it on GitHub https://github.com/ufs-community/UFS_UTILS/issues/424#issuecomment-1072784249, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYR2CVDTUQMOZYIXHHDVATTR5ANCNFSM4ZNGKJIA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718

e-mail: @.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[GeorgeGayno-NOAA]

Ran the consistency tests using bdcdb38 on WCOSS-Dell. The C768 test (that mimics GFS OPS) failed in regions with ice (tiles 1, 2, 4 and 5 were bit identical to the baseline). Differences were small. But are they significant?

Tile 3

Variable Group Count       Sum   AbsSum         Min          Max      Range        Mean      StdDev
tsea     /      9340   113.615  113.634 -0.00945343     0.248625   0.258078   0.0121643   0.0527147
hice     /       477 -0.578968 0.578968 -0.00294555 -8.92851e-05 0.00285627 -0.00121377 0.000515948
tisfc    /     10527   757.433  757.559  -0.0630229       1.6575    1.72052   0.0719514    0.332016

Tile 6

Variable Group Count        Sum     AbsSum          Min         Max       Range         Mean      StdDev
tsea     /     36232   -4.22314    4.75771     -1.55269    0.142046     1.69473 -0.000116558   0.0112281
hice     /         9 0.00153423 0.00182039 -7.16645e-05 0.000585212 0.000656876   0.00017047 0.000203502
tisfc    /     39411   -28.1543    31.7181     -10.3512    0.946975     11.2982 -0.000714376   0.0717717

@SMoorthi-emc Are these differences expected? I can create difference plots if that would help.

[SMoorthi-emc]

George, Can you explain a bit about this test? Is it on full grid? Are you setting min_ice=0.15 for both ocean and lakes? Also, can you point me to your source on dell for both opr and new so that I can actually look at the differences in the code? Thanks Moorthi

[GeorgeGayno-NOAA]

George, Can you explain a bit about this test? Is it on full grid? Are you setting min_ice=0.15 for both ocean and lakes? Also, can you point me to your source on dell for both opr and new so that I can actually look at the differences in the code? Thanks Moorthi

To see the source code, you can clone my fork:

git clone https://github.com/GeorgeGayno-NOAA/UFS_UTILS.git

Then checkout the feature/sfcsub_ccpp branch.

Code differences are here: https://github.com/ufs-community/UFS_UTILS/pull/636/files

[GeorgeGayno-NOAA]

George, Can you explain a bit about this test? Is it on full grid? Are you setting min_ice=0.15 for both ocean and lakes? Also, can you point me to your source on dell for both opr and new so that I can actually look at the differences in the code? Thanks Moorthi

The test is a global C768 grid. The data directory is on Venus: /gpfs/dell1/stmp/George.Gayno/reg-tests.sfcsub/global-cycle/test1.

The fort.35/36/37 namelists were saved. Sea ice is being applied - FNACNA is pointing to this ice file: /gpfs/dell2/emc/modeling/noscrub/George.Gayno/ufs_utils.git/reg_tests/global_cycle/input_data/gdas.t00z.seaice.5min.blend.grb.

I believe I am using a min ice of 15%.

[SMoorthi-emc]

I looked at the differences in the code. There are some minor differences in how ice is handled. So I assume there will be some differences where sea-ice/lake_ice are involved. Since the model is working fine, I have to assume the version in the model also is fine. It would be instructive if the differences and the total fields are plotted. I am neither that familiar with plotting with netcdf nor with plotting on tiles.

[GeorgeGayno-NOAA]

I looked at the differences in the code. There are some minor differences in how ice is handled. So I assume there will be some differences where sea-ice/lake_ice are involved. Since the model is working fine, I have to assume the version in the model also is fine. It would be instructive if the differences and the total fields are plotted. I am neither that familiar with plotting with netcdf nor with plotting on tiles.

I will make some plots.

[GeorgeGayno-NOAA]

Ran the consistency tests using bdcdb38 on WCOSS-Dell. The C768 test (that mimics GFS OPS) failed in regions with ice (tiles 1, 2, 4 and 5 were bit identical to the baseline). Differences were small. But are they significant?

Tile 3

Variable Group Count       Sum   AbsSum         Min          Max      Range        Mean      StdDev
tsea     /      9340   113.615  113.634 -0.00945343     0.248625   0.258078   0.0121643   0.0527147
hice     /       477 -0.578968 0.578968 -0.00294555 -8.92851e-05 0.00285627 -0.00121377 0.000515948
tisfc    /     10527   757.433  757.559  -0.0630229       1.6575    1.72052   0.0719514    0.332016

Tile 6

Variable Group Count        Sum     AbsSum          Min         Max       Range         Mean      StdDev
tsea     /     36232   -4.22314    4.75771     -1.55269    0.142046     1.69473 -0.000116558   0.0112281
hice     /         9 0.00153423 0.00182039 -7.16645e-05 0.000585212 0.000656876   0.00017047 0.000203502
tisfc    /     39411   -28.1543    31.7181     -10.3512    0.946975     11.2982 -0.000714376   0.0717717

@SMoorthi-emc Are these differences expected? I can create difference plots if that would help.

Some printout from the C768 consistency test (branch and develop). These TILE 6 points have the largest ice temperature differences (TISFC). Note that all these points had ice concentrations (SICFCS) at the minimum - 15% - before and after the call to sfccycle. At ice points, skin temperature (TSEA) is a linear combination of the ice temperature and the open water temperature (271.2) based on the ice concentration. If ice concentration does not change, I would not expect TSEA or TISFC to change. But that is what is happening with 'develop' (TSEA and TISFC become equal). The branch, which uses the CCPP version of sfcsub.F, does not change TSEA and TISFC. That makes sense to me. Does the CCPP version have a bug fix?

'develop'

These points at tile 6

                     SLIFCS             SICFCS            TSEA            TISFC
(258,201) before  2.00000000000000  0.150000000000000  269.373310985963  259.022073239755
(216,238) before  2.00000000000000  0.150000000000000  270.051486406558  263.543242710385
(324,633) before  2.00000000000000  0.150000000000000  271.347340912173  272.182272747823

                    SLIFCS             SICFCS            TSEA            TISFC 
(258,201) after   2.00000000000000  0.150000000000000  269.373310985963  269.373310985963
(216,238) after   2.00000000000000  0.150000000000000  270.051486406558  270.051486406558
(324,633) after   2.00000000000000  0.150000000000000  271.347340912173  271.347340912173

branch at db546417

                     SLIFCS             SICFCS            TSEA            TISFC
(258,201) before  2.00000000000000  0.150000000000000  269.373310985963  259.022073239755
(216,238) before  2.00000000000000  0.150000000000000  270.051486406558  263.543242710385
(324,633) before  2.00000000000000  0.150000000000000  271.347340912173  272.182272747823

                     SLIFCS             SICFCS            TSEA           TISFC
(258,201) after   2.00000000000000  0.150000000000000  269.373310985963  259.022073239755
(216,238) after   2.00000000000000  0.150000000000000  270.051486406558  263.543242710385
(324,633) after   2.00000000000000  0.150000000000000  271.347340912173  272.182272747823

[SMoorthi-emc]

I did make several updates in the code related to the ice. It happened as I was trying to take into account the fractional grid. I do not specifically recall the details as this was done some time ago. Moorthi

On Fri, Mar 25, 2022 at 3:35 PM GeorgeGayno-NOAA @.***> wrote:

Ran the consistency tests using bdcdb38 https://github.com/ufs-community/UFS_UTILS/commit/bdcdb383529db8830ba91fc1ffb889b64cce9bab on WCOSS-Dell. The C768 test (that mimics GFS OPS) failed in regions with ice (tiles 1, 2, 4 and 5 were bit identical to the baseline). Differences were small. But are they significant?

Tile 3

Variable Group Count Sum AbsSum Min Max Range Mean StdDev tsea / 9340 113.615 113.634 -0.00945343 0.248625 0.258078 0.0121643 0.0527147 hice / 477 -0.578968 0.578968 -0.00294555 -8.92851e-05 0.00285627 -0.00121377 0.000515948 tisfc / 10527 757.433 757.559 -0.0630229 1.6575 1.72052 0.0719514 0.332016

Tile 6

Variable Group Count Sum AbsSum Min Max Range Mean StdDev tsea / 36232 -4.22314 4.75771 -1.55269 0.142046 1.69473 -0.000116558 0.0112281 hice / 9 0.00153423 0.00182039 -7.16645e-05 0.000585212 0.000656876 0.00017047 0.000203502 tisfc / 39411 -28.1543 31.7181 -10.3512 0.946975 11.2982 -0.000714376 0.0717717

@SMoorthi-emc https://github.com/SMoorthi-emc Are these differences expected? I can create difference plots if that would help.

Some printout from the C768 consistency test (branch and develop). These TILE 6 points have the largest ice temperature differences (TISFC). Note that all these points had ice concentrations (SICFCS) at the minimum - 15%

• before and after the call to sfccycle. At ice points, skin temperature (TSEA) is a linear combination of the ice temperature and the open water temperature (271.2) based on the ice concentration. If ice concentration does not change, I would not expect TSEA or TISFC to change. But that is what is happening with 'develop' (TSEA and TISFC become equal). The branch, which uses the CCPP version of sfcsub.F, does not change TSEA and TISFC. That makes sense to me. Does the CCPP version have a bug fix?

'develop'

These points at tile 6

                 SLIFCS             SICFCS            TSEA            TISFC

(258,201) before 2.00000000000000 0.150000000000000 269.373310985963 259.022073239755 (216,238) before 2.00000000000000 0.150000000000000 270.051486406558 263.543242710385 (324,633) before 2.00000000000000 0.150000000000000 271.347340912173 272.182272747823

                SLIFCS             SICFCS            TSEA            TISFC

(258,201) after 2.00000000000000 0.150000000000000 269.373310985963 269.373310985963 (216,238) after 2.00000000000000 0.150000000000000 270.051486406558 270.051486406558 (324,633) after 2.00000000000000 0.150000000000000 271.347340912173 271.347340912173

branch at db54641 https://github.com/ufs-community/UFS_UTILS/commit/db54641789469eb2a0eb59ea5ada875a8f962e63

                 SLIFCS             SICFCS            TSEA            TISFC

(258,201) before 2.00000000000000 0.150000000000000 269.373310985963 259.022073239755 (216,238) before 2.00000000000000 0.150000000000000 270.051486406558 263.543242710385 (324,633) before 2.00000000000000 0.150000000000000 271.347340912173 272.182272747823

                 SLIFCS             SICFCS            TSEA           TISFC

(258,201) after 2.00000000000000 0.150000000000000 269.373310985963 259.022073239755 (216,238) after 2.00000000000000 0.150000000000000 270.051486406558 263.543242710385 (324,633) after 2.00000000000000 0.150000000000000 271.347340912173 272.182272747823

— Reply to this email directly, view it on GitHub https://github.com/ufs-community/UFS_UTILS/issues/424#issuecomment-1079373303, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALLVRYVZB55TLTS3NTC4QEDVBYIQBANCNFSM4ZNGKJIA . You are receiving this because you were mentioned.Message ID: @.***>

-- Dr. Shrinivas Moorthi Research Meteorologist Modeling and Data Assimilation Branch Environmental Modeling Center / National Centers for Environmental Prediction 5830 University Research Court - (W/NP23), College Park MD 20740 USA Tel: (301)683-3718

e-mail: @.*** Phone: (301) 683-3718 Fax: (301) 683-3718

[GeorgeGayno-NOAA]

Some graphics from the tests described here: https://github.com/ufs-community/UFS_UTILS/issues/424#issuecomment-1079373303

Note how the biggest differences in TSEA and TISFC only occur at (a very few) points where the sea ice is at the 15% threshold. As shown by the printout of several points in the previous comment. The CCPP version of sfcsub.F seems to have corrected a problem at the 15% ice threshold.

[GeorgeGayno-NOAA]

I did make several updates in the code related to the ice. It happened as I was trying to take into account the fractional grid. I do not specifically recall the details as this was done some time ago. Moorthi

Maybe you fixed something, or the 'develop' version of global_cycle was not calling sfcsub.F properly. Anyway, I think the branch which uses the CCPP version is correct. Here is another look at one point from tile 6 before and after the call to sfcsub.F.

SLIFCS SICFCS TSEA TISFC (258,201) before 2.00000000000000 0.150000000000000 269.373310985963 259.022073239755 (258,201) after (develop) 2.00000000000000 0.150000000000000 269.373310985963 269.373310985963 (258,201) after (CCPP) 2.00000000000000 0.150000000000000 269.373310985963 259.022073239755

If the ice fraction does not change, I don't see why TISFC or TSEA should change. And since TSEA is a blend of the ice skin temperature and the open water temperature (271.2), TISFC and TSEA should not be equal at ice points.