Regional inlinepost

[junwang-noaa]

The code changes are required for regional inline post capability in ufs-weather-model. The code changes include:

1. remove the MPI_COMM_WORLD in sorc/ncep_post.fd/AllGETHERV_GSD.f that blocks the inline post since the inline post is done on write grid component which is only part of the MPI_COMM_WORLD.

2. The fields on on regional domain have undefined values on some of the grid points, computation on those points causes issues. Code changes have be added to some post files to allow the computations only to occur on the grid points with valid value fields and skip the computation on grid points with undefined value fields.

[WenMeng-NOAA]

@junwang-noaa Could you sync your branch with the latest update from upstream/develop? I will start test your PR for next integration. Thanks!

[junwang-noaa]

Updated, please check.

…

On Tue, Jan 12, 2021 at 7:45 PM WenMeng-NOAA ***@***.***> wrote: @junwang-noaa <https://github.com/junwang-noaa> Could you sync your branch with the latest update from upstream/develop? I will start test your PR for next integration. Thanks! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#255 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI7D6TKDLXRKT3EEWB74CZLSZTUKBANCNFSM4V7Y6M6Q> .

[WenMeng-NOAA]

@junwang-noaa I saw the UPP standalone test for nmmb has changed result at pe_test in field HCDC:high cloud layer :

---

-rw-r--r-- 1 Wen.Meng stmp 0 Jan 13 04:01 post_nmmb_meso_2014120818_Grib2/BGDAWP03.tm00.Grib2.diff
-rw-r--r-- 1 Wen.Meng stmp 71 Jan 13 04:02 post_nmmb_meso_2014120818_Grib2_pe_test/BGDAWP03.tm00.Grib2.diff
-rw-r--r-- 1 Wen.Meng stmp 0 Jan 13 04:02 post_nmmb_meso_2014120818_Grib2/BGRD3D03.tm00.Grib2.diff
-rw-r--r-- 1 Wen.Meng stmp 72 Jan 13 04:02 post_nmmb_meso_2014120818_Grib2_pe_test/BGRD3D03.tm00.Grib2.diff
-rw-r--r-- 1 Wen.Meng stmp 69 Jan 13 04:02 post_nmmb_meso_2014120818_Grib2_pe_test/BGRDSF03.tm00.Grib2.diff
-rw-r--r-- 1 Wen.Meng stmp 0 Jan 13 04:02 post_nmmb_meso_2014120818_Grib2/BGRDSF03.tm00.Grib2.diff

[Wen.Meng@hfe04 ~/stmp2]$ cat post_nmmb_meso_2014120818_Grib2_pe_test/BGRD3D03.tm00.Grib2.diff
1364:413841521:HCDC:high cloud layer:rpn_corr=0.999525:rpn_rms=0.119973

[hu5970]

@WenMeng-NOAA @junwang-noaa After finalize the code and pass the regression test, I suggest to improve the readability of the code by indenting the block inside "if ... else ... endif".

Thanks,
Ming

[WenMeng-NOAA]

@junwang-noaa You might sync your branch with the latest commit from upstream/develop. With the UPP standalone cmake build capacity added in, you may run build script compile_upp.sh under tests subdirectory to generate all UPP executable, upp lib and include files in tests/install. The new capacity can be supported on WCOSS-dell, WCOSS-cray, Hera, Orion and Jet.

[BinLiu-NOAA]

With the updated HAFS post control files:
makefile postcntrl_hafs.xml postxconfig-NT-hafs.txt
on Orion under
/work/noaa/hwrf/save/bliu/hafs_inlinepost/sorc/hafs_post.fd/parm
both the offline and inline post produce the same number of grib2 records now. And the differences between the inline and offline post results are negligible.

So, @junwang-noaa or @WenMeng-NOAA, please help to incorporate these new HAFS post control files in this branch if possible, as well as in your ufs-weather-model level HAFS inline post regression tests.

Thanks!

Bin

[BinLiu-NOAA]

HAFS inline and offline post now produce the same number of grib2 records, and the differences between the inline and offline post results are negligible.

[WenMeng-NOAA]

@junwang-noaa Could you update VERSION file to 10.0.4? This PR would be a major upgrade for upp library. Thanks!

[junwang-noaa]

Updated.

…

On Thu, Feb 18, 2021 at 6:46 PM WenMeng-NOAA ***@***.***> wrote: @junwang-noaa <https://github.com/junwang-noaa> Could you update VERSION file to 10.0.4? This PR would be a major upgrade for upp library. Thanks! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#255 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI7D6TM6LPA5E2CZ6N22TJDS7WRFFANCNFSM4V7Y6M6Q> .

[WenMeng-NOAA]

The UPP standalone regression tests were finished on WCOSS-Dell, WCOSS-Cray and Hera. With this merge the baseline of FV3R should be re-created.

[WenMeng-NOAA]

@junwang-noaa Your PR is in a good shape. But I would like to hold off merging because there might be a potential thread issue for building upp lib via cmake without openmp disabled. Please let me if you any concerns/suggestion.

[junwang-noaa]

That is OK with me. It's good to have the openmp issue resolved.

…

On Fri, Feb 19, 2021 at 7:43 PM WenMeng-NOAA ***@***.***> wrote: @junwang-noaa <https://github.com/junwang-noaa> Your PR is in a good shape. But I would like to hold off merging because there might be a potential thread issue for building upp lib via cmake without openmp disabled. Please let me if you any concerns/suggestion. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#255 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI7D6TMXKAH6VZYYBJ5RKZ3S74AR7ANCNFSM4V7Y6M6Q> .

[junwang-noaa]

Currently POST code is not thread safe. Work needs to be done to add threading capability in POST. In this commit, the post and associated nceplibs need to be built without threading in order to get identical results.

[WenMeng-NOAA]

This PR is ready for merging.