NoahMP CCPP-compliancy #305

grantfirl · 2019-08-24T04:26:56Z

Description of changes in ccpp-physics only:

added NoahMP scheme files to repo
added metadata file for NoahMP entry point
minimal CCPP-compliancy stuff (put in module, init/run/finalize)
fix some units of surface-related variables (some variables were being converted in GFS_diag so that they had non-canonical units within the CCPP; changed so that conversion is no longer necessary and they have the correct units in the CCPP)
add errmsg and errflg for CCPP error handling
ifdef WRF error handling out
set errmsg, errflg and returns throughout internal NoahMP calling structure
send physical constants through the argument list (but see issue NoahMP uses internal constants rather than those provided from host #318 )
correct standard name and description of the 'solhr' variable
call set_soilveg in noahmpdrv_init (Note: this is also done in lsm_noah_init -- need to do something in common so as not to repeat?)

Interstitials:

move calculation of sncovr from FV3GFS_io.F90; this was needed not only because it should not have been done in an I/O routine, but there was a conflict with the namelist_soilveg module from non-CCPP code being used erroneously in the CCPP.
this calculation was moved to GFS_phys_time_vary_run because 1) it cannot go into an *_init subroutine since it was originally called after the ccpp_phys_init stage and 2) it needs to be calculated before radiation is called since sncovr is used in radiation (through the GFS sfcprop DDT); this is done during the first time step only!
see closed PR Move calculation of sncovr from FV3/io/FV3GFS_io.F90 to ccpp/physics/physics/GFS_phys_time_vary.XYZ.F90 #315 for full details (all changes from that PR were included in this one)
remove GFS DDT from GFS_time_vary_pre.F90 and do calculation of julian day there (in run) since it is not domain-dependent
move the t2m and q2m "override" into sfc_diag_post
calculate precipitation rates used in NoahMP in sfc_noahmp_pre.F90 (but see issue Precipitation rates for NoahMP are likely being calculated in the wrong place #319 )

…eadcrumb

… in meta file

…or handling dates)

grantfirl · 2019-08-24T04:31:40Z

Associated PRs:
NEMSfv3gfs: https://github.com/NCAR/NEMSfv3gfs/pull/227
FV3: https://github.com/NCAR/FV3/pull/192
ccpp-physics: #305

grantfirl · 2019-08-24T04:32:31Z

Status: Compiles successfully using Theia/Intel, but no attempt to run RTs yet.

grantfirl · 2019-08-26T16:14:09Z

physics/noahmp_pre.F90

@@ -0,0 +1,72 @@
+!> \file noahmp_pre.F90


@climbfuji This code for calculating julian day and the year length was in GFS_physics_driver.F90, so I initially put this code in a NoahMP-specific interstitial scheme, but thinking a bit more, this code does not depend on location (only time), so why couldn't it be put in GFS_time_vary_pre_run where other time-related quantities are calculated to save some CPU cycles?

This sounds like a reasonable idea to me!

OK, thanks, I'll do that.

…y_pre_run; update GFS_time_vary_pre_run.fv3.F90 to new metadata and remove DDT dependency

… because it was originally done after the call to ccpp_physics_init in FV3GFS_io.F90/sfc_prop_restart_read

…use it was originally done after the call to ccpp_physics_init in FV3GFS_io.F90/sfc_prop_restart_read); this needs to happen before radition is called

…ic_post_run

…is active

…specific interstitial to test for b4b

climbfuji · 2019-09-13T21:10:56Z

physics/GFS_time_vary_pre.fv3.F90

+
+        if(lsm == lsm_noahmp) then
+          !GJF* These calculations were originally in GFS_physics_driver.F90 for 
+          !     NoahMP. They were moved to this routine since they only depends 


I will correct the typo "they only depends" to "they only depend" in my collect-and-test PR.

For the record - done.

physics/module_sf_noahmp_glacier.f90

climbfuji · 2019-09-13T21:14:30Z

physics/module_sf_noahmp_glacier.f90

+                         tbot   ,zbot   ,zsnso  ,dzsnso ,                 & !in
+                         tg     ,stc    ,snowh  ,sneqv  ,sneqvo ,sh2o   , & !inout
+                         smc    ,snice  ,snliq  ,albold ,cm     ,ch     , & !inout
+#ifdef CCPP


Just wondering why we need ifdef CCPP in a CCPP physics scheme? Do you want to keep this for comparing it/merging our with code from other/the authoritative repositories/repository?

The ifdef CCPP in all of the code in ccpp-physics for NoahMP was an attempt to try to anticipate these changes going back to an authoritative repo outside of CCPP so that the code could continue to be used as-is outside of the CCPP.

climbfuji · 2019-09-13T21:15:54Z

physics/module_sf_noahmplsm.f90

+        if(fveg <= 0.05) fveg = 0.05
+     else
+        write(*,*) "-------- fatal called in sflx -----------"
+#ifdef CCPP


Very nice, thanks for making those code changes.

climbfuji

A massive amount of work. I did not check every line in this PR (8439 lines of code in one file alone - nature can't be this complicated).

I left a few notes for the future.

I will pull this PR into a new PR that collects MYJ, NoahMP and other developments and tests them together.

grantfirl · 2019-09-13T23:31:29Z

Issues #318 and #319 should be addressed in a followup PR sometime when a change in baseline is acceptable.

NCAR#305) * Introduction of CA domain definition. * Read/Write and restart capability of CA * Change namelist control parameter so that CA time and space scales are adjusted for various resolutions given the same CA namelist input. * Significant cleaning of old ideas. * Improved coupling to cumulus convection.

grantfirl added 9 commits August 20, 2019 11:40

initial addition of CCPP metadata for NoahMP

e48c99c

Merge branch 'fix_LSM_units' into noahmp

366378c

put sfc_noahmp_drv.f in module; add init/finalize + meta file html br…

6236796

…eadcrumb

fix soil_moisture_content unit error

747c5a4

add CCPP error variables to sfc_noahmp_drv.f and fix local name error…

aa5aec7

… in meta file

add NoahMP interstitial code (only pre routine needed for GFS suite f…

e5e6b78

…or handling dates)

send physical constants through the argument list

6c2cab8

remove WRF error handling in favor of CCPP error handling

24b8942

pass errmsg and errflg down call chain

1098500

grantfirl added work in progress do not merge labels Aug 24, 2019

grantfirl requested review from climbfuji and llpcarson as code owners August 24, 2019 04:26

grantfirl commented Aug 26, 2019

View reviewed changes

move NoahMP calculation of julian day and year length to GFS_time_var…

dab6e6d

…y_pre_run; update GFS_time_vary_pre_run.fv3.F90 to new metadata and remove DDT dependency

grantfirl mentioned this pull request Aug 27, 2019

Address issue #300 #302

Merged

grantfirl requested a review from JulieSchramm as a code owner August 30, 2019 01:26

grantfirl force-pushed the noahmp branch from 252b4e9 to 9855548 Compare August 30, 2019 01:38

add set_soilveg to noahmpdrv_init

d0af67a

grantfirl force-pushed the noahmp branch from 9855548 to d0af67a Compare August 30, 2019 01:45

grantfirl added 8 commits August 30, 2019 18:02

add calculation of sncovr to GFS_surface_generic_pre_run (not to init…

3e2f892

… because it was originally done after the call to ccpp_physics_init in FV3GFS_io.F90/sfc_prop_restart_read

add calculation of sncovr to GFS_phys_time_vary_run (not to init beca…

78523cf

…use it was originally done after the call to ccpp_physics_init in FV3GFS_io.F90/sfc_prop_restart_read); this needs to happen before radition is called

Merge branch 'fix_sncovr' into noahmp

64f70df

Merge branch 'gmtb/develop' into noahmp

6d04c51

fix GFS_time_vary_pre.fv3.F90 compilation error

af34508

add calculation of precipitation variables for NoahMP in GFS_MP_gener…

619afe0

…ic_post_run

add resetting of sfcprop%t2m and q2m to values from noahmp if noahmp …

de728a8

…is active

(temporarily?) move calculation of precip rates for NoahMP to scheme-…

775f4ff

…specific interstitial to test for b4b

grantfirl added 3 commits September 10, 2019 17:20

Merge branch 'gmtb/develop' into noahmp

a311364

revert whitespace changes to GFS_MP_generic.F90

e7bb31a

fix bug in metadata table for GFS_phys_time_vary_run for FV3

69c215f

climbfuji reviewed Sep 13, 2019

View reviewed changes

physics/module_sf_noahmp_glacier.f90 Show resolved Hide resolved

climbfuji reviewed Sep 13, 2019

View reviewed changes

climbfuji approved these changes Sep 13, 2019

View reviewed changes

grantfirl mentioned this pull request Sep 13, 2019

NoahMP uses internal constants rather than those provided from host #318

Closed

grantfirl mentioned this pull request Sep 13, 2019

Move calculation of sncovr from FV3/io/FV3GFS_io.F90 to ccpp/physics/physics/GFS_phys_time_vary.XYZ.F90 #315

Closed

grantfirl removed do not merge work in progress labels Sep 13, 2019

climbfuji mentioned this pull request Sep 18, 2019

gmtb/develop: update from master, MYJ PBL/SFC, Noah MP #320

Merged

climbfuji merged commit 69c215f into NCAR:gmtb/develop Sep 19, 2019

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

NoahMP CCPP-compliancy #305

NoahMP CCPP-compliancy #305

grantfirl commented Aug 24, 2019 •

edited

Loading

grantfirl commented Aug 24, 2019

grantfirl commented Aug 24, 2019

grantfirl Aug 26, 2019

climbfuji Aug 26, 2019

grantfirl Aug 26, 2019

climbfuji Sep 13, 2019

climbfuji Sep 13, 2019

climbfuji Sep 13, 2019

grantfirl Sep 13, 2019

climbfuji Sep 13, 2019

climbfuji left a comment

grantfirl commented Sep 13, 2019

NoahMP CCPP-compliancy #305

NoahMP CCPP-compliancy #305

Conversation

grantfirl commented Aug 24, 2019 • edited Loading

grantfirl commented Aug 24, 2019

grantfirl commented Aug 24, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

climbfuji left a comment

Choose a reason for hiding this comment

grantfirl commented Sep 13, 2019

grantfirl commented Aug 24, 2019 •

edited

Loading