gsd/develop: update from from develop 2020/01/27

[climbfuji]

Update gsd/develop from develop:

• inline post changes
• GFSv16 changes
• cmake build system changes
• updates from dtc/develop that were merged to develop previously
• replace GFS_v15 and GFS_v15plus tests with GFS_v15p2 and GFS_v16beta tests

For regression testing information, see below.

[climbfuji]

Associated PRs:
#7
NOAA-GSL/fv3atm#9
NOAA-GSL/GFDL_atmos_cubed_sphere#1
NOAA-GSL/ccpp-physics#9
NOAA-GSL/ccpp-framework#2

[climbfuji]

@tanyasmirnova @haiqinli @joeolson42 can you please have a look at this PR and associated PRs? I will update the regression testing info in a few minutes (all looking good). Thanks!

[tanyasmirnova]

Dom,
I think the last part of the following code in the GFS_MP_generic.F90 could be also done for RUC in case dtf /= dtp:
if (lsm==lsm_ruc) then
if (imp_physics == imp_physics_gfdl .or. imp_physics == imp_physics_thompson) then

  if (lsm==lsm_ruc .or. lsm==lsm_noahmp) then
        raincprv(:)   = rainc(:)
        rainncprv(:)  = frain * rain1(:)
        iceprv(:)     = ice(:)
        snowprv(:)    = snow(:)
        graupelprv(:) = graupel(:)
    !for NoahMP, calculate precipitation rates from liquid water equivalent thickness for use in next time step
    !Note (GJF): Precipitation LWE thicknesses are multiplied by the frain factor, and are thus on the dynamics time step, but the conversion as written
    !            (with dtp in the denominator) assumes the rate is calculated on the physics time step. This only works as expected when dtf=dtp (i.e. when frain=1).
    if (lsm == lsm_noahmp) then
      tem = 1.0 / (dtp*con_p001) !GJF: This conversion was taken from GFS_physics_driver.F90, but should denominator also have the frain factor?
      draincprv(:)   = tem * raincprv(:)
      drainncprv(:)  = tem * rainncprv(:)
      dsnowprv(:)    = tem * snowprv(:)
      dgraupelprv(:) = tem * graupelprv(:)
      diceprv(:)     = tem * iceprv(:)
    end if
  end if

What do you think?
Thanks,
Tanya

[DomHeinzeller]

Ok, this will have to be a follow-up change. Can you create an issue for it, please?

…

On Jan 29, 2020, at 11:49 AM, tanyasmirnova ***@***.***> wrote: Dom, I think the last part of the following code in the GFS_MP_generic.F90 could be also done for RUC in case dtf /= dtp: if (lsm==lsm_ruc) then if (imp_physics == imp_physics_gfdl .or. imp_physics == imp_physics_thompson) then if (lsm==lsm_ruc .or. lsm==lsm_noahmp) then raincprv(:) = rainc(:) rainncprv(:) = frain * rain1(:) iceprv(:) = ice(:) snowprv(:) = snow(:) graupelprv(:) = graupel(:) !for NoahMP, calculate precipitation rates from liquid water equivalent thickness for use in next time step !Note (GJF): Precipitation LWE thicknesses are multiplied by the frain factor, and are thus on the dynamics time step, but the conversion as written ! (with dtp in the denominator) assumes the rate is calculated on the physics time step. This only works as expected when dtf=dtp (i.e. when frain=1). if (lsm == lsm_noahmp) then tem = 1.0 / (dtp*con_p001) !GJF: This conversion was taken from GFS_physics_driver.F90, but should denominator also have the frain factor? draincprv(:) = tem * raincprv(:) drainncprv(:) = tem * rainncprv(:) dsnowprv(:) = tem * snowprv(:) dgraupelprv(:) = tem * graupelprv(:) diceprv(:) = tem * iceprv(:) end if end if What do you think? Thanks, Tanya — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#7>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AN7FF5AC2RHIM6NV756WAA3RAHFVNANCNFSM4KMGSZ4Q>.

[tanyasmirnova]

Dom,
I think I made a new issue for this.
Thanks,
Tanya

[DomHeinzeller]

Dom,
I think I made a new issue for this.
Thanks,
Tanya

Thank you, you did. I just transferred it from ufs-weather-model to ccpp-physics: NOAA-GSL/ccpp-physics#11

[climbfuji]

Regression testing on hera.intel: run IPD PROD tests in rt.conf against EMC baseline - all pass.

rt_ipd_prod_only.log

Next, create full baseline for rt.conf (IPD PROD, IPD REPRO, CCPP PROD) and verify against it (IPD PROD, IPD REPRO, CCPPREPRO, CCPP PROD) - all pass.

rt_full_create.log
rt_full_verify.log

No need move the full baseline to the official DTC baseline directory, this was done by NCAR#20.

[climbfuji]

Regression testing on hera.intel for rt_ccpp_dtc.conf and rt_ccpp_gsd.conf, rt_ccpp_caps.conf (first create, then verify): all pass.

rt_ccpp_dtc_create.log
rt_ccpp_dtc_verify.log
rt_ccpp_gsd_create.log
rt_ccpp_gsd_verify.log

[climbfuji]

This PR is ready to merge.

[DomHeinzeller]

Reviewed and approved by @haiqinli and @tanyasmirnova. All tests pass.