RRTMGP in fv3atm #26
RRTMGP in fv3atm #26
Conversation
Henry Juang's changes for MULTI_GASES, incorporating variable R and Cp for virtual temperature calculations. Also adds an option for tau (used in Rayleigh damping) to be negative, which allows for the GSM calculation of the time constant. Roughly, FV3 tau=0.022 is GSM tau=-10 (10-day time constant in GSM). Change-Id: I59edc57e8c0ec04390bc18158a96798e6c34f5a8
ccpp_prebuild version of multi-suite static build, cleanup validation of SDF used
…lop_merge_in_master_20190419
…ast physics through CCPP
…hub_update_vlab_master_20190417
… sncovr to match documentation
…90419 gsd/develop: bugfix for sncovr (compilation error)
… calculate Sfcprop%sr for GFDL-MP in PROD+TRANSITION mode
…b_master_20190417_BASED_ON_static_build_xml_validation_20190326 GitHub CCPP update with VLab master changes 20190419
…cumulated diagnostic fields and #61740 Incorrect units in calculation of Sfcprop%srflag for GFDL-MP; adapted for CCPP version of the code
…S_layer/GFS_restart.F90
…e variable for calculation of sr (Sfcprop%sr) to achieve b4b-identical results in PROD+TRANSITION mode
… fhgoc3d in metadata table and in comments
…_fhgoc3d gsd/develop: fix units fhzero fhcyc fhgoc3d
…b_master_20190429 GitHub CCPP update to VLab master 2019/04/29
…tests_20190502 Update/cleanup gsd/develop: rename Thompson MP
…-precision constant in TRANSITION mode to avoid b4b mismatches of Sfcprop%sr in fv3_streched_nest versus fv3_ccpp_stretched_nest regression tests
…mtb-ccpp_20190503 gsd/develop merge in gmtb/ccpp 2019/05/03
…mtb-ccpp_20190503 gsd/develop merge in gmtb/ccpp 2019/05/03
…gsd-develop_20190506 Merge gsd/develop into gmtb/ccpp 2019/05/06
… changes in results if current default non-frac landmask is used. CCPP modifications by @grantfirl and @climbfuji
…ice to replicate functionality in GFS_physics_driver
…r GFS_surface_composites_{pre,post} schemes, comment out debugging print statements
…b_master_20190507 Fractional landmask changes in GFS physics
…specific namelist sections from general gfs_physics namelist section, set MYNN EDMF namelist options automatically, depending on do_mynnedmf
…hub_cleanup_20190509
dustinswales
requested review from
climbfuji,
grantfirl and
llpcarson
as code owners
| @@ -5908,6 +6071,52 @@ subroutine interstitial_create (Interstitial, IM, Model) | |||
| allocate (Interstitial%zorl_land (IM)) | |||
| allocate (Interstitial%zorl_ocean (IM)) | |||
| allocate (Interstitial%zt1d (IM)) | |||
| ! RRTMGP | |||
There was a problem hiding this comment.
If these arrays are used exclusively by RRTMGP, they should only be allocated if RRTMGP is used. Need to make sure that if these arrays get passed into subroutines that are also used by other physics (without using the arrays inside, then its dimensions must be specified as assumed-size in the intent-declaration of the subroutines).
| @@ -23,6 +23,11 @@ CPPDEFS += -DNEW_TAUCTMAX -DSMALL_PE -DNEMS_GSM -DINTERNAL_FILE_NML | |||
|
|
|||
There was a problem hiding this comment.
No changes should be required to this file?
There was a problem hiding this comment.
I wasn't able to get FV3 to build unless the rte-rrtmgp source files were added here. This was something @grantfirl and I came up with back in the fall to get it to build.
There was a problem hiding this comment.
I remember this now ... we need to find a better way to do this. In particular because the make build is phased out as we speak in favor of a cmake build, hence the proposed changes would be meaningless.
There was a problem hiding this comment.
Thanks for making the changes. Besides the makefile question, we still have to fix the conditional allocation of the RRTMGP arrays in GFS_typedefs.F90 (i.e. allocate only if RRTMGP is used).
Regarding the timeline for the commit. We are scheduled to send dtc/develop to EMC for inclusion in the authoritative branches mid of this month. I prefer to merge the RRTMGP PRs afterwards so that the current PR going to EMC will be more digestible. I guess we will need the time to find a better way for the RRTMGP types and for setting up the ufs-weather-model regression tests?
| @@ -23,6 +23,11 @@ CPPDEFS += -DNEW_TAUCTMAX -DSMALL_PE -DNEMS_GSM -DINTERNAL_FILE_NML | |||
|
|
|||
There was a problem hiding this comment.
I remember this now ... we need to find a better way to do this. In particular because the make build is phased out as we speak in favor of a cmake build, hence the proposed changes would be meaningless.
Thanks! Looking at this particular commit, I realized that there is no reset-to-zero of these arrays in GFS_typedefs.F90's interstitial_rad_reset call. Is this intended? |
|
Great! Now we are down to the typedefs issue. Let me look at this beginning of next week (huge note on my desk). |
RRTMGP in fv3atm (#26 - based on latest code)
|
This PR has been merged as part of #32, closing it. |
PBL tendencies were missing in two schemes; now fixed. Squashed commit of the following: * fix bugs found in pbl and ozone 3d diagnostic tendencies * remove debugging prints * implied shape arrays for five variables * more block labels in CCPP * make the logging less wordy * point to hash 238c84c of ccpp/physics
This PR contains changes to the fv3atm required for CCPP enabled with the RRTMGP radiation scheme: NCAR/ccpp-physics#400