Coupling of RRTMGP to Thompson MP (plus improvements) #528
Conversation
…ster-ncar-f3e6761-addThompsonMP
dustinswales
requested review from
climbfuji,
grantfirl and
JulieSchramm
as code owners
|
We definitely need the bug fix in the RRTMGP interstitial (and perhaps the namelist-controlled improvements for the GFSv16beta+RRTMGP suite) before the RRTMGP test can start. |
|
@ligiabernardet agreed - answers will be just wrong otherwise. |
There was a problem hiding this comment.
This looks all good, just need to watch out for variables that are declared as intent(out). If you return from a subroutine early and don't set those variables beforehand, then they are intent(inout) (in both the Fortran source code and the metadata file).
I am going to add a couple of GSL folks as reviewers for this PR.
| !> - LW time-step adjustment: | ||
| if (use_LW_Jacobian) then |
There was a problem hiding this comment.
Since there are no changes to the adjsfculw_{lnd,ice,ocn} components I assume that this capability is not (yet) compatible with the fractional grid?
There was a problem hiding this comment.
@climbfuji I'm adding @RobertPincus here to keep me honest.
In GP the LW fluxes, and Jacobians of the LW, take into account the surface type, so there is no need for the LW flux adjustment between radiation calls to distinguish by surface type.
This is different in RRTMG, where it's slightly more complicated. The LW flux adjustments rely on surface temperatures from the lsm between rad calls, which distinguishes by surface type, adding this dependence.
As far as I can tell, all the frac_grid option does is combine the "by surface type" LW flux adjustments for mixed ocean/land/ice grid-points. This is not possible in GP, and I'm not sure we would want to do it? This seems to be inconsistent w/ how the surface is treated in the LW flux in the radiation schemes.
There was a problem hiding this comment.
| @@ -293,55 +293,6 @@ subroutine sgscloud_radpre_run( & | |||
|
|
|||
| endif ! timestep > 1 | |||
|
|
|||
| !> - Compute SFC/low/middle/high cloud top pressure for each cloud domain for given latitude. | |||
There was a problem hiding this comment.
Can you add a comment in the general description of your changes that you are removing (seemingly) redundant code from module_SGSCloud_RadPre.F90? Removing the ptop and gethml call seems OK, which we basically do the same thing later in progclduni. I'm a bit hesitant to give a thumbs up until I see a test run showing no change in the output (I'll queue up a test quick). I'm not certain of the impacts of alpha_dummy...
There was a problem hiding this comment.
@joeolson42
I added this to the description.
As far as I can tell, alpha_dummy was there just to avoid an error when calling gethml(), which is called later. After the cloud-MP has been configured, just prior to calling the radiation.
There was a problem hiding this comment.
That's right, I had to put that alpha_dummy parameter there to keep the code working after gethml() as updated in an earlier PR. It doesn't affect the cloud overlap calculations that are used in the GSL suites.
There was a problem hiding this comment.
Dustin, do you expect that the removal of this code will not change our cloud cover fields? I expect this to be the case. I'll approve this PR (conditionally) upon the regression tests showing that the removal of this code does not change our cloud cover fields.
There was a problem hiding this comment.
I ran the RTs on Hera. None of these changes impact the other baselines.
|
@dustinswales Have you tested if your changes lead to different answers for some of the physics suites? If so, which and do you know why? |
The only source code change that changes the results is the correction of the bug with the GP interstitials intents. |
|
@dustinswales just one clarification about my The code generator will skip any operation that is not necessary for performance reasons. This means that if a variable is declared Thus, if the temporary variable isn't set inside the routine, uninitialized data will be assigned to the original variable. |
Thanks, that is good to know. |
|
Dom, Do you want to let physics group know the impact of this PR to GFSv16
physics test?
…On Wed, Dec 16, 2020 at 4:06 PM Dom Heinzeller ***@***.***> wrote:
@dustinswales <https://github.com/dustinswales> Have you tested if your
changes lead to different answers for some of the physics suites? If so,
which and do you know why?
The only source code change that changes the results is the correction of
the bug with the GP interstitials intents.
There are new features controlled by the namelist that we can turn on for
v16 physics which will change the results. As it stands in the UFS PR, I
didn't turn these things on for the v16 RTs.
Thanks, that is good to know.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#528 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AI7D6TI2BNYLWH76II6KQG3SVEOOXANCNFSM4USUBMJA>
.
|
If I read Dustin's reply correctly, there are no changes to the results unless someone wants to turn on these new options (all off by default). I assume that these will be tested as part of the UFS R2O project and that these have been discussed and designed in accordance with the team. |
This is good to know. |
This is correct. These new options will be turned on for the GSL physics suite enabled w/ GP, for which the baselines need to be created. |
|
Sorry, maybe I misunderstood, I saw the following lines in
ufs-weather-model PR#323: " Answers to the v16beta baselines using RRTMGP
need to be updated ". Would you please add comments on which RT tests will
change results and which ones will not? Thanks
…On Wed, Dec 16, 2020 at 4:18 PM dustinswales ***@***.***> wrote:
Dom, Do you want to let physics group know the impact of this PR to GFSv16
physics test?
… <#m_-8057787860451177830_>
If I read Dustin's reply correctly, there are no changes to the results
unless someone wants to turn on these new options (all off by default). I
assume that these will be tested as part of the UFS R2O project and that
these have been discussed and designed in accordance with the team.
This is correct. These new options will be turned on for the GSL physics
suite enabled w/ GP, for which the baselines need to be created.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#528 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AI7D6TNNQWDX35TCYX4TRNTSVEPYTANCNFSM4USUBMJA>
.
|
There are 4 baseline with RRTMGP that need updating to reflect the bugfix. |
| effrin_cldliq(1:nCol,1:nLev) = re_cloud(1:nCol,1:nLev)*1.e6 | ||
| effrin_cldice(1:nCol,1:nLev) = re_ice(1:nCol,1:nLev)*1.e6 | ||
| effrin_cldsnow(1:nCol,1:nLev) = re_snow(1:nCol,1:nLev)*1.e6 | ||
|
|
There was a problem hiding this comment.
@DomHeinzeller Do we need the units conversion here? Or ccpp will take care of it.
There was a problem hiding this comment.
@climbfuji @dustinswales I got a couple comments on this PR, but is see the status Pending on them and no answers.
There was a problem hiding this comment.
@tanyasmirnova I'm just seeing these comments now. This PR was merged this morning.
| cld_rerain(1:nCol,1:nLev) = rerain_def | ||
|
|
||
| ! Compute cloud-fraction. Else, use value provided | ||
| if(.not. do_mynnedmf .or. imfdeepcnv .ne. imfdeepcnv_gf ) then ! MYNN PBL or GF conv |
There was a problem hiding this comment.
@climbfuji @dustinswales The code that follows this IF statement is for cases when MYNN PBL and GF convective schemes are not used. How do the MYNN and GF sub-grid clouds get into cld_frac? I guess it is done in a different locations, right?
There was a problem hiding this comment.
@tanyasmirnova
The treatment of sub-grid clouds in GP is the same as was in RRTMG, just organized differently.
The subgrid-scale clouds are added to the cloud fields in module_SGSCloud_RadPre.F90, called before preparing the MP inputs for radiation in both radiation schemes.
module_SGSCloud_RadPre_run() starts by computing a Xu-Randall cloud-fraction, then if there are BL clouds (do_mynnedmf), adjust the cloud-fraction to incorporate the BL clouds. Then add the convective clouds (imfdeepcnv == imfdeepcnv_gf) if they are requested.
So in GFS_rrtmgp_thompsonmp_pre.F90 we compute cloud-fraction only if it hasn't been created before.
There was a problem hiding this comment.
These changes for some reason show up as Clara's, but I think this is from merging gsl/develop with all the cloud-radiation interaction updates into master. We need to add you as reviewer for all PRs that touch RRTMG code, I fear.
There was a problem hiding this comment.
Yeah that's fine. I thought it was weird that Clara was adding to the mp-radiation coupling...
There was a problem hiding this comment.
That doesn't look at all familiar to me, so if it is mine, I have a very short memory.
| ! | ||
| precip_overlap_param = cloud_overlap_param | ||
| ! Cloud-fraction | ||
| cld_frac(1:nCol,1:nLev) = tracer(1:nCol,1:nLev,i_cldtot) |
There was a problem hiding this comment.
@dustinswales In the GFS_rrtmg_pre.F90 there is an option to use MYNN cloud fraction under certain conditions. I do not see this option here.
There was a problem hiding this comment.
@tanyasmirnova
Yes there is. I see this starting at lines 718. Weird. When did that get in there?
I went back and see that this was added to the GFDL-RRTMG coupling on July 20th, 33102ba#diff-135feb492e7ec0ea9d140b82431a8b8eda48acb4eab4ed783ca819e933476475.
@ClaraDraper-NOAA and @climbfuji
Can it be that something got added to the coupling of GFDLMP and RRTMG that needs to be propagated into the coupling of GFDLMP to GP?
* Feature/ccpp codeowners (NCAR#442) * Per-file CODEOWNERS in ccpp/physics to set up automatic review requests * Code cleanup. Remove used code/variables. Fix minor inconsistencies. (NCAR#440) * Remove ESMF Alarm and TimeInterval variables from module_fv3_config. * Variables nfhmax, nfhmax_hf are unused outside InitializeAdvertise. Declared them as local variables. * There is no need to keep duplicates of all time related variables in atm_int_state when we can easily access them from atm_int_state%Atm. * Remove redundant call to fms_init * Add few missing ESMF_LogFoundError checks in module_fcst_grid_comp.F90 * Delete time_utils.F90. Unused. * print only actual errors to stderr, everything else to stdout * Move realizeConnectedCplFields to module_cplfields from module_cap_cpl * Declare Atmos as module variable, and remove atmos_internalstate_wrapper * Move code from clock_cplIntval to InitializeAdvertise * Removed INTERNAL_FILE_NML from atmos_model.F90 * CCPP cloud cover change for Thompson MP associated with Xu Randall (NCAR#443) * Remove old comments from GFS_typedefs.F90. * Update logic that sets Model%cnvcld = .false. * Wrapper for ccpp-physics NCAR#806, NCAR#807, NCAR#813 (NCAR#447) * CCPP physics updates from PR NCAR#806(only diagnostic changes for RRTMGP - small diag-only RT impact for tests involving RRTMGP),NCAR#807 (only aborting model if sfc emis file not present when needed) ,NCAR#813(only SCM-specific physics changes) * Wrapper for ccpp-physics NCAR#808 and 816 (roughness length over ice and NoahMP tsurf bugfix) (NCAR#452) This PR contains the ccpp physics PR NCAR#808 and NCAR#816. 808 addresses an error in the momentum roughness length over tiles with ice. 816 fixes an occasional segfault bug related to the tsurf variable in NoahMP and updates to "improve snow simulation in NoahMP for P8". * Use 'model set run clock' routine in FV3 NUOPC cap. (NCAR#450) * Add 'SetRunClock' specialization routine to FV3 NUOPC cap, which sets fv3 model clock. FV3 cap does not keep local copy of clock (clock_fv3) as saved module variable anymore. * Minor code cleanup. * Clean up iovr=4 (exponential cloud overlapping method) in RRTMG (NCAR#445) Clean up the exponential cloud overlapping method in RRTMG, which reflects the practice of the pre-2018 operational HWRF model. * Bug fix for dimensions of eta level variables and WAM variables in Fortran code and CCPP metadata (NCAR#431) * Add additional diagnostic arrays for radiation-microphysics development * Bug fixes for WAM model runs with levr < levs * Update inline post with latest UPP release upp_v10.0.11 (NCAR#449) * Update UPP revision * Add foundation temperaure in GFS read interface for inline post. * Wrapper for ccpp-physics NCAR#812 (NCAR#453) * update submodule pointer for regression testing of ccpp-physics#812 * MYNN sfclay (RAP suite) restart reproducibility, P8 suite definition files (NCAR#455) * Fix uninitialized variable zmtnblck in ccpp/data/GFS_typedefs.F90 * Fix typo in CCPP standard name for ncnvwind in ccpp/data/GFS_typedefs.meta * Add ten 2d variables required for MYNNSFC restart reproducibility to ccpp/driver/GFS_restart.F90 * create initial p8 suites, P8 initial SDFs as copies of FV3_GFS_v16_coupled_nsstNoahmpUGWPv1 and FV3_GFS_v16_nsstNoahmpUGWPv1 Co-authored-by: Denise Worthen <[email protected]> * Fixes on initializing snow depth over ice and changes z0ice (NCAR#461) * modify FV3GFS_io.F90 by fixing errors associated with initializing snow depth over ice in the case where both land and water coexist (i.e. fractional grid case) * z0ice is changed to 1.0 cm from 1.1cm in atmos_model.F90 * Radar-derived microphysics temperature tendencies similar to operational HRRR (NCAR#457) - implements a feature of the operational HRRR, radar-derived microphysics temperature tendencies applied in the first N minutes of the forecast to improve clouds in the first few hours. * HRRR-like radar-derived temperature tendencies * Give a warning when convection is enabled with radar tten * Fix uninitialized variable zmtnblck in ccpp/data/GFS_typedefs.F90 * Add ten 2d variables required for MYNNSFC restart reproducibility to ccpp/driver/GFS_restart.F90 * fixing snod bug in atmos_model.F90 (NCAR#465) * Thompson MP cloud tuning (NCAR#463) * Improve cloud fraction when using Thompson MP. See NCAR#809 for more details. * Feature/hwrf legacy (NCAR#459) * Added a new suite definition file: suite_FV3_HAFS_v0_thompson_tedmf_gfdlsf.xml. This SDF is a legacy hwrf SDF but only the GFDL surface layer scheme and the Thompson scheme are kept. * CCPP: Update consistency checks and effective radii calculation for Thompson MP (NCAR#462) * This PR updates the submodule pointer for ccpp-physics for the changes described in NCAR#786 (Update consistency checks and effective radii calculation for Thompson MP) * Unified write_netcdf, add single file 'cubed_sphere_grid' output (NCAR#466) * Unify write_netcdf and write_netcdf_parallel modules. * Add support for writing 'cubed_sphere_grid' history files in a single netcdf file * Code refactoring and cleanup. * Change standard name and unit of CCPP error flag variable in CCPP framework and physics (NCAR#467) *Change standard name and unit of CCPP error flag variable in CCPP framework and physics. * Add code to enable ESMF managed threading (NCAR#469) * Implement ESMF-managed-threading for fcstComp and wrtComp's. * Revisions to repair iovr=5 cloud overlap option (NCAR#471) *This PR addresses part 2 of CCPP issue NCAR#748 to activate the exponential-random cloud overlap method (iovr=5) in RRTMG. * Add 2d decomposition on the write grid comp (NCAR#470) * update write grid comp to have 2D decomposition on output grid. * combine PR#468: Bug fix in FV3GFS_io.F90 for allocation of temp2d Co-authored-by: Ted Mansell <[email protected]> * 4DIAU bug when iau_filter_increments=T (NCAR#458) * Add lsm_cold_start variable for RUC LSM SCM support and gwdps bugfix (combined) (NCAR#475) * add lsm_cold_start variable in GFS_typedefs.F90/meta and update ccpp/physics for testing * includes @SMoorthi-emc 's bugfix for gwdps.f. * add lon info in the write group (NCAR#476) * Multiple output grids (NCAR#480) Update fv3 cap and write grid component to enable outputting multiple domains. This is done be creating an array of fcstGrids, and array of rout handles where each element of these arrays correspond to one atm domain. In the write grid component updates were made to allow grid spec parameters for each output grid to be specified separately. Co-authored-by: Gerhard Theurich <[email protected]> * GPU-enabled version of Grell-Freitas convection in ccpp-physics (NCAR#479) * Enable Thompson MP when coupling with UFS-Aerosols (NCAR#484) * Add support for Stochastically Perturbed Parameterizations (SPP) in FV3 and add the FV3_RRFS_v1alpha SDF. (NCAR#454) Adds the necessary code in fv3atm to allow for Stochastically Perturbed Parameterizations (SPP) in a set of RAP/HRRR-based physics parameterizations. Specific to the fv3atm repository, code in this PR defines the necessary variables associated with various SPP-related fields (e.g., logical to activate SPP, parameterization-specific SPP variables, etc.) that are then passed to ccpp-physics. * lateral boundary fix for regional runs (NCAR#482) * Feature: NSSL Microphysics support (NCAR#472) Supports CCPP with new NSSL microphysics scheme, which has separate hydrometeor categories for graupel and hail (Mansell et al. 2010, JAS). This PR allows fv3atm to compile correctly with CCPP+NSSL, but a further update to atmos_cubed_sphere is needed to run correctly. The NSSL scheme is used in NSSL's daily FV3 convection-allowing forecasts, and the scheme itself is aimed at this scale (dx <= 4km). Testing is underway also with a preliminary Warn-on-Forecast FV3 ensemble. * Add p8b SDF (NCAR#491) * add p8 suite files * fix 2phases intermediate restart Co-authored-by: Raffaele Montuoro <[email protected]> Co-authored-by: Jun Wang <[email protected]> * Split CCPP finalize into physics_finalize and (framework) finalize; minor CCPP cleanup and documentation updates (NCAR#487) * Minor cleanup and documentation updates for CCPP; split CCPP finalize into physics_finalize and (framework) finalize * Add 'valid time' variable using ISO string format (NCAR#492) Add 'valid time' variable using ISO string format to netcdf history files. Use double precision variable to set value of 'time' attribute in wrt comp import state Update ccpp/physics (setting surface-related interstitial variables for SCM prescribed-surface-flux mode) Update inline_post_stub.F90 subroutine interfaces to match inline_post.F90 Co-authored-by: Ted Mansell <[email protected]> Co-authored-by: Grant Firl <[email protected]> * Fix Lambert Conformal Conic mapping in the write component on the southern hemisphere (NCAR#497) Subroutine `lambert` in the write component has been fixed to do the mapping on the southern hemisphere correctly. * Rectify total number of Thompson MP species for PBL mixing (NCAR#509) * update upp revision (NCAR#505) * move some parameters in mynn surface layer to namelist options * land surface upgrades for p8c (NCAR#499) * Added capability for cdmbgwd(1) to scale GSL blocking drag (NCAR#502) * release/p8c: it is related to ccpp-physics PBL and cumulus updates plus a bug fix to the Thompson scheme PR NCAR#882 (NCAR#503) * change ccpp-physics branch to feature/p8c_pbl_cp branch * change ccpp/physics in feature/p8c_pbl_cp branch * revised GFS_typedefs.F90 for the paramters in the saSAS scheme * sync upp * Revert "sync upp" This reverts commit 429d0469c92d200fa058aca6c6bd17adac7f9568. * sync upp to 0185bb9 * Merge remote-tracking branch remote/main for ccpp-physics and update .gitmodules * Bring AQM model into UFS (NCAR#511) * Reset to zero coupling arrays for accumulated snow, large scale rain, and convective rain at the end of each coupling step if coupling with chemistry model. * Properly set kind type of literal constants defining zero and one. * Initialize to zero canopy resistance output variablein noah/osu land-surface model subdriver. * Re-implement radiation diagnostic output involving spectral band layer cloud optical depths (0.55 and 10 mu channels) to prevent floating invalid errors due to uninitialized optical depth arrays. * Temporarily disable filling export fields during the NUOPC Realize phase since it breaks coupling with aerosol component. * Increase maximum number of input aerosol scavenging factors to accommodate AQM/CMAQ 5.2.1 chemical tracers. * Remove inst_pres_height_surface from chemistryFieldNames as its imported already elsewhere Co-authored-by: Raffaele Montuoro <[email protected]> * RRTMGP for prototype 8 (NCAR#498) * Add explict treatment of convective cloud to RRTMGP. * Some cleanup and bug fixes from previous commit. working now with Thompson MP plus radiatively active convective cloud. * Updated physics. New interstitials. * Added explicit MYNN-EDMF cloud coupling to RRTMGP. * Added switches for sgs clouds in GP. * Replaced cld_mynn_ naming convention with cld_pbl_ * Fixed inconsistency between G/GP in Thompson MP. * Some more cleanup of cloud-fraction... * Pulled in Qingfu's changes to ccpp-physics * Synced physics with main * Updated physics. * Revert change from previous commits (sampling of different cloud types). New SDFs for p8c * Changed arguments to implied shape. * Renamed scheme file in SDF. Now common to both RRTMG and RRTMGP SDFs * Syntax error in ccpp-physics * Revert changes to names * Renamed p8_rrtmgp SDFs * Added bounding to temperature at layer-interface used by RRTMGP. * Reorder loop * Combined gp sw and lw aerosol routines. Modest speedup (~4%) * HAFS related moving-nesting, nest coupling, and movable output grid capabilities (NCAR#501) * Add the storm-following or specified moving-nesting capability for both the regional and global configuration. * Expand the UFS/HAFS FV3ATM-HYCOM ocean coupling to support the static and moving nesting configuration. * Expand the write grid component to support outputting the moving nest domain properly. * Five HAFS related moving-nesting regression tests are added to test these newly added features. The regional_hafs RT was removed. Co-authored-by: Dom Heinzeller <[email protected]> Co-authored-by: Dom Heinzeller <[email protected]> Co-authored-by: Grant Firl <[email protected]> Co-authored-by: Daniel Rosen <[email protected]> Co-authored-by: man.zhang <[email protected]> Co-authored-by: William Ramstrom <[email protected]> Co-authored-by: Ufuk Turuncoglu <[email protected]> Co-authored-by: Dusan Jovic <[email protected]> Co-authored-by: A-Kyle <[email protected]> Co-authored-by: Gerhard Theurich <[email protected]> Co-authored-by: Dusan Jovic <[email protected]> * Split GFS_typedefs.F90, rename CCPP_interstitial, import ccpp_t from ccpp_types (NCAR#508) * Split GFS_typedefs.F90: move GFS_interstitial_type to CCPP_typedefs.F90; Rename CCPP_interstitial to GFDL_interstitial; Import ccpp_t from ccpp_types instead of ccpp_api * Add bugfix in ccpp/data/GFS_typedefs.meta: mark input_nml_file as active only if the number of lines in the file is greater than zero * Fix uninitialized min_rand variable in Thompson MP when using SPP (NCAR#516) * Pass SPP namelist entries outside of stochastic physics routines. * Fixes to SPP metadata * Add active attribute to spp_prt_list * Add active metadata flag to spp_stddev_cutoff * Add active metadata flag for spp_var_list * A bug fix for wet scavenging of aerosols and a minor modification for moisture property calculation (NCAR#520) * Update ccpp_prebuild_config.py for one-to-one CCPP physics schemes/files (NCAR#519) * point to gsl/merge-develop branch * Point to gsl/merge-develop-to-community branch for ccpp/physics * Corrections to merge + joe updates * Point .gitmodules to Sam's fork * Correct issues found by reviewers * Rename mynnpbl to mynnedmf * Remove some empty _init and _finalize routines and update a comment in sgscloud_radpre * Feature/lndp noahmp2 : add land perturbation scheme for Noah-MP fractional veg (NCAR#513) Updates to allow land perturbation scheme to be applied for Noah-MP. Main code changes: -added lsm_noahmp model option to lndp_type==2 land perturbation scheme (fv3-atm, stochastic_physics) -updated comments -cleaned up the namelists to make the different options clearer. Removed misleading lndp_each_step variable from gfs_physics_nml, and replaced it with "lndp_model_type" in the nam_sfcperts namelist for the different forecast types (cycling DA, short forecasts, perturbing only initial conditions) (fv3-atm, stochastic_physics, ufs-weather-model) NOTE: see note below, re: specification of smc perturbation for RAP/HRRR/etc (lndp_model_type==2) -revised code in stochastic_physics_wrapper to only allocate arrays that will be used when passed into lndp_apply_perts (fv3-atm) -deleted unused albedo arrays in stochastic_physics_wrapper and lndp_apply_perts (fv3-atm, stochastic_physics) -fixed bug causing precision errors when calculating the soil moisture ice content (often ~0.0; stochastic_physics) -fixed bug in which lndp_apply_perts wasn't called for nscyc = 0. (fv3-atm) -increased max_nvar_lndp at Jeff Ator's request (fv3-atm) -added a new test for the Noah-MP and lndp==2 combination (ufs-weather-model) * Add \file to smoke files * Add hailcast code in atmos_cubed_sphere (NCAR#528) * Combine PROD and REPRO build options into 'Release' build type (NCAR#524) Combined PROD and REPRO build modes into 'Release' build type. Now only 'Debug' and 'Release' build type are supported. Both build types must produce bit-for-bit reproducible outputs using different number of threads, mpi tasks, different domain decomposition, have reproducible restarts etc. * Unify inline post (NCAR#521) * Unify global and regional inline posts and add bug fix for dx/dy computation. * Update upp revision to dc8bc68 * Set up different configurations of read max/min 2m T for global and regional FV3. * adding bug fix from jili dong and removing U* averaging * adding bug fix from jili dong and removing U* averaging * Remove some commented-out code from cu_gf_driver in ccpp-physics. * Turn rrtmg smoke band 10 into a model namelist variable * Switch ccpp-physics smoke arguments to implied shape * Remove some stops and gotos from ccpp-physics smoke * Remove some hard-coded constants and rename some meta entries * Change RRTMGP to RRTMG in two P8 SDFs (NCAR#527) * Change RRTMGP to RRTMG in suite_FV3_GFS_v17_p8 and suite_FV3_GFS_v17_coupled_p8 * deleted or modified some SDFs related to RRTMGP or Thompson schemes * added a new SDF file for P8 with rrtmgp * Rename some standard_names, change a unit, and remove a "use physcons" * NSSL ccpp-physics bugfixes and new ccpp-framework debugging feature (NCAR#529) * Set num3d correctly when using radar_tten * Only add cnv_3d_ud_mf to the restart files when needed * Pass some chemistry varibles to mynn_bl_driver correctly * Correct argument passing within mynn edmf * point to NCAR main ccpp/physics * merge gsl institutional fork (NCAR#526) Merge in GSL institutional fork * point to sam's repo * ccpp/physics: Change to GSL CODEOWNERS * Returned the comments on soil resistance back. * point to gsl/develop for ccpp/physics Co-authored-by: Dusan Jovic <[email protected]> Co-authored-by: Dom Heinzeller <[email protected]> Co-authored-by: Grant Firl <[email protected]> Co-authored-by: mzhangw <[email protected]> Co-authored-by: WenMeng-NOAA <[email protected]> Co-authored-by: Denise Worthen <[email protected]> Co-authored-by: SMoorthi-emc <[email protected]> Co-authored-by: ChunxiZhang-NOAA <[email protected]> Co-authored-by: Jun Wang <[email protected]> Co-authored-by: Ted Mansell <[email protected]> Co-authored-by: Jeff Whitaker <[email protected]> Co-authored-by: Gerhard Theurich <[email protected]> Co-authored-by: DomHeinzeller <[email protected]> Co-authored-by: Raffaele Montuoro <[email protected]> Co-authored-by: JeffBeck-NOAA <[email protected]> Co-authored-by: MatthewPyle-NOAA <[email protected]> Co-authored-by: Ted Mansell <[email protected]> Co-authored-by: Jessica Meixner <[email protected]> Co-authored-by: Jun Wang <[email protected]> Co-authored-by: joeolson42 <[email protected]> Co-authored-by: HelinWei-NOAA <[email protected]> Co-authored-by: mdtoyNOAA <[email protected]> Co-authored-by: Brian Curtis <[email protected]> Co-authored-by: dustinswales <[email protected]> Co-authored-by: Bin Liu <[email protected]> Co-authored-by: Dom Heinzeller <[email protected]> Co-authored-by: Grant Firl <[email protected]> Co-authored-by: Daniel Rosen <[email protected]> Co-authored-by: man.zhang <[email protected]> Co-authored-by: William Ramstrom <[email protected]> Co-authored-by: Ufuk Turuncoglu <[email protected]> Co-authored-by: Dusan Jovic <[email protected]> Co-authored-by: A-Kyle <[email protected]> Co-authored-by: ClaraDraper-NOAA <[email protected]> Co-authored-by: BinLi-NOAA <[email protected]>
This PR contains several advancements to the existing v16beta coupling and introduces a new coupling for RRTMGP and the Thompson MP scheme.
For RRTMGP enabled v16beta physics there are several new (optional) capabilities, controlled via physics namelist:
-) Include scattering in LW all-sky rte calculation (doGP_lwscat = .true.)
-) Use of internal RRTMGP cloud-optics (doGP_cldoptics_LUT (default) or doGP_cldoptics_PADE (Pade Approximation used) = .true).
-) Adjust surface upwelling LW surface fluxes between RRTMGP calls using the Jacobian of the upwelling LW flux (use_LW_jacobian = .true.). F_adj = F_o + dT*Jacobian, where Jacobian = dLW/dT upward at surface.
These new functionalities are also available in the RRTMGP coupling with the Thompson MP.
The coupling of GP to the Thompson MP is isolated to GFS_rrtmgp_thompsonmp_pre.F90 and based on the existing coupling to the RRTMG radiation scheme. A common function for the Xu-Randall prognostic cloud-fraction is included.
This PR also include a bugfix in the RRTMGP interstitial fields. The intent needs to be inout, not simply out, for routines in the radiation loop. Some fields are used in calls in between radiation calls and were getting corrupted with the intent(out).
Additionally, redundant code was removed from module_SGSCloud_Rad_Pre.F90. @joeolson42