Leverage changes from EMC to use updated stochastic physics by MingjingTong-NOAA · Pull Request #81 · NOAA-GFDL/SHiELD_physics

MingjingTong-NOAA · 2026-03-05T22:45:16Z

Description

This PR include changes made to use updated stochastic physics

Fixes # (issue)

A forecast test was run to compare with the main branch and the results are bit-wise identical.

Checklist:

Please check all whether they apply or not

[x ] My code follows the style guidelines of this project
[x ] I have performed a self-review of my own code
[] I have commented my code, particularly in hard-to-understand areas
[] I have made corresponding changes to the documentation
[x ] My changes generate no new warnings
[x ] Any dependent changes have been merged and published in downstream modules

lharris4

Thank you for submitting this code for review. I do believe that there is a lot of complexity added to the SHiELD physics driver for a mostly self-contained module, and in particular a number of options and variables could be moved within the stochastic_physics_wrapper module.

lharris4 · 2026-03-10T15:31:00Z

      print *, 'basic control parameters'
      print *, ' me                : ', Model%me
      print *, ' master            : ', Model%master
+      print *, ' communicator      : ', Model%communicator%mpi_val


I am concerned about adding MPI communicators to the physics, as much of the SHiELD physics doesn't manage its own communication. If this is specific to the stochastic physics, then it should be contained in that module. @bensonr can you comment on this?

I've removed MPI communicators from physics and changed stochastic physics code to define communicator inside the that module.

lharris4 · 2026-03-10T15:36:50Z


    real(kind=kind_phys) :: rbcr            !< Critical Richardson Number in the PBL scheme
    logical              :: mix_precip      !< Whether to apply PBL mixing to precipitating hydrometeors
+    logical              :: cap_evap        !< Whether to apply limiter to evaperation from latent heat flux


Thank you for explicitly adding this option

lharris4 · 2026-03-10T15:38:14Z

    integer              :: nca             !< number of independent cellular automata
    integer              :: nlives          !< cellular automata lifetime
    integer              :: ncells          !< cellular automata finer grid
+    integer              :: nca_g           !< number of independent cellular automata


Eventually the stochastic physics options should go in a separate stochastic_physics_nml

Current only three stochastic physics namelist variables are used. We can make this change when we explore other options of stochastic physics.

lharris4 · 2026-03-10T15:43:50Z

                                            !< ivegsrc = 2   => UMD  (13 category)
    integer              :: isot            !< isot = 0   => Zobler soil type  ( 9 category)
                                            !< isot = 1   => STATSGO soil type (19 category)
+    real(kind=kind_phys), pointer :: pores(:) => null() !< max soil moisture for a given soil type for land surface model


Variables only used within stochastic_physics_wrapper would ideally be contained entirely within that module, possibly in a datatype defined for that module.

I adopted pores and resid from UFS. They are from land surface models not entirely with in stochastic_physics_wrapper. I will leave it there for future applying perturbations to land state.

lharris4 · 2026-03-10T15:54:08Z

@@ -2654,36 +2723,48 @@ subroutine control_initialize (Model, nlunit, fn_nml, me, master, &
 !---Cellular automaton options


What do these CA parameters do? In particular pert_radtend is referenced from GFS_physics_driver.F90 and a brief explanation would be helpful to understand the driver logic.

The meaning of those variables has been added

JosephMouallem · 2026-03-10T15:47:33Z

@@ -1,5 +1,6 @@
 module GFS_typedefs

+       use mpi_f08


All MPI related functionalities should be done through FMS

@JosephMouallem Can you show me an example how to do this through FMS? Thanks.

@MingjingTong-NOAA You could directly import any related mpi functionality from fms modules, their name usually start with mpp_, here is an example:
use mpp_mod, only: mpp_pe, mpp_sync_self, mpp_sync, ...
If you could let me know what is needed here, I can point you to the right calls.

@JosephMouallem I need to pass a MPI communicator to stochastic physics routines. Please take a look at stochastic_physics/stochastic_physics_wrapper.F90 to see how I pass the communicator. The subroutines called in stochastic_physics/stochastic_physics_wrapper.F90 can be found here /gpfs/f6/bil-coastal-gfdl/scratch/Mingjing.Tong/SHiELD/shield_da_merge/SHiELD_SRC/stochastic_physics/

Thanks.

I've removed the MPI communicator from physics.

JosephMouallem · 2026-03-10T15:47:44Z

  type GFS_init_type
    integer :: me                                !< my MPI-rank
    integer :: master                            !< master MPI-rank
+    type(MPI_Comm) :: fcst_mpi_comm              !< forecast tasks mpi communicator


same as above

Same as above

JosephMouallem · 2026-03-10T15:49:01Z


    integer              :: me              !< MPI rank designator
    integer              :: master          !< MPI rank of master atmosphere processor
+    type(MPI_Comm)       :: communicator    !< MPI communicator


Same comment as above and for all related code that follows.

Same as above

JosephMouallem · 2026-03-10T16:00:10Z

+    logical :: param_update_flag
+
+#ifdef _OPENMP
+    nthreads = omp_get_max_threads()


I think this should be done upstream and not here

…c namelist parameters, pass nthreads into stochastic_physics/stochastic_physics_wrapper.F90 from upstream routine

lharris4

I think this is good. I'm a bit concerned about the profusion of new variables but the SHiELD physics needs a general cleaning up anyway.

lharris4 · 2026-06-16T15:40:31Z

@JosephMouallem @laurenchilutti can you take a look at @MingjingTong-NOAA's PR?

Leverage changes from EMC to use updated stochastic physics

faf9f84

MingjingTong-NOAA requested review from JosephMouallem, laurenchilutti and lharris4 March 5, 2026 22:45

MingjingTong-NOAA mentioned this pull request Mar 6, 2026

Correct the use iau_offset and add new functions to output restart files NOAA-GFDL/FMScoupler#176

Open

lharris4 requested changes Mar 10, 2026

View reviewed changes

JosephMouallem reviewed Mar 10, 2026

View reviewed changes

Remove MPI communicator from the physics, add definition of stochasti…

c0ecda7

…c namelist parameters, pass nthreads into stochastic_physics/stochastic_physics_wrapper.F90 from upstream routine

laurenchilutti requested review from JosephMouallem and lharris4 March 30, 2026 19:13

lharris4 approved these changes Apr 8, 2026

View reviewed changes

		@@ -2654,36 +2723,48 @@ subroutine control_initialize (Model, nlunit, fn_nml, me, master, &
		!---Cellular automaton options

Conversation

MingjingTong-NOAA commented Mar 5, 2026

Uh oh!

lharris4 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lharris4 left a comment

Choose a reason for hiding this comment

Uh oh!

lharris4 commented Jun 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants