Clubb coriolis by HingOng · Pull Request #19 · larson-group/clubb_release

HingOng · 2025-08-06T19:20:04Z

These changes implement the nontraditional and traditional Coriolis terms, each controlled by a flag. The nontraditional ones are explicitly integrated in the up2, wp2, and upwp equations, and so are the traditional ones in the upwp and vpwp equations. The Coriolis parameters are either passed through the host model interface or calculated from the latitude in the standalone namelist. The Coriolis terms are ignored by the default flag setting.

With the default setting, the only change is the setting of the traditional Coriolis parameter from the standalone namelist. Previous versions allow user to set a Coriolis parameter inconsistent with the latitude. The present version ignores the "fcor_nl" and calculates "fcor" from "lat_vals".

The implementation of the Coriolis terms has been validated with an analytical benchmark.

…g it with an unstructured data statement. I added this due as a fix before thinking qclvar was a local variable, which happened because I mixed up the CLUBB_CAM and CLUBBND_CAM flags. larson-group/cam#175

…t was breaking the ECT test, see larson-group/cam#175

…contains pointers, and these pointers will be different on the host and device, so it is bad in theory to copy the structure back to the CPU as it might overwrite cpu memory pointers with gpu memory pointers. In practice though I've seen no problems caused by this, I'm just making this commit preemptively, and it has already been tested with the ECT test.

…oriolis

vlarson

Is this pull request a superset of #20 ?

Do these code changes leave the results BFB identical when the flags are set to zero?

I can make these code changes if you don't have time or interest.

vlarson · 2025-08-07T19:11:07Z

input/stats/host_budgets_stats.in

 'vp2_bt', 'vp2_ma', 'vp2_ta', 'vp2_tp', 'vp2_dp1', 'vp2_dp2', 
 'vp2_pr1', 'vp2_pr2', 'vp2_sdmp', 'vp2_cl', 'vp2_pd', 'vp2_sf', 'vp2_splat',
-'up2_bt', 'up2_ma', 'up2_ta', 'up2_tp', 'up2_dp1', 'up2_dp2', 
+'up2_bt', 'up2_ma', 'up2_ta', 'up2_tp', 'up2_dp1', 'up2_dp2', 'up2_nct',


We need to update the budget balance test:

https://github.com/larson-group/clubb_release/tree/master/postprocessing/check_budgets_balance

vlarson · 2025-08-07T19:13:36Z

src/CLUBB_core/version_clubb_core.txt

@@ -0,0 +1,312 @@
+commit 51107487c8ae0f96461e5a882a203d3019aa6aa5 (HEAD -> clubb_Coriolis_cam, newremote.00/clubb_Coriolis_cam)


We won't need to commit this file because it's automatically generated.

vlarson · 2025-08-07T19:15:43Z

src/CLUBB_core/clubb_api_module.F90

    !$acc end data
    !$acc end data
+
+#ifdef CLUBB_CAM


Why is this needed?

Hing thinks that this was included by accident, but he hasn't tested the code on a GPU, so he's not 100% sure.

vlarson · 2025-08-07T19:18:41Z

input/tunable_parameters/coriolis_test_parameters.in

@@ -0,0 +1,110 @@
+! $Id$


Why do we need this file? Can we just add any new parameters to

https://github.com/larson-group/clubb_release/blob/master/input/tunable_parameters/tunable_parameters.in ?

coriolis_test_parameters.in shuts off all the pressure terms so that Hing can compare with his analytic solution.

We could consider creating a Jenkins test that runs the Coriolis test and compares to the analytic answer.

vlarson · 2025-08-07T19:21:56Z

src/SILHS/version_silhs.txt

@@ -0,0 +1,288 @@
+commit 46cac2787c3802be0418762948ad6d1231dde182


Likewise, we won't need to commit this file because it's automatically generated.

vlarson · 2025-08-07T19:25:35Z

src/clubb_driver.F90

-      fcor_nl
+      fcor_nl ! obsolete
+      ! fcor_nl can be read from a namelist but is not used for any calculation.
+      ! fcor and fcory are calculated from lat_vals from a namelist.


Is fcor always generated from lat_vals? I think we want to retain the capability to directly specify fcor.

Because of this change, we don't expect the results to be BFB. We'll have to see how much the answers change. If not much, then we could delete fcor_nl. But if the answers do change, then maybe we do want to keep the option of specifying fcor, just so that we match our benchmark LES better.

vlarson · 2025-08-07T19:26:13Z

src/clubb_driver.F90

    sens_ht   = 0._core_rknd
    latent_ht = 0._core_rknd
-    fcor_nl   = 1.e-4_core_rknd
+    fcor_nl   = 1.e-4_core_rknd ! obsolete


Did you make this obsolete?

vlarson · 2025-08-07T19:27:37Z

src/clubb_driver.F90

    p_sfc = p_sfc_nl
    T_sfc = T_sfc_nl
-    fcor = fcor_nl
+    fcor  = two * omega_planet * sin ( lat_vals*radians_per_deg ) ! Retire fcor_nl


I think that we want this to be optional, if fcor_nl is not specified.

I have run a cgils_s6 test with this line restored as "fcor = fcor_nl" and compared the results against those from the "larson-group:master" branch. They match BFB.

A problem is that cgils_s6 test is extremely sensitive to fcor.
If prescribed from fcor_nl, fcor = 4.264000000000000E-005, the cgils_s6 test completes normally.
If calculated from lat_vals, fcor = 4.263948942216714E-005, the cgils_s6 test aborts with the following error:
iteration = 118; time = 35400.0
Fatal error in CLUBB - MPI Process 1 / Chunk index 1
Column index not av
ailable. Latitude/Longitude for column 1: 17.00/******
Latitude range from -90
to 90 degrees, Longitude range from -180 to 180 degrees
mixt_frac * precip_frac_1 + ( 1 - mixt_frac ) * precip_frac_2 /= precip_frac wi
thin numerical roundoff
level = 15
mixt_frac * precip_frac_1 + ( 1 - mixt_frac ) * precip_frac_2 =
8.254983776519802E-002
precip_frac = 8.254983776519802E-002
Fatal error in CLUBB in - MPI Process 1 / Chunk index 1
Grid column index i
= 1, Latitude 17.00 / Longitude 17.00
Latitude range from -90 to 90 degrees
, Longitude range from -180 to 180 degrees
precip_frac_assert_check failed
Fatal error in CLUBB - MPI Process 1 / Chunk index 1
Column index not av
ailable. Latitude/Longitude for column 1: 17.00/******
Latitude range from -90
to 90 degrees, Longitude range from -180 to 180 degrees
in setup_pdf_parameters after calling precip_fraction
Fatal error in CLUBB - MPI Process 1 / Chunk index 1
Column index not av
ailable. Latitude/Longitude for column 1: 17.00/******
Latitude range from -90
to 90 degrees, Longitude range from -180 to 180 degrees
Fatal error after setup_pdf_parameters_api
Fatal error in CLUBB - MPI Process 1 / Chunk index 1
Column index not av
ailable. Latitude/Longitude for column 1: 17.00/******
Latitude range from -90
to 90 degrees, Longitude range from -180 to 180 degrees
Fatal error after pdf_hydromet_microphys_prep
CLUBB-TIMER time_loop_init = 0.0022
CLUBB-TIMER time_clubb_advance = 0.0456
CLUBB-TIMER time_clubb_pdf = 0.0283
CLUBB-TIMER time_SILHS = 0.0001
CLUBB-TIMER time_microphys_scheme = 0.0103
CLUBB-TIMER time_microphys_advance = 0.0171
CLUBB-TIMER time_loop_end = 0.1028
CLUBB-TIMER time_output_multi_col = 0.0001
CLUBB-TIMER time_total = 0.2073
Fatal error in clubb, check your parameter values and timestep

I had to reduce the timestep to normally complete the run with calculated fcor.

vlarson · 2025-08-07T19:31:20Z

src/clubb_driver.F90

    do k = 1, gr%nzm
      do i = 1, ngrdcol
-        upwp(i,k)             = zero          ! vertical u momentum flux
+        upwp(i,k)             = upwp_init(k)  ! vertical u momentum flux


Is upwp_init initialized for all cases? Usually, it should be zero, no?

upwp_init is set automatically from upwp in clubb_driver.F90.

vlarson · 2025-08-07T20:17:56Z

Is this pull request a superset of #20 ?

No, #20 is a pull request for the CAM tag. #19 is a pull request for CLUBB master.

Do these code changes leave the results BFB identical when the flags are set to zero?

Hing hasn't done this test.

vlarson · 2025-08-07T20:39:08Z

src/clubb_driver.F90

+        em   = em + 1.5_core_rknd * w_tol_sqd
+        do i = 1, ngrdcol
+          do k=1,gr%nzm
+            upwp(i,k) = 0.5_core_rknd * dt_main * fcory(i) * ( up2(i,k) - wp2(i,k) )


This is where upwp_init is set.

Rename fcory as fcor_y and add description

Rename fcor as fcor_y and add description. Also, add description for the coriolis_test

Rename fcory as fcor_y and add description.

Add description for the Coriolis test case

HingOng · 2025-11-26T08:57:30Z

This branch is merged with up-to-date larson-group/clubb_release/master
When l_nontraditional_Coriolis and l_traditional_Coriolis are .false. (default), this merged version outputs additional variables up2_nct, wp2_nct, upwp_nct, upwp_tct, and vpwp_tct to *_zm.nc and wp3_nct and wp2up to *_zt.nc. However, the other output variables remain identical with those generated by larson-group/clubb_release/master for the bomex case.
When l_nontraditional_Coriolis and l_traditional_Coriolis are .true., this merged version outputs an additional flag l_hole_fill = "T". However, the other output variables remain identical with the unmerged version of this branch for the atex_long case (known to be sensitive to the nontraditional Coriolis terms).

vlarson · 2025-12-02T20:45:36Z

@HingOng, do you happen to have code changes for clubb_intr.F90 in CAM that correspond to these code changes to CLUBB?

vlarson · 2025-12-02T21:03:29Z

@HingOng, with the newest version of your branch, do the budgets balance, according to this script?:

https://github.com/larson-group/clubb/blob/master/postprocessing/check_budgets_balance/checkBudget.py

HingOng · 2025-12-02T21:10:36Z

@HingOng, do you happen to have code changes for clubb_intr.F90 in CAM that correspond to these code changes to CLUBB?

Yes, I have them on a local branch.

HingOng · 2025-12-02T21:16:36Z

@HingOng, with the newest version of your branch, do the budgets balance, according to this script?:

https://github.com/larson-group/clubb/blob/master/postprocessing/check_budgets_balance/checkBudget.py

Yes, checkBudget.py returns "Budgets successfully balance!"

vlarson · 2025-12-02T21:28:26Z

@HingOng, do you happen to have code changes for clubb_intr.F90 in CAM that correspond to these code changes to CLUBB?

Yes, I have them on a local branch.

Is the branch on github? Or is there another way to share it?

HingOng · 2025-12-02T22:00:00Z

@HingOng, do you happen to have code changes for clubb_intr.F90 in CAM that correspond to these code changes to CLUBB?

Yes, I have them on a local branch.

Is the branch on github? Or is there another way to share it?

I just pushed the branch to my fork of CAM:
https://github.com/HingOng/CAM/tree/clubb_Coriolis

Note that this branch is based on an outdated version of CAM and CLUBB.

LarsonGroupSysAdmin and others added 16 commits June 5, 2024 01:58

Updating the clubb core and silhs version files

4b3e80f

Fixing bug. qclvar copying is handled by the api, but we were deletin…

928998b

…g it with an unstructured data statement. I added this due as a fix before thinking qclvar was a local variable, which happened because I mixed up the CLUBB_CAM and CLUBBND_CAM flags. larson-group/cam#175

Fixing GPU bug. There is a vertical dependency with wpxp. This is wha…

59a16b3

…t was breaking the ECT test, see larson-group/cam#175

Moving PosInf to clubb_model_settings.F90

15e8020

Updating the clubb core and silhs version files

4fd8ccd

Add flags for the Coriolis terms.

873f5c1

Pass fcory from api module into advance module

55e508d

Implement the traditional Coriolis terms

5e24524

Implement the nontraditional Coriolis terms

5110748

Merge remote-tracking branch 'origin/clubb_Coriolis_cam' into clubb_C…

90d009c

…oriolis

Add stats variables for the Coriolis terms

d8008d2

Compute Coriolis parameters from namelist latitude

4d3a6d2

Create the coriolis_test case

32067ed

Tune the coriolis_test case. However, oscillation frequency is too slow

826e2c2

Correct a coefficient in Coriolis. "coriolis_test" matches analytical

a39946b

LarsonGroupSysAdmin force-pushed the master branch from 9c35ec3 to 46cac27 Compare August 7, 2025 04:31

vlarson requested changes Aug 7, 2025

View reviewed changes

vlarson reviewed Aug 7, 2025

View reviewed changes

HingOng and others added 10 commits August 8, 2025 14:53

Update advance_clubb_core_module.F90

ee843d2

Rename fcory as fcor_y and add description

Update advance_wp2_wp3_module.F90

5f5c9b2

Rename fcory as fcor_y and add description

Update advance_xm_wpxp_module.F90

e11b4b9

Rename fcory as fcor_y and add description

Update advance_xp2_xpyp_module.F90

4df8aaf

Rename fcory as fcor_y and add description

Update clubb_api_module.F90

21b6bdb

Rename fcory as fcor_y and add description

Update spurious_source_test.F90

06794ce

Rename fcory as fcor_y and add description

Update clubb_driver.F90

d6c0750

Rename fcor as fcor_y and add description. Also, add description for the coriolis_test

Update generalized_grid_test.F90

6941185

Rename fcory as fcor_y and add description.

Update coriolis_test_model.in

7a523d2

Add description for the Coriolis test case

Stablize the cgils_s6 case with smaller time step

d07eac4

HingOng and others added 12 commits August 27, 2025 02:35

Create the ekman case

e9981ae

Tune the ekman case to match LES

5fec22a

Compute wp2up with PDF and pass along with wp2thvp

5e90977

Implement the nontraditional Coriolis term for wp3

3f4cc18

Fine-tune parameters for the ekman case

5145f75

Create the atex_long case

7681ea8

Retune ekman case with l_vary_convect_depth = .true.

6140416

Fine-tune the atex_long case

58764c2

Add atex_long case to pyplotgen

ad573ab

Fine-tune atex_long case with less radical changes

147b522

Updating the clubb core and silhs version files

6d04a4f

Merge remote-tracking branch 'upstream/master' into clubb_Coriolis

d816536

HingOng requested a review from vlarson November 26, 2025 08:29

vlarson requested a review from huebleruwm November 26, 2025 12:53

		@@ -0,0 +1,312 @@
		commit 51107487c8ae0f96461e5a882a203d3019aa6aa5 (HEAD -> clubb_Coriolis_cam, newremote.00/clubb_Coriolis_cam)

		@@ -0,0 +1,288 @@
		commit 46cac2787c3802be0418762948ad6d1231dde182

Conversation

HingOng commented Aug 6, 2025

Uh oh!

vlarson 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!

vlarson commented Aug 7, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

HingOng commented Nov 26, 2025

Uh oh!

vlarson commented Dec 2, 2025

Uh oh!

vlarson commented Dec 2, 2025

Uh oh!

HingOng commented Dec 2, 2025

Uh oh!

HingOng commented Dec 2, 2025

Uh oh!

vlarson commented Dec 2, 2025

Uh oh!

HingOng commented Dec 2, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants