Bug fix and rework the scale-aware code #1840

weiwangncar · 2023-03-30T01:43:34Z

TYPE: bug fix

KEYWORDS: scale-aware ntiedtke

SOURCE: internal

DESCRIPTION OF CHANGES:
Problem:
The original PR to add scale-awareness to the new Tiedtke scheme had a bug: the introduction of the namelist control didn't work correctly. Also the introduction of the namelist control made the code look ugly. After discussing with the original developer, we agreed if the changes to scale the cloud to rain conversion coefficient (which helps to reduce very light precip), and saturation check as a condition for triggering mid-level convection (which helps to the number of grid points that are classified as mid-level convection) are removed, the rest of the code can be added without a namelist control.

Solution:
Removing the namelist introduced in PR#1806, and the changes related to cloud-to-rain conversion coefficient and saturation check for mid-level convection, and only keeping the scale-awareness only for convective adjustment time scale and shallow mass fluxes.

LIST OF MODIFIED FILES:
M Registry/Registry.EM_COMMON
M dyn_em/module_first_rk_step_part1.F
M phys/module_cu_ntiedtke.F
M phys/module_cumulus_driver.F

TESTS CONDUCTED:

Works as expected in two cases.
The Jenkins tests have passed.

weiwangncar · 2023-03-30T02:49:57Z

The Jenkins test results:

Test Type              | Expected  | Received |  Failed
= = = = = = = = = = = = = = = = = = = = = = = =  = = = =
Number of Tests        : 23           24
Number of Builds       : 60           57
Number of Simulations  : 158           150        0
Number of Comparisons  : 95           86        0

Failed Simulations are: 
None
Which comparisons are not bit-for-bit: 
None

weiwangncar · 2023-03-30T20:35:52Z

The Jenkins test results:

Test Type              | Expected  | Received |  Failed
= = = = = = = = = = = = = = = = = = = = = = = =  = = = =
Number of Tests        : 23           24
Number of Builds       : 60           57
Number of Simulations  : 158           150        0
Number of Comparisons  : 95           86        0

Failed Simulations are: 
None
Which comparisons are not bit-for-bit: 
None

phys/module_cu_ntiedtke.F

hichunxi · 2023-03-31T07:00:07Z

phys/module_cu_ntiedtke.F

@@ -1039,14 +1035,10 @@ subroutine cumastrn &
 if(ldcum(jl).and.ktype(jl).eq.1) then
 ikb = kcbot(jl)
 ikt = kctop(jl)
- ztau = ztauc(jl) * scale_fac(jl)
+ ztau = ztauc(jl) * scale_tau(jl)


Maybe it is better to set ztauc (360., 10800.), then ztau=ztauc*scale_fac

@weiwangncar please check the code here. My suggestion is to add (360, 10800) to ztauc. There will be no additional time limit for ztau.

@hichunxi I thought that is what I did based on your recommendation before. Here is the section in the new code:

ztauc(jl) = max(ztmst,ztauc(jl)) ztauc(jl) = max(360.,ztauc(jl)) ztauc(jl) = min(10800.,ztauc(jl)) ztau = ztauc(jl) * scale_fac(jl)

@weiwangncar Sorry I missed it. It looks good now. I will approve this PR very soon.

hichunxi · 2023-03-31T07:10:51Z

phys/module_cu_ntiedtke.F

@@ -1085,10 +1077,8 @@ subroutine cumastrn &
 else
 zmfub1(jl) = zmfub(jl)
 end if
+ zmfub1(jl) = zmfub1(jl)/scale_mf(jl)


scale-awareness for shallow convection at 15 km grid spacing is questionable.

weiwangncar · 2023-03-31T19:24:34Z

@hichunxi Thanks for the comments. If I change the constant in the scaling formula a bit to make the value at 15 km to be equal to that in the original formula, I get this plot:

Would this be acceptable. As you can see, the green line in the plot is for scaling shallow convection, and it is a very weak scaling.

hichunxi · 2023-03-31T20:33:02Z

@hichunxi Thanks for the comments. If I change the constant in the scaling formula a bit to make the value at 15 km to be equal to that in the original formula, I get this plot:

Would this be acceptable. As you can see, the green line in the plot is for scaling shallow convection, and it is a very weak scaling.

For deep convection, it looks ok. But for shallow convection, the scale-factor is around 2 at 8 km grid spacing, which might be a little bit too high.

weiwangncar · 2023-03-31T20:39:11Z

@hichunxi You might be right. How about we leave it like this now, and make adjustment if we have evidence that it doesn't work well or is not realistic in the future.

hichunxi · 2023-04-03T01:29:02Z

@weiwangncar Ok, leave it as is now and make potential changes in the future.

…tau before scaling

weiwangncar · 2023-04-06T18:53:34Z

The Jenkins test results:

Test Type              | Expected  | Received |  Failed
= = = = = = = = = = = = = = = = = = = = = = = =  = = = =
Number of Tests        : 23           24
Number of Builds       : 60           57
Number of Simulations  : 158           150        0
Number of Comparisons  : 95           86        0

Failed Simulations are: 
None
Which comparisons are not bit-for-bit: 
None

weiwangncar · 2023-04-06T19:02:54Z

@hichunxi Can you take a look at this final change? Thanks.

weiwangncar · 2023-04-06T20:58:29Z

@hichunxi Thanks!
@dudhia @smileMchen Can you review and approve if it is ok? Thanks!

TYPE: bug fix KEYWORDS: scale-aware ntiedtke SOURCE: internal DESCRIPTION OF CHANGES: Problem: The original PR to add scale-awareness to the new Tiedtke scheme had a bug: the introduction of the namelist control didn't work correctly. Also the introduction of the namelist control made the code look ugly. After discussing with the original developer, we agreed if the changes to scale the cloud to rain conversion coefficient (which helps to reduce very light precip), and saturation check as a condition for triggering mid-level convection (which helps to the number of grid points that are classified as mid-level convection) are removed, the rest of the code can be added without a namelist control. Solution: Removing the namelist introduced in PR#1806, and the changes related to cloud-to-rain conversion coefficient and saturation check for mid-level convection, and only keeping the scale-awareness only for convective adjustment time scale and shallow mass fluxes. LIST OF MODIFIED FILES: M Registry/Registry.EM_COMMON M dyn_em/module_first_rk_step_part1.F M phys/module_cu_ntiedtke.F M phys/module_cumulus_driver.F TESTS CONDUCTED: 1. Works as expected in two cases. 2. The Jenkins tests have passed.

weiwangncar added 2 commits March 17, 2023 16:47

fixed a bug that caused zprcdgw not defined

ae96632

rework the scale-aware code

66684e8

weiwangncar requested review from a team as code owners March 30, 2023 01:43

weiwangncar changed the title ~~Fix sa ntk~~ Bug fix and rework the scale-aware code Mar 30, 2023

weiwangncar added the release-v4.5 label Mar 30, 2023

dudhia previously approved these changes Mar 30, 2023

View reviewed changes

smileMchen previously approved these changes Mar 30, 2023

View reviewed changes

hichunxi reviewed Mar 31, 2023

View reviewed changes

phys/module_cu_ntiedtke.F Outdated Show resolved Hide resolved

hichunxi reviewed Mar 31, 2023

View reviewed changes

make scaling factor more continous at reference grid size and limit z…

b16031a

…tau before scaling

weiwangncar dismissed stale reviews from smileMchen and dudhia via b16031a April 6, 2023 17:55

hichunxi approved these changes Apr 6, 2023

View reviewed changes

dudhia approved these changes Apr 10, 2023

View reviewed changes

smileMchen approved these changes Apr 10, 2023

View reviewed changes

weiwangncar merged commit cfc2801 into wrf-model:release-v4.5 Apr 10, 2023

weiwangncar deleted the fix-sa-ntk branch April 10, 2023 20:06

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Bug fix and rework the scale-aware code #1840

Bug fix and rework the scale-aware code #1840

weiwangncar commented Mar 30, 2023 •

edited

Loading

weiwangncar commented Mar 30, 2023

weiwangncar commented Mar 30, 2023

hichunxi Mar 31, 2023

hichunxi Apr 6, 2023

weiwangncar Apr 6, 2023

hichunxi Apr 6, 2023

hichunxi Mar 31, 2023

weiwangncar commented Mar 31, 2023

hichunxi commented Mar 31, 2023

weiwangncar commented Mar 31, 2023

hichunxi commented Apr 3, 2023

weiwangncar commented Apr 6, 2023

weiwangncar commented Apr 6, 2023

weiwangncar commented Apr 6, 2023

Bug fix and rework the scale-aware code #1840

Bug fix and rework the scale-aware code #1840

Conversation

weiwangncar commented Mar 30, 2023 • edited Loading

weiwangncar commented Mar 30, 2023

weiwangncar commented Mar 30, 2023

hichunxi Mar 31, 2023

Choose a reason for hiding this comment

hichunxi Apr 6, 2023

Choose a reason for hiding this comment

weiwangncar Apr 6, 2023

Choose a reason for hiding this comment

hichunxi Apr 6, 2023

Choose a reason for hiding this comment

hichunxi Mar 31, 2023

Choose a reason for hiding this comment

weiwangncar commented Mar 31, 2023

hichunxi commented Mar 31, 2023

weiwangncar commented Mar 31, 2023

hichunxi commented Apr 3, 2023

weiwangncar commented Apr 6, 2023

weiwangncar commented Apr 6, 2023

weiwangncar commented Apr 6, 2023

weiwangncar commented Mar 30, 2023 •

edited

Loading