Add support for NSSL 2-moment cloud microphysics scheme #761

MicroTed · 2021-10-15T04:16:22Z

NSSL 2-moment Cloud Microphysics (National Severe Storms Laboratory)

Features consistent with WRF-ARW release (same code base):

Fully 2-moment (mass and number), for droplets, rain, ice crystal, snow, graupel, and hail (hail can be deactivated)
Predicted bulk concentration of cloud condensation nuclei (CCN) (optional)
Allows residual supersaturation (no saturation adjustment) for reasonably physical aerosol response
Adaptive 2-moment sedimentation to prevent excessive size sorting
Predicts bulk density of graupel and hail
Hail species intended as “true” hail, characterized by size and source rather than simple density
Designed primarily for convection-allowing grid spacing (dx <= ~4km)
Computes effective radii for radiation (sets effr_in=.true.)
Computes radar reflectivity assuming Rayleigh scattering (sets ltradar=.true.)

Main reference: Mansell, E. R., C. L. Ziegler, and E. C. Bruning, 2010: Simulated electrification of a small thunderstorm with two-moment bulk microphysics. J. Atmos. Sci., 67, 171-194, doi:10.1175/2009JAS2965.1

Namelist options:

nssl_cccn - CCCN background CCN concentration. Default: 0.6e9 (per m^3 at STP)
nssl_alphah - graupel shape parameter, default = 0.
nssl_alphahl - hail shape parameter, default = 1.
nssl_hail_on - hail activation flag, default = .false.
nssl_ccn_on - CCN prediction flag, default = .true.
nssl_invertccn - store CCN as number of activated instead of number remaining, default = .true. (avoids need for specified boundary conditions)

Tested with SCM (pull request initiated) and earlier version tested with UFS-LAM
e.g.,
./run_scm.py -c twpice -n input_GSD_v1nssl_nohailnoccn.nml -t tracers_gsd_nssl_nohail_noccn.txt -s SCM_GSD_v1nssl

Related PRs:
For SCM: NCAR/ccpp-scm#277
For FV3: NOAA-EMC/fv3atm#472

…n, cloud ice, snow, graupel, and hail. Graupel and hail have predicted bulk density via the particle volume. Hail can be deactived. Simple CCN concentration can be predicted, either as the count of unactivated or activated nuclei. (Mansell et al. 2010, JAS)

…ers)

- Split NSSL conditional in GFS_rrtmg_pre.F90 from Thompson for now - Text comments in radiation_clouds.f

…g calcnfromq from running, which creates number concentration from the initial condition hydrometeor mass

…rious points of very small mass with large reflectivity were showing up, perhaps because of the very large time step in UFS (40s). This helps eliminate those.

…t cleaner. Changes to calcnfromq to set droplet number as 9 micron radius droplets, and then deplete CCN if turned on. Also set masses to zero if less than qxmin.

…s way)

…er mixing ratios masses are transferred to water vapor. Also added second estimate for graupel number conc. and take minimum. Added air density limit in setvtz and nssl_2mom_gs to limit fall speed or rhovt. Added limit on Bigg freezing to only act if freezing radius is 8mm or less.

…d flag for nssl_ccn_on - Updataed microphysics - Radiation (rrtmg) includes calculated rain radius. Test code to compute radii in the subroutine, but something not right with incoming number concentrations - Renamed mp_nsslg to mp_nssl

MicroTed · 2021-10-15T04:19:04Z

FYI @tsupinie

MicroTed · 2021-10-15T04:43:42Z

Companion SCM pull request is NCAR/ccpp-scm#277

grantfirl · 2021-10-18T19:02:44Z

@MicroTed, thanks for your contribution. We'll start to review it now. Thanks also for the SCM PR. Is there an associated PR for fv3atm? Also, since this PR touches many other schemes, we may need to add those schemes' respective code owners as reviewers to OK your changes too.

…n NSSL MP

Minor changes to prepare for merge

grantfirl · 2022-02-17T16:51:32Z

@climbfuji Since you requested changes on this PR, you will need to review again.

grantfirl · 2022-02-17T16:53:00Z

@MicroTed This will need to be updated to main until it is your turn in the UFS merge queue. I'm going to ask that it be next behind #820

climbfuji

I found a few places where the code could be optimized for better computational performance, and there may be more (I did not look through the actual NSSL mp code). It will work as-is, but in the few places I highlighted these changes are easy to make.

climbfuji · 2022-02-17T17:03:10Z

physics/GFS_PBL_generic.F90

+               dqdt(i,k,ntgv) = dvdftra(i,k,14) 
+               dqdt(i,k,nthv) = dvdftra(i,k,15) 
+               dqdt(i,k,ntoz) = dvdftra(i,k,16) 
+               IF ( nssl_ccn_on ) THEN


I would move this if loop outside these nested do loops - have another if block and then inside the if block you loop over k and i.

That’s a good point, but don’t modern compilers usually take care of this as part of optimization? Basically the compiler should make two full loops, one with ccn and the other without, so that only one loop runs. A separate loop for one variable might go slightly slower?

Possible. And yes, you are right, the compiler does all sorts of things. It might even do the same with my suggested approach of having two loops. Please ignore my comments regarding those optimizations.

I had implemented this for GFS_PBL_generic.F90. Should I not commit this now?

Sorry. Ted made a good point that I wasn't thinking of.

Yeah, back in the day one had to consider such things! I have probably gotten dumber as compilers have gotten smarter...

So, @climbfuji, does this mean that you approve of the code as-is?

Yes, I will. Need to check the apparently old block of code ...

physics/GFS_PBL_generic.F90

physics/GFS_rrtmg_pre.F90

physics/GFS_suite_interstitial.F90

physics/module_MYNNPBL_wrapper.F90

grantfirl · 2022-02-17T17:17:22Z

@MicroTed I'll make the changes that @climbfuji requested, retest on Hera/intel, and submit a PR into your branch.

physics/GFS_rrtmg_pre.F90

Minor update in microphysics to lower the supersat. threshold at which it allows droplet nucleation at low temperature (T < -36C). This alleviates rare high supersaturation in very deep strong (supercell) updrafts.

MicroTed · 2022-02-24T21:06:13Z

Some internal documentation has been added (comments for subroutines, etc. for doxygen). Plus added file in docs/pdftxt for the NSSL scheme. Thanks to Man Zhang for help!

grantfirl · 2022-02-25T18:24:10Z

@MicroTed This PR is the next one in the queue. Please pull down the latest main, merge it into this branch, and push back to GitHub. If there are merge conflicts (which it appears there may be since this PR and main have both modified a couple of files), it looks like you just need to accept the union of the two changes, and make sure that there is no duplication.

MicroTed · 2022-02-25T21:28:57Z

@grantfirl It is updated and a test run of FV3 was successful.

…heir preceding paths

grantfirl · 2022-03-03T16:21:37Z

@MicroTed We need to update this pull request with #869 in order to fix a compilation memory error on wcoss_dell that is largely unrelated to anything having to do with NSSL MP. Here are the steps that I would suggest:

Starting from within ccpp/physics on your local machine, let's add Dom Heinzeller's fork as a remote (since that is where Add logic to reduce optimization for host-model provided GFS_typedefs.F90 #869 is coming from): git remote add dom-fork https://github.com/climbfuji/ccpp-physics
Fetch Dom's branches: git fetch dom-fork
Checkout Add logic to reduce optimization for host-model provided GFS_typedefs.F90 #869 PR branch: git checkout feature/reduce_opt_gfs_typedefs
Merge this into your feature/nsslmicro branch: git checkout feature/nsslmicro, git merge feature/reduce_opt_gfs_typedefs
Push to Github

Once this is done, you'll need to update your FV3 PR branch as well:

cd up to FV3
git add ccpp/physics, git commit -m "update ccpp/physics submodule pointer"
Push to GitHub

Once that is done, I'll update the ufs-weather-model PR branch.

MicroTed · 2022-03-03T18:07:51Z

@grantfirl OK, ccpp/physics and fv3atm have been pushed.

MicroTed added 21 commits October 14, 2021 21:05

- Fixes subroutine end statements (causes error on some older compil…

a13afa1

…ers)

Add missing 'nthl' to call interface

8c25e52

- Pass effrr into NSSL driver

ef29545

- Split NSSL conditional in GFS_rrtmg_pre.F90 from Thompson for now - Text comments in radiation_clouds.f

Turned off a print statement.

dc2a827

Restore the incorrectly removed flags.

9e30e90

Turn off setting rain radius for now.

7049163

Fixed typo in meta file

315489b

Fixed typo and missing declaration

2b95bde

- Fixed setting of itimestep on first model step. This was preventin…

4b35ce9

…g calcnfromq from running, which creates number concentration from the initial condition hydrometeor mass

Turned on zeroing out of hail for low number concentration. Some spu…

1cfe2c8

…rious points of very small mass with large reflectivity were showing up, perhaps because of the very large time step in UFS (40s). This helps eliminate those.

Added extra printout info for large fall speeds.

c27901c

Call calcnfromq on every time step, which helps keep boundaries a bi…

21a2875

…t cleaner. Changes to calcnfromq to set droplet number as 9 micron radius droplets, and then deplete CCN if turned on. Also set masses to zero if less than qxmin.

Removed re_rain from interface (not used and not planning to use thi…

06fc773

…s way)

Changed itimestep to a purely local variable (i.e., not saved)

6d484f7

Fixed bug in setting array values of "rain" (noticed by E. Aligo)

635e028

Fixed missing setting of save arrays for NSSL.

e18f790

Update to newer base code plus some cleanup of NSSL microphysics

b565f5f

Made IF test on tracer indices in post_run consistent with pre_run

3b7b139

MicroTed requested review from climbfuji, grantfirl and llpcarson as code owners October 15, 2021 04:16

MicroTed mentioned this pull request Oct 15, 2021

Add support for NSSL cloud microphysics scheme (CCPP) NCAR/ccpp-scm#277

Merged

MicroTed added 2 commits October 15, 2021 20:00

Switched 'vertical_dimension' to 'vertical_layer_dimension'

7e41947

Added convert_dry_rho flag

7e38a90

MicroTed requested a review from joeolson42 as a code owner February 11, 2022 18:50

MicroTed and others added 4 commits February 12, 2022 19:13

Reduce potential sedimentation computation with interval_sedi_vt=2

2c02bc5

update some flag units and remove GFS_suite_interstitial dependency o…

779c432

…n NSSL MP

convert argument arrays in mp_nssl.F90 to assumed-shape

b104184

Merge pull request #2 from grantfirl/feature/nsslmicro_gjf

a2a7f8b

Minor changes to prepare for merge

grantfirl approved these changes Feb 17, 2022

View reviewed changes

climbfuji reviewed Feb 17, 2022

View reviewed changes

grantfirl mentioned this pull request Feb 17, 2022

Feature: NSSL Microphysics support ufs-community/ufs-weather-model#1032

Merged

16 tasks

climbfuji reviewed Feb 17, 2022

View reviewed changes

physics/GFS_rrtmg_pre.F90 Outdated Show resolved Hide resolved

Merge branch 'main' into feature/nsslmicro

5da3050

climbfuji requested review from climbfuji and removed request for llpcarson February 17, 2022 20:03

climbfuji approved these changes Feb 17, 2022

View reviewed changes

Added internal documentation (for doxygen).

a9349ed

Minor update in microphysics to lower the supersat. threshold at which it allows droplet nucleation at low temperature (T < -36C). This alleviates rare high supersaturation in very deep strong (supercell) updrafts.

Merge branch 'main' into feature/nsslmicro

580df44

climbfuji added 2 commits March 2, 2022 22:06

Add logic to reduce optimization for multiple files, independent of t…

44a2bd9

…heir preceding paths

No optimization for GFS_typedefs.F90

4d808f8

grantfirl mentioned this pull request Mar 3, 2022

Add logic to reduce optimization for host-model provided GFS_typedefs.F90 #869

Merged

Merge branch 'feature/reduce_opt_gfs_typedefs' into feature/nsslmicro

dc5ed83

grantfirl merged commit 979324a into NCAR:main Mar 4, 2022

This was referenced Mar 4, 2022

Add NSSL microphysics to the CCPP ufs-community/ufs-weather-model#1079

Closed

Radiation Improvements for prototype 8 #871

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add support for NSSL 2-moment cloud microphysics scheme #761

Add support for NSSL 2-moment cloud microphysics scheme #761

MicroTed commented Oct 15, 2021 •

edited

Loading

MicroTed commented Oct 15, 2021

MicroTed commented Oct 15, 2021

grantfirl commented Oct 18, 2021

grantfirl commented Feb 17, 2022

grantfirl commented Feb 17, 2022

climbfuji left a comment

climbfuji Feb 17, 2022

MicroTed Feb 17, 2022

climbfuji Feb 17, 2022

grantfirl Feb 17, 2022

climbfuji Feb 17, 2022

MicroTed Feb 17, 2022

grantfirl Feb 17, 2022

climbfuji Feb 17, 2022

grantfirl commented Feb 17, 2022

MicroTed commented Feb 24, 2022

grantfirl commented Feb 25, 2022

MicroTed commented Feb 25, 2022

grantfirl commented Mar 3, 2022

MicroTed commented Mar 3, 2022

Add support for NSSL 2-moment cloud microphysics scheme #761

Add support for NSSL 2-moment cloud microphysics scheme #761

Conversation

MicroTed commented Oct 15, 2021 • edited Loading

MicroTed commented Oct 15, 2021

MicroTed commented Oct 15, 2021

grantfirl commented Oct 18, 2021

grantfirl commented Feb 17, 2022

grantfirl commented Feb 17, 2022

climbfuji left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

grantfirl commented Feb 17, 2022

MicroTed commented Feb 24, 2022

grantfirl commented Feb 25, 2022

MicroTed commented Feb 25, 2022

grantfirl commented Mar 3, 2022

MicroTed commented Mar 3, 2022

MicroTed commented Oct 15, 2021 •

edited

Loading