Change Rayleigh damping settings to improve model instability and acoustic wave time step for C768 to improve computational efficiency #1612
Comments
|
@XiaqiongZhou-NOAA Is the fast_tau_w_sec=0.2 used by all the resolutions (C48-C768)? Would you please create an issue in ufs weather model repo so that the UFS RT will use the same value in the namelist? I saw the default value for fast_tau_w_sec is 0. |
|
I would use this for HR2 at first and need more experiments to confirm. |
|
@XiaqiongZhou-NOAA the settings used in HR2 should match whatever settings we're using in the ufs-weather-model cpld_control_gfsv17 test so if it's not already there, so when you're ready to confirm what settings we should use, they should be updated in both that test in ufs-weather-model and here in the g-w. |
XiaqiongZhou-NOAA
changed the title
|
@XiaqiongZhou-NOAA Kate, please add a link to the Google Slides showing the test and evaluation of implicit Rayleigh damping you have done. This PR contains two parts. The credit of the work goes to Kate (Xiaqiong) Zhou, Wei Li, Kevin Viner and our GFDL collaborators. |
|
@XiaqiongZhou-NOAA I removed the HR2 milestone from this issue as we don't have a corresponding update to ufs-weather-model yet. |
|
Jessica,
There is no model update needed for this PR. The dycore was updated by
GFDL a while ago
On Wed, Jul 19, 2023 at 7:47 AM Jessica Meixner ***@***.***> wrote:
@XiaqiongZhou-NOAA <https://github.com/XiaqiongZhou-NOAA> I removed the
HR2 milestone from this issue as we don't have a corresponding update to
ufs-weather-model yet.
—
Reply to this email directly, view it on GitHub
<#1612 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AKY5N2L2UE6T6R2WYKRBAITXQ7JMTANCNFSM6AAAAAAYDVTRGQ>
.
You are receiving this because you commented.Message ID:
***@***.***>
--
*Fanglin Yang, Ph.D.*
*Chief, Physics and Dynamics Division*
*NOAA/NWS/NCEP Environmental Modeling Center*
*https://www.emc.ncep.noaa.gov/gmb/wx24fy/fyang/
<https://www.emc.ncep.noaa.gov/gmb/wx24fy/fyang/>*
|
|
By model changes, maybe I should be more specific by saying "the coefficient updates need to be updated in the GFSv17 ufs-weather-model regression test". Does that make more sense? If it's C768 setting only, then there's not a corresponding regtest. Otherwise, we try to keep things in sync. |
|
The time step change can only be applied to C768. Rayleigh damping can be
applied to other resolutions but is optional
On Wed, Jul 19, 2023 at 8:14 AM Jessica Meixner ***@***.***> wrote:
By model changes, maybe I should be more specific by saying "the
coefficient updates need to be updated in the GFSv17 ufs-weather-model
regression test". Does that make more sense? If it's C768 setting only,
then there's not a corresponding regtest. Otherwise, we try to keep things
in sync.
—
Reply to this email directly, view it on GitHub
<#1612 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AKY5N2I4FVFG26YMCEACOA3XQ7MSLANCNFSM6AAAAAAYDVTRGQ>
.
You are receiving this because you commented.Message ID:
***@***.***>
--
*Fanglin Yang, Ph.D.*
*Chief, Physics and Dynamics Division*
*NOAA/NWS/NCEP Environmental Modeling Center*
*https://www.emc.ncep.noaa.gov/gmb/wx24fy/fyang/
<https://www.emc.ncep.noaa.gov/gmb/wx24fy/fyang/>*
|
|
I only created an issue in the UFS-weather model not an PR and assume that the code manager of the ufs-weather-model will take care that since it is only a namelist change: |
|
The namelists for the regression tests are version controlled on github and therefore require a PR for any changes. Could you please submit one to the ufs-weather-model and then after that's merged we can also update the global-workflow? |
|
Dear @JessicaMeixner-NOAA and @XiaqiongZhou-NOAA, I'm taking care of the ufswm namelists to be matched with G-W side. Currently I'm updating the ufswm C$resolution tests to GFSv17. I'll take care of this in that PR or another one. Thanks |
|
@SadeghTabas-NOAA, When can PR#1833 " ufswm C$resolution tests to GFSv17" be completed ? We'd like to include the new Rayleigh damping and new time-stepping for acoustic waves in the upcoming HR experiments sooner than later. Thanks. |
|
@yangfanglin |
@yangfanglin Sorry I missed this massage, I've updated the tests C48 and C96 however for the tests C192 and C384 we need inputs and ICs of the common time that are not currently available, so I don't know how long it may take to merge PR#1833. As @aerorahul mentioned, if this is a priority please open a new PR. Thanks and sorry for inconvenience. |
|
Since HR2 has started to run, let's try to get this update into HR3. After the HR2 tags for workflow and model are created, @XiaqiongZhou-NOAA please create a PR to update model regression tests and another PR to update workflow. |
|
@XiaqiongZhou-NOAA @yangfanglin Is this something that you're still looking to change in workflow? I haven't seen a PR for it. |
|
I think the issue just wasn't linked. Here's a recent input.nml: /scratch1/NCEPDEV/climate/Jessica.Meixner/forlisa.input.nml (from a branch but no input.nml parts were updated). I see fast_tau_w_sec=0.2 updated. But it'd be good to confirm everything else is as expected. |
|
It is updated with this PR.
Update functionality for Prototype HR3 UGWP by mdtoyNOAA · Pull Request
#1987 · NOAA-EMC/global-workflow (github.com)
<https://github.com/NOAA-EMC/global-workflow/pull/1987/files>
…On Tue, Jan 16, 2024 at 2:41 PM Walter Kolczynski - NOAA < ***@***.***> wrote:
@XiaqiongZhou-NOAA <https://github.com/XiaqiongZhou-NOAA> @yangfanglin
<https://github.com/yangfanglin>
Is this something that you're still looking to change in workflow? I
haven't seen a PR for it.
—
Reply to this email directly, view it on GitHub
<#1612 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ALQE7UUQVCABXA7LNRXD743YO3JUNAVCNFSM6AAAAAAYDVTRGSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOJUGM4TMNRXHA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
Thanks @XiaqiongZhou-NOAA @JessicaMeixner-NOAA Completed in #1987 |
Description
This request is to add "fast_tau_w_sec" in dynamics input namelist (&fv_core_nml). This parameter controls implicit Rayleigh damping of vertical velocity introduced to UFS with PR#1720.
The suggested default value:
fast_tau_w_sec=0.2
tau=0
rf_cutoff=10
n_split=4
n_split is changed from 6 to 4 to save computational resources
and meantime the model remains stable.
The results from the corresponding sensitivity tests are here: https://docs.google.com/presentation/d/1n4DUr_4KqbJ5t04n2kgZDpLCJV_KpvXS/edit?usp=sharing&ouid=117079466243404562293&rtpof=true&sd=true
The impact of the change of Rayleigh damping on C384L127:
https://docs.google.com/presentation/d/1KMAuJ4H5uczhE5dZPrTiMhapktVMtGwKIa2ahGJYLO4/edit?usp=sharing
Requirements
HR2 can run with this new option.
Acceptance Criteria (Definition of Done)
The modification of the parameters is determined based on the verification of sensitivity experiments.
**Related to
ufs-community/ufs-weather-model#1824
The text was updated successfully, but these errors were encountered: