Separate use of initial increment/perturbation file from REPLAY/+03 ICs #3119
Conversation
changes. Lines starting
…lobal-workflow into gefs_replay_ci
Fix a bug in a variable check in the extractvars task where parenthesis in the variable name will prompt a false warning message.
Ice variables are extracted from the history files instead of the product files. This is necessary because there are certain ice variables that are needed that are only available in the history files and not the product files.
Add suggested patches to gefs_replay_ci
Shellcheck disable has been added. Also, f000 is skipped in extractvars for ocn and ice since there are no products for f000 for ocn and ice. This change will remove warning messages.
NeilBarton-NOAA
requested review from
WalterKolczynski-NOAA,
aerorahul,
junwang-noaa,
KateFriedman-NOAA,
DavidHuber-NOAA and
TerrenceMcGuinness-NOAA
as code owners
| cdate="${PDY}${cyc}" | ||
| if (( cdate < 2017072000 )); then | ||
| export NST_SPINUP=1 | ||
| fi |
There was a problem hiding this comment.
Curious why this is being removed? No longer need?
There was a problem hiding this comment.
The ICs used for the reforecasting effort have NSST spun up.
Co-authored-by: Walter Kolczynski - NOAA <[email protected]>
|
@CatherineThomas-NOAA @JessicaMeixner-NOAA Would you or someone else on your teams please review the model setting changes that impact GFS to confirm the new settings are correct? |
Thanks @WalterKolczynski-NOAA - We + @RuiyuSun will look into this and see what the impact is. Since it's not obvious what the changes are, we'll have to run a couple tests but will try to run those tomorrow. |
I don't know much about the parameters in the FV3. Could @KevinViner-NOAA help reviewing these changes (dddmp, hord_, kord_)? Thanks!
I checked the forecast_predet.sh in the hr.v4 version of the global workflow and didn't see this hord_dp
I don't know much about the dynamic parameters in FV3. Could @KevinViner-NOAA help to review the changes in these parameters? Thanks! I checked the forecast_predet.sh in the hr.v4 but didn't see hord_dp being set? When this is added? |
|
hord_dp is set to -5 for non-hydrostatic GFSv17/GEFSv13 apps and 10 for hydrostatic SFS app. The settings are correct. I believe @XiaqiongZhou-NOAA added the option for the hydrostatic case. |
|
Well, appears there are changes been made to set hord_dp differently for mono and non_mono. @XiaqiongZhou-NOAA please check and confirm. We used to have MONO defined in config.fcst. In the latest workflow I saw MONO is used in forecast_predet.sh but I do not see anywhere it is defined. The logics around the blocks near line 329 and line 359 then become problematic. |
|
"MONO" and "TYPE" are a yaml variable for SFS and GEFS which are defined at ~ci/cases/yamls/sfs_defaults.yaml and parm/config/gefs/yaml/defaults.yaml, but they are not chargeable in GFS and defined in ~parm/config/gfs |
|
Running the C48_S2SW CI test of the g-w, we get one configuration change in input.nml: Original: I don't think this is expected behavior based on the comments from @XiaqiongZhou-NOAA I also tried to run the C96C48_hybatmDA CI test and am having trouble running that one for some reason. But first we can check on the change of hord_dp |
| fi | ||
| else # non-monotonic options | ||
| d_con=${d_con_nonmono:-"1."} | ||
| do_vort_damp=".true." | ||
| if [[ "${TYPE}" == "nh" ]]; then # non-monotonic and non-hydrostatic | ||
| hord_mt=${hord_mt_nh_nonmono:-"5"} | ||
| hord_xx=${hord_xx_nh_nonmono:-"5"} | ||
| hord_dp=${hord_xx_hydro_mono:-"-5"} | ||
| hord_dp=${hord_xx_nh_nonmono:-"-5"} |
There was a problem hiding this comment.
it should be changed to hord_dp=${hord_dp_nh_nonmono:-"-5"}.
Then add "export hord_dp_nh_nonmono=-5" in parm/config/gfs/config.fcst
I guess have to do the same for gefs/config.fcst.
Description
This PR is a continuation of the previous SFS PR. The use of initial perturbations/increments for ensemble long forecasts (e.g., SFS and GEFS) was missing from the previous PR because this option was included with the REPLAY_IC option.
In this PR, the REPLAY_IC option is now strictly for the +03 ICs, and the variables USE_ATM_ENS_PERTURB_FILES and USE_OCN_ENS_PERTURB_FILES are used for initialization with the initial perturbation/increment files used for SFS and GEFS reforecasting. This separation will also make it easier to remove the +03 IC optoin.
Type of change
Change characteristics
How has this been tested?
I tested long forecast runs on hercules and multiple CI tests.
Checklist