Made ERI F1850C test work with gregorian calendar. #1437
Conversation
Several more changes are needed in cime, which will have a commit with the same title. This was tested using ERI_D_C3_CG_Ld8.ne30pg3_g17.F1850C_LTso.derecho_intel.cam-outfrq9s_leapday bld/namelist_files/use_cases/1850_cam_lt.xml Change year of some forcing files to 1852 so that leap day is handled. cime_config/testdefs/testmods_dirs/cam/outfrq9s_leapday/shell_commands Change start year to 2008, so that adding 4 (cime) yields a date that's within the time span of all of the forcing files.
|
I can't request reviewers, assign the issue, or select labels, so please add those as appropriate. |
| <!-- LBC Files : --> | ||
| <flbc_file>atm/waccm/lb/LBC_17500116-20150116_CMIP6_0p5degLat_c180905.nc</flbc_file> | ||
| <flbc_cycle_yr>1850</flbc_cycle_yr> | ||
| <flbc_cycle_yr>1852</flbc_cycle_yr> |
There was a problem hiding this comment.
I'm not sure that it will be acceptable to change the start year for this usecase, could you create a new usecase instead? 1852_cam_lt.xml perhaps
There was a problem hiding this comment.
I've moved the new 1850_cam_lt.xml to leapyears_cam_lt.xml and reverted 1850_cam_lt.xml.
My push of these changes to my fork is failing with "The requested URL returned error: 500".
Should I push it somewhere else instead?
| ./xmlchange GLC_NCPL=\$ATM_NCPL | ||
| fi | ||
| ./xmlchange RUN_STARTDATE=2012-02-29 | ||
| ./xmlchange RUN_STARTDATE=2008-02-29 |
There was a problem hiding this comment.
Why this change? Aren't these both leapyears?
There was a problem hiding this comment.
Some of the external forcing files end in 2014. Since I'm changing the increment between the forecasts from 2 to 4 years I wanted to avoid problems from requesting a year that's beyond
the time span of those files. If that was too pre-emptive, I'll try a test with 2012.
The F1850 test of the gregorian calender in ERI requires that some forcing files provide data from 1852, while others can only provide 1850.
|
@jedwards4b |
|
You can see on the conversation tab of the pull request that your commit was made before my review was made. |
| <?xml version="1.0"?> | ||
| <namelist_defaults> | ||
|
|
There was a problem hiding this comment.
I don't see any hooks to this use_case. That probably should be added with this PR or the accompanying cime PR ESMCI/cime#4899
There was a problem hiding this comment.
I don't see any hooks to this use_case. That probably should be added with this PR or the accompanying cime PR ESMCI/cime#4899
Yes, I ended up going in a different direction, which doesn't require this use_case.
I've removed it from my branch, but not committed that yet.
|
A git(hub) question; I'm trying to update my cam branch (gregorian_ERI) before committing and pushing changes. |
|
In your local clone, I would do the following: You might need to do a fresh clone of your branch to make sure that remnants from your previous attempts are no longer there. Do these commands work for you, or is this what you already did? |
outfrq9s_leapday/user_nl_clm Prevented CLM failure due to year check outfrq9s_leapday/user_nl_cice outfrq9s_leapday/user_nl_mosart Added to make the history output daily, so that comparison files exist even when the forecasts span a month end. outfrq9s_leapday/shell_scripts Kept the change of start year from 2012 to 2008 This test also requires changes to cime Replacement of modifier _CG with _cG Test with HIST compsets, not 1850 (or other spinup) These were tested using ERI_D_C3_cG_Ld8.ne3_ne3_mg37.FHISTC_LTso.derecho_intel.cam-outfrq9s_leapday. No standard resolution or B compsets have been tested yet.
3 participants
Motivation
Enable testing of the gregorian calendar for all of the start modes of a multi-instance case,
which will help ensure that DART can be used with every release (maybe tag) of CESM.
Description
The ERI test cannot use both the _CG (gregorian calendar) _C# (multi-instance) modifiers
because some conditionals in cime refer only to _C and unsuitable years are requested
from some CAM external forcing files.
This PR makes the test work for leap days.
It should require no changes to other uses of the ERI test,
but I have not run other tests because I don't know which tests to run.
Fixes
Several more changes are needed in cime, which are being handled in CIME PR #4899.
This was tested using
ERI_D_C3_CG_Ld8.ne30pg3_g17.F1850C_LTso.derecho_intel.cam-outfrq9s_leapday
bld/namelist_files/use_cases/1850_cam_lt.xml
cime_config/testdefs/testmods_dirs/cam/outfrq9s_leapday/shell_commands