Cleanup of CatchCN restart routines #629
Conversation
|
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged. |
1 similar comment
|
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged. |
|
This PR is being prevented from merging because you have not added one of our required labels: 0 diff, 0 diff trivial, Non 0-diff, 0 diff structural, 0-diff trivial, Not 0-diff, 0-diff, automatic, 0-diff uncoupled. Please add one so that the PR can be merged. |
2 similar comments
|
This PR is being prevented from merging because you have not added one of our required labels: 0 diff, 0 diff trivial, Non 0-diff, 0 diff structural, 0-diff trivial, Not 0-diff, 0-diff, automatic, 0-diff uncoupled. Please add one so that the PR can be merged. |
|
This PR is being prevented from merging because you have not added one of our required labels: 0 diff, 0 diff trivial, Non 0-diff, 0 diff structural, 0-diff trivial, Not 0-diff, 0-diff, automatic, 0-diff uncoupled. Please add one so that the PR can be merged. |
|
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged. |
|
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged. |
|
Nightly tests summary @gmao-rreichle, @weiyuan-jiang : Comparing LDAS output for nc4 files failed, almost all vars in "lnd" and "lfs" collections differ (for global run 48/56 differ):
For example SFMC maximum with branch went up for 0.00008. All looks in realm of roundoff. Comparing for restarts is zero diff. GEOSgcm develop vs branch (with new MAPL) is zero diff. |
|
Just adding for extra information that the paired GEOSgcm_GridComp branch lacked an update to the most recent 'develop' branch reflecting changes that Anton made. This is likely the cause of the differences. I just merged develop into my branch again and we are currently re-running the tests. |
|
Not sure why Anton's changes (new IGNI GridComp) should give us non-0-diff with GEOSldas runs. It was successfully tested 0-diff for the GCM (I assume) and should have no effect unless DO_FIRE_DANGER is set to .true. |
|
I've run clean set of tests now, after Jana's update to sync with develop, and with MAPL v2.35.1. |
|
@biljanaorescanin, please run a GEOSldas test using GEOSldas "develop" with just one change -- to MAPL 2.35. That'll tell is if it's Jana's change or MAPL. I fear it's the latter. |
|
Yeah, Jana suggested same thing I am setting that now. Stay tuned. |
|
CONUS just failed for only MAPL v2.35.1 and develop, it ailed same way as on the branch for "lnd" and "lfs" comparisons. I will leave all tests to finish but I expect all fails we have seen here are due to MAPL. Not sure but few things here could be candidates for changes we see @weiyuan-jiang: |
MAPL v2.35.1 (and presumably also v2.35.0) is apparently not zero-diff for GEOSldas, contradicting the MAPL release notes. I suspect we skipped testing MAPL v2.35 with GEOSldas. Or didn't we? The challenge now is to understand which of the MAPL changes caused the diffs, to determine if the revisions in MAPL can/should be zero-diff by patching MAPL v2.35, and -- if non-0-diff is the only option -- to verify that the diffs are acceptable ("roundoff"). |
|
I didn't test MAPL v2.35.0. |
|
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged. |
|
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged. |
1 similar comment
|
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged. |
This PR addresses GEOSgcm_GridComp issue #667 (https://github.com/GEOS-ESM/GEOSgcm_GridComp/issues/667) and is paired with GEOSgcm_GridComp PR #682 (https://github.com/GEOS-ESM/GEOSgcm_GridComp/pull/682).
It removes the duplication of Catchment-CN specific constants in CatchmentCNRst.F90 and instead imports them through use-statements. It also removes the duplication of orbital constants and replaces the use of hard-coded orbital subroutines with the relevant MAPL subroutines.
This PR should be zero-diff for Catchment and is zero-diff for Catchment-CN when used with
RESTART: 1. It is not zero-diff for Catchment-CN used withRESTART: 2. When running Catchment-CN withRESTART: 2, there are differences in the day length variable (dayx) with maximum differences on the order of 10 seconds.UPDATED 7 Mar 2023 by @gmao-rreichle:
PR is not 0-diff for diagnostic (HISTORY) output from all current "model" GEOSldas tests because of the update to MAPL v2.35.2.
MAPL v2.35 fixed a bug in the bit shaving tool, and the GEOSldas "model" tests happen to use bit shaving.
Restarts are 0-diff (except for simulations with CatchCN using
RESTART: 2as noted above.