Opened 6 years ago

Closed 3 years ago

#1314 closed Enhancement (wontfix)

Implement assimilation interface SETTE test

Reported by: djlea Owned by: djlea
Priority: low Milestone:
Component: OCE Version: trunk
Severity: Keywords:
Cc: Review:
MP ready?:
Progress:

Description

Create a SETTE test configuration which includes observation operator and increments tests, to ensure that future NEMO developments do not break OBS and ASM.

Commit History (1)

ChangesetAuthorTimeChangeLog
4727djlea2014-07-24T18:40:12+02:00

#1314: Creating a new branch to contain developments to SETTE for testing the assimilation interface.

Change History (6)

comment:3 Changed 6 years ago by djlea

Revision 4751 contains initial development of an observation operator test for the assimilation code. The test observations are added using "fbgenerate" utility (also added to branch in OBSTOOLS).

comment:4 Changed 6 years ago by djlea

Summary of changes:

  1. New SETTE test added: ORCA2_LIM_OBS which tests NEMO OBS and ASM with simulated observations and increments. The intention is that the observation and increments files will also be added to the standard "FORCING" data. (I can provide them by email on request also).
  1. The observations were created with fbgenerate which is also added to TOOLS/OBSTOOLS. The increments are generated using some python code (added to TOOLS/OBSTOOLS/python). TO DO: to document these new utilities in the NEMO book.
  1. The standard ORCA2_LIM and ORCA2_LIM3 builds have the keys_diaobs and key_asminc set to try to prevent future developments from inadvertently breaking at least the builds of OBS and ASM.
  1. The reference namelist is updated to prevent runtime errors with key_diaobs and key_asminc set.
  1. Default array settings added back into diaobs since the removal of these caused serious errors.
  1. Changed to code to use ctl_stop in a couple of places to match standards elsewhere in the code.
  1. Fix to jul2greg to use FLOOR rather than INT. This change was motivated by a problem which arose when running the test example. This runs on the 1 Jan 1AD which implies a negative "julian" date. Before this change "julian" days close to midnight were being put into the wrong day. (E.g. -711855.000039 should be Jan 4, 23:59:59 but was instead being output as Jan 4, 0:00:00).
  1. To aid with SETTE testing of OBS and ASM I added some output (to ocean.output) of the overall mean and RMS observation model differences for the various observation types (done in obs_write).

comment:5 Changed 6 years ago by djlea

Summary of tests:

Performed the SETTE tests detailed in the review wiki: ​https://forge.ipsl.jussieu.fr/nemo/wiki/2014WP/ticket/1314/Review

The new developments all build and run without errors and perform in the expected way.

Last edited 6 years ago by djlea (previous) (diff)

comment:6 Changed 3 years ago by clevy

  • Resolution set to wontfix
  • Status changed from new to closed

Suppressing tickets opened on the trunk since more than 2 years… cleaning started!

Note: See TracTickets for help on using tickets.