Add the ocx experiment - #38
Conversation
The observationally constrained experiment (C011) was missing from experiments_ismip7.csv, so the checker refused every ocx file. It now starts in any year from 1950, the earliest year of any OCX forcing, to 2015, and ends in 2025. OCX has no ESM: its atmosphere comes from ERA5, the Greenland ocean from EN4 and the Antarctic ocean from expert judgement. An ocx file may leave the ESM field out or put any name there but a CMIP model's. A name without the field is normalized to ten fields with an empty ESM by _file_name_fields, which every place that reads fields by index now uses, and the companion lookup rebuilds the name from the file's own fields so that it keeps the omission. Like historical, ocx chooses its own start year, so litemp must carry a snapshot at the run's first year as well as its last. The generator writes that snapshot, and leaves the ESM field out when given an empty ESM id. An experiment named in the wrong case (OCX, as the forcing directories spell it) is now told that experiment names are lower case. The 9-field match is case-insensitive so that such a file reaches that hint rather than a field-count error. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Testing
Posted by Claude Code on @xylar's behalf. The testing, analysis and wording above are AI-authored; please check them accordingly. |
|
@hgoelzer, I'd really like you to take a look at this if you can. The "ESM" field for OCX is a particular problem. I decided to allow it to be anything other than a CMIP model or nothing at all. But would appreciate your thoughts. We clearly should have been on this much earlier! |
|
Hi Xylar, I think about two option for the ESM in OCX.
1 would cover all modellers that have already submitted/finished their CORE. If we think of ESM in loose terms as "the model providing the climate forcing", a lot of options could be possible here. Therefore leaving it open to allow for those options. We provide products (EN4, ERA), but should be open for that people chose something else, like another reanalysis product. 2 would make the it easier to parse the OCX in downstream processing. All other CORE exps have a well defined ESM and having that per-model information in the filename doesn't really help. Since modellers as of today do not know how to name it, prescribing something at this stage seems still possible. People that want to be using the checker right now for OCX are arguable still busy with post-processing and could adjust. |
|
Okay, it sounds like skipping the ESM field will make downstream workflows harder. Let's not allow that. I lean toward option 1 where we leave it open. |
hgoelzer
left a comment
hgoelzer
left a comment
There was a problem hiding this comment.
OK, go ahead with ESM left open.
Following the discussion on ismip#38: leaving the field out would make ocx names harder to parse downstream, and groups may use forcing other than the products ISMIP7 provides, so an ocx file carries the ESM field with any name in it, CMIP model names included. The name must be letters, digits and hyphens, which rules out an empty field and the dots other fields of the convention also forbid. This drops _file_name_fields and the generator's empty ESM id, since a name always has ten fields again. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2 participants
The checker refused every file of the observationally constrained experiment (C011) because ocx was missing from experiments_ismip7.csv. A modeler reported this. This PR adds ocx with a start in any year from 1950 to 2015 and an end in 2025, and bumps the version to 0.6.0.
For review:
Other changes:
Posted by Claude Code on @xylar's behalf. The testing, analysis and wording above are AI-authored; please check them accordingly.