I’ve added details for the 1/4-degree and 1-degree runs that I think should be tracked in this way (i.e. major IAF runs that people will be interested in analysing, and major RYF runs that they may be interested in branching perturbations off). Nearly everything else I don’t think is worth tracking in this way (except perhaps one of Hakase’s 1-degree IAMIP runs??).
I don’t think we need to add the full metadata.yaml description here. I feel that this is a place to give a short overview of the runs available and have links to where the data is and where to find the configs if you want to rerun/run a perturbation.
Some details, which i don’t know, from the 1/10th are still missing.
I’m not sure how many experiments used sync_data.sh to copy the outputs from /scratch, but those that did will have a run summary .csv and a clone of the git history (git-runlog) in the same location as the output file. This should provide everything that’s needed to re-do or branch a run from any point.
@aekiss in other words, we should all just read the instructions! That automatic git-runlog sync is great.
I’ve added a short paragraph at the beginning of the tutorial indicating different ways in which the configuration files for the previous experiment can be found. See the start of Tutorials · COSIMA/access-om2 Wiki · GitHub
Of course, the git-runlog won’t help if people aren’t using sync_data.sh. A quick check with find suggests that there are quite a few experiments where this is the case, including important runs such as 01deg_jra55v13_ryf9091 and all the 1-degree OMIP runs. These should be fixed if possible.
I’m also updating /g/data/hh5/tmp/cosima/access-om2-run-summaries for the first time in years. The list of runs summarised is in do-run-summaries. This is far from comprehensive and I’m not sure it’s the best approach.