Combining different realizations from different experiments for the same model

[malininae]

@npgillett asked me a question but I don't know the answer, so maybe someone else has a solution. In GISS-E2-1-G for DAMIP experiment the runs were set up in such a way, that one needs to merge r1i1p1f3 hist-nat (1850-2020) with r1i1p1f2 ssp245-nat (2021 onwards) to obtain the continuous dataset for 1850-2100. Now, the problem is, that we haven't found an ESMValTool syntaxis to combine those two experiments to combine them in the preprocessor. - { dataset: GISS-E2-1-G, exp: [hist-nat, ssp245-nat], ensemble: [r1i1p1f3, r1i1p1f2], grid: gn } doesn't seem to do the trick. Any suggestions?

[tepmo42]

@valeriupredoi can you help with this?

[valeriupredoi]

Hi Liza, so either [hist-nat, ssp245-nat] or [r1i1p1f3, r1i1p1f2] works, but not both. We can probably plug in functionality for both, @bouweandela @schlunma was there a particular reason why we didn't go the extra mile and get them both to work at the same time? I can't remember, but then again, am old 😁

[malininae]

Thanks for the reply, that's what I suspected. I think it is relatively exotic to want to combine [hist-nat, ssp245-nat] and [r1i1p1f3, r1i1p1f2], so I'm not sure that is worth implementing just for us.

[npgillett]

Thanks for the replies! In favour of adding the functionality, one of the examples on this page indicates that you can do this:

datasets:
  - {dataset: CanESM2, project: CMIP5, exp: [historical, rcp85], ensemble: [r1i1p1, r1i2p1], start_year: 2001, end_year: 2004}

Though I suppose you could also revise the documentation!

[valeriupredoi]

@npgillett good catch, and thanks for the docs pointer! We should probably revise that, but before I do that (or, alternatively, implement the combo feature), Liza, would it be possible to get a full stack traceback of the error you get by using that in your recipe, please? 🍺

[bouweandela]

This used to work with older versions of the tool.

[malininae]

@valeriupredoi Sorry, forgot to do it. Here's the output. Nathan tried it on the older version of core/tool, I tried at 2.10 core and still have the same error. @npgillett tried with and without the quotes, but still the output is the same. Thanks for looking into it.
nathans_recipe_main_log_debug.txt
nathans_recipe_test_yml.txt

[bouweandela]

That looks like a bug in the code that is combining multiple ensemble members of the same dataset into a single entry for the recipe without wildcards and other magic that is written out for reproducibility.