Using recipe_martin18grl.yml with CMIP6 data

[sarcol-bgs]

Apologies for the very simple question! I successfully ran this drought recipe with CMIP5 data, but when I try to replace with CMIP6 data I get the attached error. I don't know if anyone would know what is causing it?

Thank you!

Sarah

recipe_drought_CMIP6.txt
log.txt
main_log_debug.txt

[valeriupredoi]

hi @sarcol-bgs glad to see people are still running on JASMIN 😁 I'll have a look at it, and will try running your posted recipe 👍

[valeriupredoi]

@sarcol-bgs sorry for the delay, CEDA are having issues with the SOF storage today, that also hinders file access in general, but I managed to reproduce the issue you are reporting; a few pointers:

• note that your reference dataset ACCESS1-3 is CMIP5
• the diagnostic is not able to generate a correct filename for the reference because it assumes it's CMIP6
• @katjaweigel can one actually run such a variation of the recipe as Sarah here is trying to? The diagnostic is droughtindex/diag_save_spi.R and Sarah is using ACCESS1-3 as her reference dataset (CMIP5) with ACCESS-CM2 (CMIP6) as additional dataset

[valeriupredoi]

I have successfully ran the recipe by replacing ACCESS1-3 with ACCESS-CM2 (as proof of concept, it doesn't make much sense to compare same models 😁 )

[sarcol-bgs]

Ah that's interesting. I replaced with BCC-CSM2-MR and got the same error.

[valeriupredoi]

you're right! Same with UKESM1-0-LL not working - I think this is some hardcoded feature related to CMIP5 data in the diagnostic, unfortunately, since it's not outputting much of a debug output, I can't figure it out, sorry - @katjaweigel could you help out here, please, as maintainer of the diagnostic? 🍺

[sarcol-bgs]

Yep, no worries. Thanks very much for your help!

[sarcol-bgs]

Thank you for taking a look! It does seem that when I change the reference dataset to a CMIP6 dataset, it gives the same error.
log.txt

[sarcol-bgs]

Perfect, thank you, that works for me. I also got it running with the other models I mentioned.

[valeriupredoi]

very many thanks @katjaweigel 🍺 And thanks to @sarcol-bgs for persevering, would you consider this solved then? If so, we'd appreciate you marking Katja's answer as the answer, please, so we keep updated tabs on resolved discussion items. Cheers both!

[valeriupredoi]

also, Katja, is there a need for a PR that updates the recipe? It'd be very useful to have one, if that's the case :)

[katjaweigel]

Dear @valeriupredoi
Thanks for asking! If the issue is solved now it would not really require an update, maybe an update of the documentation/ comment lines to better explain how to use the reference data set here. But I plan a PR on the recipe and the diagnostic soon (because there is a new feature in it for Ranjinis paper, which will be published soon) and could include it there and include a CMIP6 example (maybe even get rid of the reference data set, but the would mean to change both diagnostics). In addition @lukruh and me have several updates, but for them it will probably take more time until we include them (but then it would fit better to remove the reference data set).

[valeriupredoi]

sounds good, take yer time, Katja 🍺