recipe_extreme_index.yml : trying to apply on monthly datasets

[mpmoine]

Hi,

I'm new there, it's my first SOS discussion ;-)

I am testing recipe_extreme_index.yml. It works well on daily datasets but when trying to apply to monthly data, I get the following error (that occurres whatever the CMIP6 model source I use):

Error in dim(historical_data) <- c(model = 1, var = 1, lon = length(lon),  :
  dims [product 9058400] do not match the length of object [297600]
Execution halted

From the doc, I undertood that il should be possible to apply to monthly data. Did I missunderstood?

Thanks in advance.

[valeriupredoi]

Hi @mpmoine and welcome to ESMValTool! Unfortunately I am a technical maintainer, and don't know a great deal about this recipe, but let me try ping some people that may know for you @schlunma @sloosvel @LisaBock @katjaweigel Also, it'd be good to attach the full log (main log or main log debug) so we can see where the stack trace starts and how it evolves 🍺

[mpmoine]

Thanks @valeriupredoi for your prompt reply! 👍
Here are the log files!
slurm-31492.out.txt
log.txt

[katjaweigel]

Hi @mpmoine
I tried to run it, changing only mip: day to Amon in the original recipe and got the same error like you.
I'm afraid the documentation text is wrong, which states, that any time step should be possible.
The R script called in this recipe (magic_bsc/extreme_index.R) contains the line (104):
time <- seq(start_date, end_date, "day")
which I'd assume to work only for daily data. Some of the indices used are in any way defined for daily data (rx5day, consecutive dry days (cdd)). Is it important for you to use monthly data / which index are you aiming at? Or can you use daily data and we should only update the documentation? Since the recipe is labels "unmaintained" I'm not sure if the any of the authors, e.g @nperezzanon @alasdairhunter @nmanuben could help?

[mpmoine]

Hi @katjaweigel
Many thanks for testing and for the confirmation.
That's fine for me to run this recipe on daily data. An update of the documentation sounds largely sufficient!
Thanks!

[valeriupredoi]

cheers fro stopping by @katjaweigel 🍺 @mpmoine would you be kind enough to open an issue about fixing the documentation please - I can assign people after that, and also if you could mark the question answered - we are tracking the answered/unanswered ones. Many thanks both!

[mpmoine]

@valeriupredoi : issue just posted! This discussion can be closed now.
thanks to both of you @valeriupredoi @katjaweigel