Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Salvage cmip5-ng #90

Merged
merged 4 commits into from
Aug 26, 2021
Merged

Salvage cmip5-ng #90

merged 4 commits into from
Aug 26, 2021

Conversation

leabeusch
Copy link
Collaborator

@leabeusch leabeusch commented Aug 24, 2021

Fixes issues with reading in data from cmip5-ng archive for all RCPs

  • Introduces use_cftime=True to deal with incompatible calendar for default xarray approach (-> thanks to @mathause for suggesting this solution!)
  • Adds ability to deal with fact that r1i1 member does not always exist (which if exists is always moved to first position due to legacy reasons)
  • Removes tas run with bug in time axis in scenario that wasn't used before

leabeusch and others added 4 commits August 24, 2021 18:52
The time encoding in cmip5-ng is apparently not compatible with the
default approach used by xarray 0.17.0 to read in data. Hence, the use_cftime
argument needed to be introduced.
In case no *r1i1* run exists leave the runs in their original order.
Only reason *r1i1* is moved to front is to make it possible to compute
anomalies by baselining every run with respect to the first run (i.e.,
a legacy from the original MESMER description paper approach of training
on a single run).
r14 of rcp45 of EC-EARTH has a faulty time axis in the cmip5-ng archive
with 10 years always sharing the same time stamp. Bug is reported which
will likely eventually lead to other changes (i.e., this exclusion will
become irrelevant again because either the run will be removed from the
archive or it will be fixed).
@mathause
Copy link
Member

Looks ok to me. The changes should not cause a problem for #55 I think - we don't touch the same functions. Tests would be nice but testing IO is one of the more difficult things to do so I'd leave it as is.

@leabeusch
Copy link
Collaborator Author

Great, thanks! I removed the test item from the task list now (considering the effort that would be needed to test this in the current code setup). & just for future reference: In a later PR, we should replace the part of the code that moves the r1i1 run to the first position (if r1i1 exists) with the cleaner sorting solution proposed by you in issue #91.

@leabeusch leabeusch merged commit fbd734d into master Aug 26, 2021
@leabeusch leabeusch deleted the salvage_cmip5ng branch August 26, 2021 09:17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

does loading cmip5 data actually work?
2 participants