Matching time coordinates in different variables

[morobking]

I have several different variables I load in my ESMValTool diagnostic. However, one of them has missing time steps so only has a time coordinate of length 22 not 24 like all the others.
Is there a way in ESMValTool to a) check for this, maybe as a preprocessor? and b) correct it so the cube of data has the same time coordinate as the other data but with 'fill value' as the data at its missing times?

[valeriupredoi]

yes, regrid_time just recently overhauled by @schlunma to account for exactly these sort of corner cases, a preprocessor from esmvalcore._time - that can be used to fix that, Rob. To check is a bit of a faff since those time steps may actually be correct by CMOR standards

[schlunma]

Unfortunately regrid_time cannot help you if you have a different number of time points 😬. It will NOT add missing values or change the data in any other way, but simply set the time points to a common value.

[valeriupredoi]

ah yes, I realized that there are actually missing time points after I've written the reply, but thought it's still useful since masking would probably be easier once all time points are on the same axis - or, would using regrid_time would actually be destructive in this case ie may move time points in positions where they actually are missing?

[schlunma]

Hmm, good question. That probably depends on the actual data 🤔

[bouweandela]

check for this, maybe as a preprocessor

Missing coordinate points are checked as part of the CMOR checks performed by the preprocessor. So unless the points are missing at the start or end of the range, I would expect that you get an error about a discontinuous coordinate. Is that not happening?