Closed juliettelavoie closed 1 year ago
CFTime indexes were causing more problems so the util check_timeindex()
converts it to a standard calendar. This was implemented in timeseries()
but not in gridmap()
. Adding it should do the trick.
If I may, have this (check_timeindex
) been tested with 360_day
data ? I think the principal line would fail (it's missing the align_on
arg). This could be ambiguous as the "best" align_on
value may differ according to the data frequency and what you're trying to do.
Also, I suggest modifying plot_coords
to stringify CFtime objects, rather than convert to numpy datetime. To be sure the plotted date is not modified or dropped. (Ex : gridmap
with data on the 29 Feb. Rare, but possible with freq='M'
).
Both types have da.time.dt.strftime
available.
Thanks @aulemahal, your suggestion looks good!
About support for 360_day
data, check_timeindex
doesn't exist for the sole purpose of plot_coords
; there was an issue with plotting timeseries with a CFtime index. So I'll revisit this in a separate issue.
I have a dataset with my time axis as cftime object, not np.datetime. This breaks
show_time
ingridmap
. It is a common type, spiro should be able to handle it.@Beauprel I can provide you with an example in private.