Dark masked data in homogenous patterns about xeofs HOT 3 CLOSED

Ah! True, that makes sense if you use the p-values to mask the homogeneous patterns! :)

Sorry for the somewhat unclear docstring of heterogeneous patterns. For two data fields X and Y, the homogeneous patterns calculate the correlations corr(PC_x, X) and corr(PC_y, Y), while heterogeneous patterns are using the opposite field, i.e. corr(PC_x, Y) and corr(PC_y, X). Hope that's clear, if not don't hesitate to open another issue. I close this one as the original question was answered :)

from xeofs.

Just by looking at it: sorry no clue! If these were NaNs in the data they should be masked out consistently for different modes, but here the black patches are different for mode 2 and 4. Did you look into the underlying DataArrays to check what type of data it is? That could give you some hints on what you are looking at.
Without any minimal working example or the raw data of your analysis, I cannot offer you more feedback, sorry!

EDIT: as to plotting bars instead of lines you probably consult the matplotlib library, e.g. here. From what I know, xarray does not support bar graphs out of the box.

from xeofs.

Well, thanks for the information on bars.
I think the masked regions aren't any errors. Looks like after changing transparency I relate they are based on p-values for homogenous and heterogenous arrays

So, it all looks good, sorry for bothering you - I came back here to delete the issue.
Another point (hope it is relevant )

What does opposite i/p here mean?

from xeofs.