Hi @nperez
This fix is for issue #245 (closed). To calculate the per-paired climatology, the NA positions along sdate of exp and obs should be consistent. You can see the old code here, it uses extra steps to find the NA positions, which causes the error mentioned in issue 245. Because some mean functions use narm = TRUE, so if that dimension has NA, the NA is removed and the average result contains no NA anymore.
I fixed the bug, but I wonder why the old code was written in that way. Do you have any idea?
Best,
An-Chi