While working on #156 and per discussion with @dcherian, I learned that esmlab's codebase is predominantly full of hacks.

I don't know the code that well at all (!) but this stuff looks sane.

I myself have started having a hard time understanding the existing codebase. It's been a while since I looked at esmlab's code base. So, I spent the last four days looking at the existing code base, and my take-away is that esmlab's codebase is full of hacks. And this makes it difficult to debug or even maintain :(

I am personally in favor of helping with pushing most of the general functionality into xarray and only keeping things that are domain-specific. I just found out that there has been recent progress in pydata/xarray#2922, and I am excited about this. Once pydata/xarray#2922 is merged, we can move most if not all existing functionality from https://github.com/NCAR/esmlab/blob/master/esmlab/statistics.py into xarray.

Originally posted by @andersy005 in #156 (comment)