issue_comments: 687336494

I find xarray so useful in many ways, for which I'm grateful. But there are some current limitations that force me to hesitate before recommending it to colleagues. One is this issue - lack of support (or rather, I suspect simply no "clearly stated support"?) for curvilinear coordinate systems, which are pretty much ubiquitous in the work I do. The other issue which causes me to pause before recommending xarray wholeheartedly is the complexity (and frequent slowness and errors, still - all previously reported) in dealing with GRIB2 file formats that include multiple vertical coordinate systems (e.g., products from the NCEP Unified Post Processing System - UPP). But that's an issue for another thread... Any movement on wrapping scipy griddata (or some suitably more sophisticated scipy tool) within xarray's interface?