Make `xarray` datasets discoverable about numfocusfalldev HOT 2 OPEN

We were actually thinking about the Pangeo datasets. The term "large" is subjective of course, and large enough to warrant a catalog, as in: pangeo-data/pangeo#39. We experimented with something along these lines a few weeks ago at the Pangeo workshop, https://gist.github.com/rsignell-usgs/88cfae22896bf9fed5bd36a6689e7210. The goal would be to facilitate discovery of these datasets through their attributes/metadata.

from numfocusfalldev.

There are a large and growing number of publicly-available datasets that are loadable into xarray from buckets in the Cloud.

Can you give some examples of this?

The ones I know about are the datasets we have put online in zarr format in Pangeo. (Some docs about this process here: http://pangeo.io/data.html#data-in-the-cloud). Cataloging these datasets is an open issue (pangeo-data/pangeo#39)

The current problem with hosting xarray data in the cloud is that hdf does not play well with cloud storage. This is a technical obstacle that is being discussed in many places across xarray, zarr, netCDF, etc. That's why I'm curious about your claim that there are already a large number of publicly available cloud datasets that play well with xarray.

All that said, I am supportive of this idea in general.

from numfocusfalldev.