A few points for improving the indexing notebooks

[negin513]

From today's presentation, I can think of a few things to improve the whole indexing materials:

• Advanced indexing :

• Add a schematic image on the subsetted points to explain the difference between the orthogonal vs. point-wise indexing.
  Something similar to this:

The right figure here shows it for Numpy/ poinwise indexing:
Here is another one showing the orthogonal indexing:

• Adding an exercise for this notebook.
• Adding an example for this warning:
  If an indexer is a DataArray(), its coordinates should not conflict with the selected subpart of the target array (except for the explicitly indexed dimensions with .loc/.sel). Otherwise, IndexError will be raised!

• Boolean indexing:

• Adding a plot of all grid locations to show how dims and physical coordinates are different for the curvilinear grid.

[negin513]

Here is another one I have found from Matlab:

[dcherian]

These images are great! Thanks Negin.

[negin513]

Okay, so I created this for adding to the tutorial, to better explain these concepts:

np_arr = np.arange(1, 26).reshape(5, 5)
np_arr [[0, 2, 4], [0, 2, 4]]

da = xr.DataArray(np_arr)
da [[0, 2, 4], [0, 2, 4]]

I will add these to our notebooks soon...

[scottyhq]

closed by #264 !