DataTree tutorial with GPM_3IMERGHH_07

[eni-awowale]

I am still waiting on this to get merged pydata/xarray-data#31 so I can fetch the data with pooch. Some other ideas, I want to add an example of DataTree.from_dict to make a time series tree of the hurricane going inland. This would mean I would need to add a few more granules to the data repository. I was also thinking it could be good to add an example of open_groups.

Let me know what you all think
@TomNicholas, @shoyer, @flamingbear, @keewis, @owenlittlejohns and @aladinor

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[github-actions]

🎊 PR Preview 16cb113 has been successfully built and deployed to https://xarray-contrib-xarray-tutorial-preview-pr-307.surge.sh

🕐 Build time: 0.01s

_{🤖 By surge-preview}

[eni-awowale]

This line is going to change once this is merged pydata/xarray-data#31

[scottyhq]

Thanks for this @eni-awowale, looks great! It'll be great to have DataTree included on this website. For your notebook to be rendered, please include a link to it in the table of contents. Organizationally I think it makes sense to add the notebook under 'fundamentals' here alongside the existing data structures notebook

xarray-tutorial/_toc.yml

Lines 16 to 18 in 869be4a

	- file: fundamentals/01_data_structures.md
	sections:
	- file: fundamentals/01_datastructures

We'll have to add metpy to the environment as well.

[dcherian]

For a fundamental data structure notebook, I would skip the complex colormap setup. IME it's best to skip extraneous details, that kind of thing really confuses new users.

[eni-awowale]

Thanks @scottyhq! Is there a specific naming convention I should follow then? Maybe something like 01_datastructures_datatree

[scottyhq]

Thanks @scottyhq! Is there a specific naming convention I should follow then? Maybe something like 01_datastructures_datatree

No convention currently, whatever seems reasonable to you! Also, I recommend merging with the main branch b/c I updated the contributing guide last week with more details on the environment. Any additions you think would be useful there would be welcome:
https://github.com/xarray-contrib/xarray-tutorial/blob/main/CONTRIBUTING.md

[shoyer]

I would suggest something short and descriptive. Maybe something like "satellite_precipitation" or "imerg"?

…

On Wed, Feb 26, 2025 at 2:29 PM Scott Henderson ***@***.***> wrote: Thanks @scottyhq <https://github.com/scottyhq>! Is there a specific naming convention I should follow then? Maybe something like 01_datastructures_datatree No convention currently, whatever seems reasonable to you! Also, I recommend merging with the main branch b/c I updated the contributing guide last week with more details on the environment. Any additions you think would be useful there would be welcome: https://github.com/xarray-contrib/xarray-tutorial/blob/main/CONTRIBUTING.md — Reply to this email directly, view it on GitHub <#307 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJJFVWZXNMJEELJWX64UU32RY54NAVCNFSM6AAAAABXL3TBLCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOBWGM2TEMJVGA> . You are receiving this because you were mentioned.Message ID: ***@***.***> [image: scottyhq]*scottyhq* left a comment (xarray-contrib/xarray-tutorial#307) <#307 (comment)> Thanks @scottyhq <https://github.com/scottyhq>! Is there a specific naming convention I should follow then? Maybe something like 01_datastructures_datatree No convention currently, whatever seems reasonable to you! Also, I recommend merging with the main branch b/c I updated the contributing guide last week with more details on the environment. Any additions you think would be useful there would be welcome: https://github.com/xarray-contrib/xarray-tutorial/blob/main/CONTRIBUTING.md — Reply to this email directly, view it on GitHub <#307 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJJFVWZXNMJEELJWX64UU32RY54NAVCNFSM6AAAAABXL3TBLCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOBWGM2TEMJVGA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[TomNicholas]

Sorry for the slow input here. After @eni-awowale showed me the printed representation of this IMERG data yesterday, we discussed how ideally we would have our tutorial dataset more obviously show something that can only be done with DataTree.

My understanding of this data is that it contains multiple groups, but only as grandparent-parent-child, without multiple siblings at any level. The requirement for parent-child alignment means that by definition these 3 groups could be collapsed into one, meaning this data doesn't really need a DataTree, as opposed to just using a Dataset.

Instead, I suggested we want our example to be something that cannot even be represented without DataTree. For example by having a level with >1 sibling, containing differently sized dimensions. We could probably massage the IMERG data we have here (e.g. by up and/or downsampling) to get some data like that.