Update landsat tutorial notebook

[jpolchlo]

Related Issue(s):

• Closes Update landsat example notebook #1056

Description:
This updates the Landsat tutorial notebook to use collection 2 data, since the old collection 1 data no longer appears on AWS. This notebook sources its data from MS planetary computer, and so requires the planetary-computer python package to be installed.

PR Checklist:

• pre-commit hooks pass locally
• Tests pass (run scripts/test)
• Documentation has been updated to reflect changes, if applicable
• This PR maintains or improves overall codebase code coverage.
• Changes are added to the CHANGELOG. See the docs for information about adding to the changelog.

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: +0.07 🎉

Comparison is base (9b363db) 91.99% compared to head (e72410b) 92.06%.

❗ Current head e72410b differs from pull request most recent head 3021242. Consider uploading reports for the commit 3021242 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1152      +/-   ##
==========================================
+ Coverage   91.99%   92.06%   +0.07%     
==========================================
  Files          51       50       -1     
  Lines        6792     6708      -84     
  Branches     1000      990      -10     
==========================================
- Hits         6248     6176      -72     
+ Misses        368      357      -11     
+ Partials      176      175       -1     

see 6 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[pjhartzell]

General comments:

• The notebook was initially written for Landsat 8. This PR uses both Landsat 8 and 9 scenes. There are a number of references to "Landsat 8" in the notebook that should be updated to just "Landsat".
• There is at least one mention of "AWS" that needs to be removed.
• The stac-browser example at the end - you can't actually browse the STAC. Or at least I couldn't. It serves up the Collection, but you can't link to the Items. But maybe that's normal for a local catalog? I don't use stac-browser enough to know.
• The notebook isn't formatted per black[jupyter]. pre-commit isn't set up quite right, so the notebooks are silently not being formatted. I'm opening a new issue for this. In the meantime, pip install black[jupyter] locally and pre-commit should apply the formatting when committing.

Macro comment:

• Getting the extra raster and classification extension requirements in the notebook is good from the standpoint of modernizing to use the latest extensions. But it reminds me of a pet peeve of mine, which are "unit" tests that try to do too much, rather than focusing on one thing. I feel like an example/tutorial of how to create STAC should certainly show how to add extensions, but only one or two simple ones. A new user has less chance of getting lost (like me wandering around test code muttering about how I can't remember what is being tested anymore) if we keep it simple. I'm not opposed to keeping it all in there, as it makes for a more complete example. But wondering if any others (@gadomski?) have thoughts.

[pjhartzell]

Looking good. Just a few comments of note.

[jpolchlo]

On your macro comment: it feels to me that this notebook is a little special. It's not just an example of how to build a STAC, but how to build a real-world STAC, in some sense. There are already examples of how to build a Catalog and how to use extensions. This is specifically about talking through the process of taking a real data set and building a Catalog for it. For this data set, that means a complete job is going to be a bit heavier in terms of the extensions that are required. I agree that examples should get the point across parsimoniously, in general, but this isn't strictly an example about how to use the API, but what it takes to pack a real data set into a complete STAC.

If the argument is that there are better places to do this, then I might not disagree. Maybe the stactools-packages repos serve this function better. In which case, the question is do we remove this notebook entirely and just add a pointer to stactools-packages into the docs instead?

W.r.t. the STAC browser, I wonder if @gadomski has a pointer about why that's not working as expected. I looked at the endpoint and saw a catalog served, which I thought was the expected outcome.

[gadomski]

A new user has less chance of getting lost (like me wandering around test code muttering about how I can't remember what is being tested anymore) if we keep it simple.

Agreed. IMO both raster and classification are a bit complex, since they live down in assets and have a hard-to-read object-in-list structure. proj is probably the best to demonstrate extensions w/o too much complexity.

The stac-browser example at the end - you can't actually browse the STAC. Or at least I couldn't. It serves up the Collection, but you can't link to the Items. But maybe that's normal for a local catalog? I don't use stac-browser enough to know.

You should be able to browse items as well, so if you can't something is mis-wired or stac-browser is broken. I'll take a look.

UPDATE: I took a look, and we should just point people to Radiant's browser instance, e.g.: https://radiantearth.github.io/stac-browser/#/external/http:/localhost:5555/LC80140322022305/LC80140322022305.json

[pjhartzell]

On your macro comment: it feels to me that this notebook is a little special. It's not just an example of how to build a STAC, but how to build a real-world STAC, in some sense. There are already examples of how to build a Catalog and how to use extensions. This is specifically about talking through the process of taking a real data set and building a Catalog for it. For this data set, that means a complete job is going to be a bit heavier in terms of the extensions that are required. I agree that examples should get the point across parsimoniously, in general, but this isn't strictly an example about how to use the API, but what it takes to pack a real data set into a complete STAC.

If the argument is that there are better places to do this, then I might not disagree. Maybe the stactools-packages repos serve this function better. In which case, the question is do we remove this notebook entirely and just add a pointer to stactools-packages into the docs instead?

I think your point is valid. Let's leave it all in there.

[pjhartzell]

I pushed a few changes to the stac-browser section. Feel free to edit as you see fit. One or two other minor comments, but looks pretty much good to go.

[jpolchlo]

Thanks for the changes to the stac-browser section, @pjhartzell ! I've incorporated the remaining comments that you offered, and I think that means this is GTG...

[pjhartzell]

Just need a quick fix for CI to pass.

pystac/validation/__init__.py:97: error: Unused "type: ignore" comment  [unused-ignore]

[jpolchlo]

Fixing this in #1166