Implement Pangeo notebooks

[tinaok]

Identify the notebooks from pangeo-openeo-BiDS-2023/tutorial
/part3
Update and implement them.

[annefou]

for now I worked on the data access using Pangeo approach and created the corresponding notebooks for the "Data Access" section. See https://github.com/annefou/cubes-and-clouds/tree/main/lectures/2.3_data_access/exercises/_alternatives

[annefou]

Also added & tested snow cover on Pangeo@EOSC:

• Jupyter notebook: https://github.com/annefou/cubes-and-clouds/blob/main/lectures/3.1_data_processing/3.1_exercises/_alternatives/31_data_processing_pangeo.ipynb
• Rendered notebook: https://annefou.github.io/cubes-and-clouds/3.1_data_processing/3.1_exercises/_alternatives/31_data_processing_pangeo.html

We would need to test it on EOX and better integrate it into the EO-College course.

[clausmichele]

@annefou thanks for the PR!
I now had finally time to go through them and I have a couple of feedbacks and thoughts:

• Filtering with the cloud cover is also possible directly when querying the STAC Catalog. Do you think it's worth to mention or not? Notebook https://github.com/annefou/cubes-and-clouds/blob/main/lectures/2.3_data_access/exercises/_alternatives/pangeo_data_access_filter.ipynb Code to do that:

URL = "https://earth-search.aws.element84.com/v1"
catalog = pystac_client.Client.open(URL)
items = catalog.search(
    bbox=spatial_extent,
    collections=["sentinel-2-l2a"],
    query={"eo:cloud_cover":dict(lt=50)}
).item_collection()

datacube = stackstac.stack(items,
                     bounds_latlon=spatial_extent,
)
datacube

• In the reduce notebook https://github.com/annefou/cubes-and-clouds/blob/main/lectures/2.3_data_access/exercises/_alternatives/pangeo_data_access_reduce.ipynb there is still the openEO process reduce_spatial mentioned. Additionally, the groupby method is mentioned at the beginning but never used. Maybe it would be interesting to show various approaches? The xarray.DataArray.reduce is also valid and has the same name as in openEO.

[clausmichele]

Also added & tested snow cover on Pangeo@EOSC:

* Jupyter notebook: https://github.com/annefou/cubes-and-clouds/blob/main/lectures/3.1_data_processing/3.1_exercises/_alternatives/31_data_processing_pangeo.ipynb

* Rendered notebook: https://annefou.github.io/cubes-and-clouds/3.1_data_processing/3.1_exercises/_alternatives/31_data_processing_pangeo.html

We would need to test it on EOX and better integrate it into the EO-College course.

It looks really good, I like how you improved it compared to the old version.

[clausmichele]

Also added & tested snow cover on Pangeo@EOSC:

* Jupyter notebook: https://github.com/annefou/cubes-and-clouds/blob/main/lectures/3.1_data_processing/3.1_exercises/_alternatives/31_data_processing_pangeo.ipynb

* Rendered notebook: https://annefou.github.io/cubes-and-clouds/3.1_data_processing/3.1_exercises/_alternatives/31_data_processing_pangeo.html

We would need to test it on EOX and better integrate it into the EO-College course.

I have some feedbacks in this PR annefou#1

[annefou]

I have some feedbacks in this PR annefou#1

Thanks for your feedback. I would like to follow up and discuss what would be the most relevant approach for our learners.

My view is that we should decide which approach to teach and use in the exercises. Since we have already OpenEO & Pangeo, we should try not to complexity further.

Then we can elaborate and show/mention other packages in the new episode 2.4 - Data Formats and Performance.

How does it sound?

[clausmichele]

Hi @annefou, the notebooks I created in the PR wanted to try to improve the code efficiency. In my opinion, using odc-stac instead of stackstac is not increasing the complexity, it might even reduce it, because sometimes the behavior of stackstac is not what the user might expect, see like gjoseph92/stackstac#241 and gjoseph92/stackstac#196 . We can discuss about this tomorrow.

[clausmichele]

@annefou @tinaok there are still two notebooks to convert:

https://github.com/annefou/cubes-and-clouds/blob/main/lectures/3.2_validation/3.2_exercises/32_validation.ipynb

and

https://github.com/annefou/cubes-and-clouds/blob/main/lectures/3.3_data_sharing/3.3_exercises/33_data_sharing.ipynb

for the data_sharing one, we can use the raster2stac package as I mentioned here: #46 (comment)

[clausmichele]

For the data_sharing notebook: https://github.com/annefou/cubes-and-clouds/blob/main/lectures/3.3_data_sharing/3.3_exercises/33_data_sharing.ipynb

It is necessary to resue the workflow defined in the 3.1 exercise, creating the snow map for a single date. The result has to be stored as a COG, metadata about bbox and temporal coverage extracted from it and the STAC Item created.

To store the xarray object to a COG it's possible to use rioxarray with .rio.to_raster(driver="COG") (test it)