Add documentation for distant AOIs for grid generation

[jtmiclat]

Closes #35
Closes #59
Closes #60

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[github-actions]

Visit the preview URL for this PR (updated for commit 162909d):

https://geowrangler--pr55-docs-document-grid-p-q6ichi76.web.app

_{(expires Thu, 07 Jul 2022 01:59:27 GMT)}

_{🔥 via Firebase Hosting GitHub Action 🌎}

[jtmiclat]

Hi @joshuacortez, Can you check if this aligns with what you mentioned in #35
https://geowrangler--pr55-docs-document-grid-p-q6ichi76.web.app/tutorial.grids.html#Generating-grids-for-multiple-distant-areas

[joshuacortez]

Hello @jtmiclat , tried to run the colab notebook but encountered an error when loading the admin_grids geojson. It couldn't find the file in the data directory. Thanks!

grid_gdf = gpd.read_file("../data/region3_admin_grids.geojson")

[jtmiclat]

@joshuacortez Fixed! Pls, try it out again. Please use this link as the button that opens collab points to master and not this branch https://colab.research.google.com/github/thinkingmachines/geowrangler/blob/docs/document-grid-perfomance/notebooks/tutorial.grids.ipynb

[joshuacortez]

Thanks for the corrections!

I looked at it and the example of having a ~50 square AOIs (divided into large cells of 1km) might not be representative of some actual use cases? Curious about the performance of dividing irregularly shaped polygons across a country into granular cells (e.g. 100m instead of 1km). That's where performance concerns (speed, memory) might be more apparent.

I can help out with this if ever too.

[jtmiclat]

@joshuacortez Can you provide some irregular polygons we can try?

[joshuacortez]

Alright, provided test polygon AOIs in this GPKG file

Parameters I used:

• PHL bounding box as custom boundaries [-236221.24150117, 507465.65086805, 1119703.39754333,
  2336815.97643073] in epsg:3123
• cell_size of 100 meters

Some benchmarked runtime results using previous code:

• ~40 sec for the first 100 AOIs
• ~43 min for all the AOIs

[jtmiclat]

@joshuacortez I ran the benchmark on colab (Single core with ~14GB ram). It is very similar to what you mentioned.

[joshuacortez]

Awesome, thanks @jtmiclat! This looks pretty good. I ran it on a local machine with similar specs

[joshuacortez]

I wonder if it makes sense to have a function that takes in a geodataframe of multiple AOIs instead of manually iterating through each AOI and making separate geodataframes. I don't feel too strongly about this but might look "cleaner" from a user perspective

[jtmiclat]

I wonder if it makes sense to have a function that takes in a geodataframe of multiple AOIs instead of manually iterating through each AOI and making separate geodataframes. I don't feel too strongly about this but might look "cleaner" from a user perspective

Moving this to the discussion!
#63