Add OSM / OSMnx examples

[emiliom]

From my post to Slack channel: Reminded me of how cool OSMnx seemed the first time I learned about it! I've been tickled to see how applicable the network/graph/NetworkX tools would be to handling OSM river networks and other water bodies. I've been impressed seeing the quality of coverage for smaller water bodies in OSM in some developing countries. I just saw this cool paper from the main OSMnx developer, on City Street Orientations around the World. The plot of street grid orientations in cities across the world is awesome!

• https://github.com/gboeing/osmnx
• https://osmnx.readthedocs.io/en/stable/index.html
• https://geoffboeing.com/publications/osmnx-complex-street-networks/
• https://github.com/gboeing/osmnx-examples

[emiliom]

OSMnx is focused on street networks, though it does facilitate other types of OSM data. Still, other Python APIs may be more appropriate:

• https://osmcode.org/pyosmium/
• https://github.com/osmlab/awesome-openstreetmap
• https://github.com/metaodi/osmapi
• Using the Overpass API
  • https://janakiev.com/blog/openstreetmap-with-python-and-overpass-api/
  • https://github.com/DinoTools/python-overpy
  • https://github.com/mvexel/overpass-api-python-wrapper
• Nominatim API
• https://www.giscourse.com/download-openstreetmap-data-using-quickosm-plugin-in-qgis-3-4-madeira/

Accessing natural features

• https://osmdata.openstreetmap.de/ has data downloads for land polygons, waterbodies, coastlines, etc. See also https://blog.jochentopf.com/2019-03-07-the-new-osmdata-service.html and
• https://help.openstreetmap.org/questions/66460/would-like-to-extract-all-the-waterways-in-europe
• https://github.com/ngageoint/mrgeo/wiki/osm-waterbodies
• https://gis.stackexchange.com/questions/163514/how-to-get-all-rivers-from-osm-using-overpass
• With OSNnx: https://stackoverflow.com/questions/55033816/plotting-importing-rivers-in-osmnx

[emiliom]

Nice, recent notebook from Shay, where she uses OSMnx for buildings, areas and streets: https://github.com/shaystrong/sagely/blob/master/osm_ml_training.ipynb

[emiliom]

From Friedrich, May 5:

I think the main addition to the existing tutorials is pulling in raster / vector data on the fly from open data sources, using a predefined lat lon box. This will allow us to easily switch between hotosm campaign target areas, as needed, and provide participants with a useful starting point moving beyond Geohack to explore these data at any location around the world.

Attached is a gist (srtm_example.html) of what that might look like, in this case using the elevation package to pull in SRTM raster data. The target area corresponds to hotosm task 5678 with the following bounding box [104.917171, 16.330147, 105.805508, 15.294198]

The hotosm task used to kickoff Geohack will cover a different area. Chase Stephens from OSM Seattle (CC'ed) will try to secure a clean task for us (~1 month in advance) which focuses on delineating buildings, as is the focus for task 5678.

From here I leave it to you which open data sources to pull in and how to work that into the tutorials. Some ideas on general objectives:

Vector tutorial:

• pull in OSM vector data on the fly for the specified bounding box using the OSM API
• learn how to explore the data using geopandas and shapely etc.
• groom and clean - ready for ML (again, this might just entail dropping unnecessary field and changing the projection)
• learn how to add additional points and lines to the dataframe

[emiliom]

I've placed the first version of the notebook here: https://github.com/geohackweek/tutorial_contents/blob/master/vector/notebooks/OSM_OSMnx_GHWhotosm.ipynb

It runs with a conda environment created with https://github.com/geohackweek/tutorial_contents/blob/master/vector/environment.yml