[Talk-us] Limited import of buildings and addresses for Orange County, California

Tod Fitch tod at fitchfamily.org
Mon Jul 27 00:27:36 UTC 2020


When I moved to my current city, I walked all the publicly accessible streets within several miles of my house to collect and map address information. And those buildings that I collected address data for I traced outlines using OSM compatible imagery. But there are private gated communities and areas farther from home that I was unable to access. So my mapping was far from complete.

I recently noticed the MapWithAI data available for JOSM and see it has building outlines for areas in my city. Unfortunately they are horrible, worse than nothing. But that got me looking and I found that the county has a public domain licensed dataset with building outlines, building heights, spot elevations at building locations and addresses. [1]

Looking at it, I don’t think the building outlines are good enough to be imported blindly. Each building will need to be checked and maybe up to a quarter of them will require some adjustment. That said, they appear to be much, much better than the building shapes available via MapWIthAI (which look like the Microsoft/Bing contribution).

With that background, I propose to use this data county dataset to fill in the blanks within my city. My proposed workflow (and current status) is:
Convert the county SHP file to an OSM file. Tag and value processing to convert the county’s fields to OSM tags including expanding abbreviations of street names. (Done)
Create limited extracts of above file for areas I was unable to map locally. (Done)
Work with others in the community to verify my proposed workflow makes sense and follows OSM guidelines (this email is starting that process).
Editing workflow (yet to start):
Select a street in my city that does not yet have all of its houses and/or addresses mapped.
For each house on that street in the Orange County dataset:
Verify that the building outline appears correct based on OSM usable imagery. Fix up as required.
Verify that the addr:street value makes sense based on existing OSM and Tiger 2020 road data.
If there is no node or polygon mapped in OSM, merge the building with address into OSM.
If there is an existing node, verify complete address matches, then merge building and “replace geometry” to add building outline to OSM.
If there is an existing building without address, copy the address from the Orange County data and place it on the existing building.
Upload changes to street.
Repeat for each other street in the city not yet completely mapped in OSM.
I envision this as a slow process but one that should be doable by one person since the area is reasonably small and I’ve already mapped significant parts of it.

If/when I get done with this, setting up an import for the rest of the county (breaking things into a grid of work to be done and using the tasking manager, etc.) might be in order. But I want to start small as I’ve never done this before.

Questions:
Do I need to setup a separate OSM user ID for this type of import. I will be spending as much time on review and verification on each object as I normally do for adding a building or address from a survey.
Does this limited import require a dedicated wiki page?
Should I remove the elevation from each building? I normally associate elevation tags with significant geographical/geological features (surface of a lake, summit of a mountain, etc.) having it on potentially hundreds of houses in an area seems unneeded and possibly unwise.
Older imports usually added one or more source=* or source:something=* tags to each object. Is having the attribution in the change set okay nowadays like it is for “normal” editing?
Thank you for your suggestions and guidance!

Tod

[1] https://data-ocpw.opendata.arcgis.com/datasets/8db4b58e6bbf4f6cac676f477348be48_0
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20200726/0603c7ce/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20200726/0603c7ce/attachment.sig>


More information about the Talk-us mailing list