[Talk-ca] Importing buildings in Canada

[Nate Wessel]

Hi all,

Daniel, thanks for continuing to prod the conversation along :-)

I guess the point of disagreement here is that I generally don't agree 
that the ODB buildings are of higher quality than what's already in OSM. 
The ODB data I've seen is quite messy, and IMO only marginally better 
than having no data in an area, especially if not properly checked and 
conflated with surrounding OSM data. People seem to generally disagree 
with that perspective though, so I'll assume that other areas are better 
represented by their respective ODB data sources. Central Toronto may 
just not have been well mapped by the City's GIS dept; it certainly 
isn't the easiest thing to get right.

My real worry here is that someone will be carelessly going through an 
import replacing geometries and will destroy the work of an editor like 
myself who carefully contributed their time to make a neat and accurate 
map. I know for certain I've contributed better data in Toronto than 
what's available from government sources for the same area.

We must recall that governments produce building footprints in the same 
way that we do - usually by tracing imagery, and there is little reason 
to suppose that their data is better or more accurate just because it 
comes from a seemingly authoritative source. It comes from interns - 
likely interns with outdated software and low-resolution surveys.

All that being said, I think my real reluctance to go with the flow here 
stems from the haste and carelessness of the original importers in the 
GTA. We're working toward a process that should be very different from 
theirs though and I probably need to just trust that our process will be 
calmer, slower, and more thoughtful. If it is, I can get on board.

But also - sorry this is such a long email - we certainly should not be 
manually doing re-conflation where buildings are very dense (e.g. 
downtown Toronto) or where they have already been imported extensively 
(much of the rest of the GTA). The import plan I'm working on for 
Toronto will take care of most of this automatically by importing only 
in the sparse gaps between existing OSM buildings. For other parts of 
Canada, this may not be much of an issue at all - I wouldn't know.

Best,

Nate Wessel, PhD
Planner, Cartographer, Transport Nerd
NateWessel.com <https://www.natewessel.com>

On 2020-01-18 5:24 p.m., Daniel @jfd553 wrote:
>
> Bonjour groupe,
>
> Here is a sequential summary of the last exchanges. I inserted some 
> comments […] within these exchanges description and summarize what I 
> understand from it at the end.
>
> *Nate*asked not to confuse the process of importing new data with that 
> of updating/modifying existing OSM data in order to keep things simple 
> for this import.
>
> *Daniel*(I) responded that importing new data and updating/modifying 
> existing ones at the same time (when necessary) is not unusual in OSM 
> [/and would be more efficient/].
>
> *John*replied that importing new data and updating/modify existing 
> data when required worked quite nicely when importing Ottawa.
>
> *Nate *explained he believes that the buildings will not be compared 
> manually since there are hundreds of thousands of them in OSM for 
> Toronto alone. In other words, he thinks there will be automated 
> edits, and these edits are not governed by the same policies as 
> imports. [/This is an important consideration. What has happened in 
> Ottawa and Toronto so far? Have automatic processes been used?/]
>
> *Tim*replied that in most cases it might be appropriate to replace OSM 
> data, as he believes [/as I do for most of the cases/] that the ODB 
> footprints will be more accurate than existing buildings. However he 
> considers it is a case-by-case decision [/then no automation process 
> should be used/].
>
> *John*couldn’t resist digressing toward the “Microsoft buildings 
> import” but had to bring back the discussion on ODB import after 
> reactions from *Tim* and *Pierre* (LOL).
>
> I think that, somehow, we could all agree on how the import should be 
> done:
>
> -Align images on ODB, unless evidence ODB data are ill located.
>
> -Align existing OSM content with the image, *only* if necessary after 
> aligning the image with ODB.
>
> -Import non-existent buildings in OSM.
>
> -Conflate ODB and OSM buildings, *only* if necessary.
>
> -
>
> To address Nate’s legitimate concerns, we could agree that there will 
> be *no* automatic changes/conflation of existing buildings. Having a 
> local import manager and by using the task manager, we should be able 
> to ensure that there will be no unauthorized import (i.e. not 
> responding to the above).
>
> Am I too optimistic?
>
> Daniel
>
>
> _______________________________________________
> Talk-ca mailing list
> Talk-ca at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ca
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20200118/052ab747/attachment.htm>