[Imports] [Talk-us-newyork] Draft proposal for import of New York State GIS SAM Address Points

[Skyler Hawthorne]

Jan 9, 2021 12:49:02 Kevin Kenny <kevin.b.kenny at gmail.com>:

> I still have some questions about conflation, because there are some odd cases that are pretty common in that data set.
> 
> There's some delay in actually propagating updates into the distributed E911 address points.  For instance, the addresses along https://www.openstreetmap.org/way/813453708 were surely not in the database when I mapped them. The subdivision isn't finished yet, and I have no aerials showing it.  I put on my "urban camouflage" (orange vest, hard hat, work boots, clipboard) and walked through. It was sort-of open to the public. The roads were done and there were a couple of model houses built. You could stop in and get shown around. I was followed for a while by a late-model Lincoln driven by a lady who had 'realtor' written all over her. Eventually she got bored.  Every building lot was marked with a stake with the house number, and I snagged coordinates from those.  How likely am I to have caused trouble down the road with failed conflations?

Unlikely. If you added the addresses within 300m of where they end up in the dataset, it will find it and skip it. It is a goal to leave existing addresses alone, except when the change would be purely additive, i.e. the existing address has a house number, but no units.

> Apartment and condominium complexes are mapped inconsistently.  For instance, the building numbers for https://www.openstreetmap.org/way/491144551 are just strung out at more-or-less equally-spaced intervals on the entrance road. The housenumbers were visible on the buildings, and I mapped them on the buildings. Likely to cause trouble?

I took a look at this area and ran my program on it; the only changes to these buildings are to add missing units.

> (Also, there's a spurious "1515 Hillside Avenue" on the junction of the entrance road.

Yeah, so this is an example of an address point that is on a driveway. They actually have a special "PointType" value of 3, which is defined by them as the "driveway" type. I think it's still useful to have.

> 
> The next subdivision over, https://www.openstreetmap.org/way/491144554, has the county showing neat clusters of address points - that are not even close to being aligned with the buildings. The building numbers stop with 608. Christian Court is not shown, and there's a spurious "1503 Hillside Avenue" on the entrance road.

This was a good one. With this example, I actually found a bug in my conversion code! I had not encountered this tagging practice before of putting multiple house numbers on one building, so it was adding address points when it should have seen that they existed already. I just fixed this. After the fix, there are no changes to the existing buildings, and the address points are skipped.

(Actually it does add 3 of the points because this building's tags are typo'd and used . instead of ; https://www.openstreetmap.org/way/491160758)

> 
> The next one over, https://www.openstreetmap.org/way/553341585 has a spurious '1453 Hillside Avenue' in the middle of the complex. I know that  the buildings have addresses on Alice Wagner Way, that are not shown. I didn't trouble to record them when I walked the neighbourhood, because ordinarily I can recover housenumbers from the county. Not this time! One of these months I'll make it back in there and grab them. 

This is another driveway point. The building is somewhere down this road.

> 
> 
> There are also some fantasies in the data.  In the vicinity of https://www.openstreetmap.org/way/553341585 there is a chain of points strung out along an entirely fictitious 'Greylock Avenue' (which I presume belongs to a subdivision for which planning permission was refused. Some of the nodes are in land that now belongs to the township.) The varous government GIS departments have goofed here as well. There are two buildings with Becker Drive addresses and one with an Alandale Avenue address that are misidentified as being on this fictitious street. (Caveat - I actually need to get in there and recheck. I'm quite certain that the street does not exist, but less certain that there are no buildings with addresses on it.)

I think you may have typo'd the way there? Otherwise they moved the points. The only ones I found in the dataset are near 42.7696561988882, -73.8841511470806

You can download this zip to see the .osm of existing relevant data, the .osc that was generated, and the log of addresses that were skipped because they were found in existing data.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20210110/54e100ad/attachment.htm>