Add addresses per mile

[jidanni]

https://opendata.gis.utah.gov/datasets/utah-address-system-quadrants/about is great,
but each quadrant group also needs to specify:

• Addresses per mile (e.g., 800. Usually the same vertically and horizontally...)
• Odd on N and E, Even on S and W sides of street...
• Baselines are 100 or 0 (does the town have numbers below 100? Usually yes.)

Only then can this be envisioned as a more complete specification.

Otherwise it is a pity to go through all that work and leave out the small final details needed to get a "full spec".

Yes, there will be quirks in some towns. Maybe have a quirks field too to mention them.

Anyway I would do about 10 Utah towns right now for you, except this week I'm on this mountain with this itty bitty cell phone.

[steveoh]

I'm not sure I understand your concern. We haven't designed a spec for addressing authorities to follow, rather we have "reverse engineered" what already exists to be able to visualize what is reality.

[jidanni]

Yes quadrants are just one aspect. But only knowing the quadrants but and not knowing how many numbers per mile is not enough. All it is is just a fun part .... one third of the full picture.

It's just like if I made a list of how many numbers per mile each town uses. Great, but where do they start? That's where the quadrants come in.

And then there's a final aspect of which side of the street is odd and even.

With three these three components you've pretty much fully described each town.

I can't think of any more components. Any extras would be quirks.

I'm saying going through all the trouble to make the quadrants is just doing most of the work but it's a a shame not to just add the final details.

[jidanni]

It's not that tough. Just takes a couple looks at Google Street view, for each town. In a few weeks I'll do a few of them for you.

[eneemann]

This dataset was derived to capture the approximate extent of address systems/quadrants so we could use them to improve our geocoding services, which rely on the address point and road centerline data to locate specific addresses within a system. Can you explain more about how the additional data would be used/useful? Are you attempting to automate locating specific, address system-based coordinates? Or is it strictly informational? We're curious because the additional information, while nice, would require a lot of manual work to add with no clear benefit or use within our processes.

[jidanni]

Actually for my personal project,
https://www.jidanni.org/geo/house_numbering/grids/define.html
, I was go the whole hog, and tie them into the PLSS.

In other words also recording the township range and section corner where each origin point lies.

But never mind my personal interest. Just let me ask you a question, how would you predict where some 8844 W 5500 N might be? We see it is critical to know how many addresses per mile are being used. 800? 1000? Yes when the police car arrives they can easily find which side of the street the incident occurred usually, so odd versus even aren't that important. But if it's a 1,000 versus 800 mistake, that'll put them miles away.

So I guess I'm saying that your database needs to work for both geocoding and reverse geocoding. Otherwise that's an awful lot of work to only reap half the benefits. And with a little more work recording work you then have all the benefits.

Personally I think the United States, and Utah with its excellent addresses in particular, should be a beacon to other countries of how to do addresses right.,
https://www.jidanni.org/geo/house_numbering/grids/ph/ncr/bands/high.html .
But that's beside the point today.

[eneemann]

For the law enforcement situation, we have a geocoding system that relies on address points (first) and road centerlines (second) to more accurately pinpoint locations, as opposed to predicting locations using a coordinate-based system. There are inevitable skews, shifts, and imperfections in the "real world" grid, so the prediction would always have an error element, increasing with distance from the grid origin. While your suggestions are interesting and would be nice to have, there are 238 unique address systems in Utah and we don't have the resources available to pursue them at this time.

[jidanni]

I see.
OK, I'll only do one:
Minersville:

• 1000 numbers per mile,
• odd on south and west sides of streets.

Yes, one can rely on address points, except if there are no addresses yet, e.g., a new town plan.