Another 10 minutes of checking and I’ve found some more problems (and fixed a couple of them):

1. way/665353951 had the wrong addr:street added, and a potentially incorrect postcode too. The postcode matches the street, and it matches the UPRN, but the chapel is definitely not on that street. I’ve fixed addr:street and removed addr:postcode for it in changeset/177586361

2. way/1075996208 is a bus stop but you’ve added an address to it. Bus stops can’t receive post, so I don’t think this is correct. I believe the postal_code tag is for this instead? (e.g. as used here way/162989911)

3. Same here: way/1075996234

4. UPRN added to the school building rather than the surrounding area (which is where the main address tagging is) at way/368739303

5. Address added to a slurry tank at way/922614550 — this is probably a case for postal_code too?

6. Wrong addr:street added to way/583271805. This house doesn’t have an addr:street as far as I know.

7. Same for way/583014950

8. Address details added to a substation at way/558600627, but substations can’t receive post, so I think this is another job for postal_code.

9. I’m not sure that way/525923986 is actually on Towneley Close, but would need to survey it to be sure.

10. Address and UPRN details for Booths (way/430176527) probably belong on the surrounding shop area.

11. Why does way/486427547 have two UPRNs? From Bing Streetside imagery it looks like there might be a cellar flat which hasn’t been mapped yet, but it would need a survey to confirm.

Grand :) Thanks for all your work on the map!

Thank you for fixing the canal/bridge issues.

I think if it’s not clear which of several UPRNs relates to a site or individual buildings on the site, then we shouldn’t be adding those UPRNs to OSM without finding another source of information which can clarify what each UPRN relates to.

I’ve yet to find time to review the rest of the changes in this changeset, so might have further queries later.

Yeah, the imagery isn’t super clear, but it is indeed a terrace with steps in height as the hill slopes down to the west.

As per the message I just sent you, this is why I think geometry changes should be kept separate from UPRN additions. There’s so much more to check with them to get them right, if you don’t have local knowledge.

Could you please pause and respond to the comments on your previous changesets before continuing to make the same huge sets of changes across the North West?

Adding UPRNs is undoubtedly good, but people need to be able to verify your work to trust it. Mixing 8000 changes of different types all together in a single changeset, with no information about the method you used to generate the changes (it has to be at least partially automated, right?), makes this very hard.

Are you sure it’s the petrol station roof which has the UPRN, and not something else? It seems a bit weird for a roof to have a UPRN.

This one? way/1264021493

Interesting example. It had me confused for a few minutes, until I enabled the Bing Streetside photo overlay (in ‘Map Data’ on the right in ID). This is like Google Streetview, but the license is compatible so we’re allowed to use it as a source for OSM. It’s not available on every road, but when it is available it’s very useful.

It shows that the bit of the ‘L’ which is on the wrong side of the Cadastral Parcel line is actually the garage for the right-hand semi of way/1264021492. It looks like it’s part of the next house down the road due to parallax error, but it’s actually a separate building.

Does that fit with your understanding of it?

Generally I treat the Cadastral Parcels as accurate. There have been a few instances where they don’t match the aerial imagery or my local knowledge, but in most cases where the Cadastral Parcels seem to diverge from aerial imagery, it’s the Cadastral Parcels which are right and they’re indicating something more complex is happening than what I’d first assumed.

Overall, though, I raised this in the hope of eliminating systematic error from misaligned aerial imagery when adding lots of geometry. We’re never going to make the map 100% perfect, and any errors with individual houses can always be fixed in future when someone with more information notices them.

Just in case it wasn’t clear, the Cadastral Parcels layer shows the property extents (i.e. land parcels from the Land Registry), not the building outlines. I think you figured this, but I wanted to double check!

The land parcels typically rarely change, although they can change, and they can also not line up perfectly with the aerial imagery everywhere, either where there’s an error in the Land Registry data, or someone has sold or annexed some land.

If you could link me to a specific building you’d like to query I can try and explain using that as an example, if that would help?

Heya, thanks for the additions to Lancaster, although mixing up UPRN addition, geometry changes and changes of other tags into one changeset makes it quite hard for others to review.

There are a few things I have questions about, hopefully they’re straightforward oversights!

1. Why remove addr:street from way/641817940?

2. There are several UPRNs which have been changed from those done in a recent import by rskedgell, e.g. way/563111800. Was that intentional? I haven’t looked to see which is correct.

3. As in your other changeset on Lancaster, you’ve added a UPRN to an area of the canal, way/845717554. Are you sure that’s right?

4. What Cadastral Parcels offset from the aerial imagery did you use? I assume it will have varied significantly across the city.

Ta

I came here to say the same thing! It seems odd to me that the canal would have a UPRN (and then several different ones in different places).

Heya, thanks for filling in these gaps.

You’ll have noticed that the houses already mapped recently in Seascale are offset from the aerial imagery. This is because I aligned the imagery with the OSMUK Cadastral Parcels layer before editing.

If you’re going to be adding a lot of geometry, it would be great if you could do the same. :) The aerial imagery is not guaranteed to be aligned to ground truth, and its alignment changes across the country (from one kilometer to the next) and across time as the imagery is updated. The easiest way to align geometry to ground truth is to align the imagery to the Cadastral Parcels, which are guaranteed to be aligned as they come from the government.

You can find the Cadastral Parcels layer in ‘Overlays’ under ‘Background Settings’ on the right in ID.

The offset I was using for the Bing imagery for the middle of Seascale was -1.27,-2.03 metres (see the changeset comments from my edits). The offset will likely be different at the northern and southern ends of the village.

Another advantage of using the Cadastral Parcels layer is that it can often help determine whether a building is a detached/semi/terraced house, as it will be split into separate parcels.

Hope that makes sense, happy to answer any questions you might have. Happy editing :)

Fixed in changeset/177423212, thanks

Thanks for the fast reply. I’ll partially revert the changes so the tagging gets re-added with history.

Hiya, welcome to OpenStreetMap. Why did you delete the tagging for the holiday cottages here? Have they closed down?

Their website is still running and there are reviews up from last month (https://www.holidaycottages.co.uk/cottage/79466-shearers--kirkland).

(See discussion on changeset/176898454)

(See discussion on changeset/176898454)

I’ve gone ahead and reverted this change as changeset/177419363, since the on-the-ground signage starts at the art by the picnic benches on the west side of the dock (https://www.cyclestreets.net/location/188250/cyclestreets188250.jpg).

I’ve also then made changeset/177419593 to merge the two disjoint relations for the cycle route.

Happy to discuss further if you think this is definitely not the right solution :)

Well, the big sign / artwork which lakedistrict linked to. It’s huge and says ’Bay Cycle Way’.

heh, fortunate timing with me checking osmcha this evening. I hope I didn’t cause you any problems!

I think it would make sense to have the swing bridge included in the route relation, given that the art is on the west side of the dock. It’s a big bit of art and I would be miffed to not visit it if I were cycling the route without local knowledge!

Also, somehow it looks like OSM has ended up with two disjoint route relations for it:
- relation/7024538/
- relation/14960194/