Confusing title, but (hopefully) relatively simple problem. See here.
Wellfield Bungalows is a set of four houses (1–4 Wellfield Bungalows) on Wellfield Road. On the same road, house numbers 1–4 Wellfield Road also exist. This creates, for example, 1 Wellfield Road and 1 Wellfield Bungalows, Wellfield Road. I would like to know the best practice for tagging them so they are not confused with each other.
At present: 1 Wellfield Road is tagged normally: addr:housenumber=1; addr:street=Wellfield Road. 1 Wellfield Bungalows is tagged: addr:housename=Wellfield Bungalows; addr:unit=1; addr:street=Wellfield Road.
Note:Nominatim does not return any results for “1 Wellfield Bungalows” with these tags.
If I instead tag the bungalow with addr:housenumber and addr:housename then, according to Addresses in the United Kingdom, the address will resolve to Wellfield Bungalows, 1 Wellfield Road which I assume would be conflated with the house and would be incorrect.
Should I move both the number and name into addr:unit or addr:housename (e.g., addr:unit=1 Wellfield Bungalows), or is there another solution? I imagine this problem is solved in a similar way groups of flats/apartments are tagged, but I have not yet learned how to tag them.
This is a clear case of an addr:substreet. As the address would be written as e.g. 3 Wellfield Bungalows, Wellfield Road, this means that ‘Wellfield Bungalows’ is not the housename of an individual house, but a grouping before the street name. I would tag as follows: addr:housenumber=3, addr:substreet=Wellfield Bungalows, addr:street=Wellfield Road. You can see an example of one I’ve mapped already here Way: 23 St Albans Road (794946111) | OpenStreetMap
There is also more information about addr:substreet on the Addresses in the United Kingdom Wiki page. This advice doesn’t apply to apartment buildings, where we use addr:housename and addr:flats instead.
I’m not sure it is. ‘Wellfield Bungalows’ is not a housename, as it is shared between 4 separate properties. & addr:unit is usually used for industrial estates and the like. I’ve not seen it used for residential properties before.
Bad practice I’m sure, but I’ve edited each four of the bungalows with a different schema with the aim to test which addresses populate correctly when searched in Nominatim. I wonder if there are other ways to validate the addresses.
Nominatim is just one consumer of the data. Just because one way of tagging works better than the others does not make it the correct one. Apps like CoMaps do not acknowledge the existence of addr:substreet, but we should not change our tagging to accommodate that.
I wouldn’t say addr:unit is restricted to industrial buildings. For me it comes down to what it looks most like. If it’s a series of buildings then the substreet is probably the best way to go. But if it’s a set of addresses in a large named building then addr:unit is preferred.
Although to be honest, any mapping that results in a correct looking adress as per the wiki page (and linked envelopes tool) is better than not mapping addresses.
Great! Using the Address Envelopes map and sanity-checking with Nominatim, I’m happy that using addr:substreet is the solution here. Thank you all for your input
Last night I was attempting to map Tudor Court and George Lee Court flats (only mentioning them if you’d like to check my work ) and, while searching for tagging help, came across this post by @Robert_Whittaker.
For a similar scenario, they suggested either one of these tag layouts.
addr:housenumber=1
addr:street=Wellfield Bungalows
addr:parentstreet=Wellfield Road
or
addr:unit=1
addr:housename=Wellfield Bungalows
addr:street=Wellfield Road
@Robert_Whittaker Would you still discourage the use of addr:substreet in situations like this?
You shouldn’t split the landuse and name it - all it needs is a addr:housename set on the building (duplication is fine if there are 2 blocks) and then addr:flats with the range of flats e.g. 5-8. If a large apartment building has multiple stairwell entrances you can add an entrance=yes node on the outline at each doorway and tag addr:flats on those, leaving addr:housename on the building.
If the house/flat numbering is within the subordinate street/building, then yes. The standard address tagging assumes that addr:housenumber is numbering with addr:street or addr:place, and that addr:unit is numbering within addr:housename.
If you did something like
Then the address would be interpreted as property number 1 on Wellfield Road, and also part of a group of dwellings known as Wellfield Bungalows on that road.
This is different from what either of the two options you quoted in your post give you, i.e. the address is property number 1 within Wellfield Bungalows, which itself is a group of dwellings on Wellfield Road.
I think you’re suggesting — and this was my original assumption — that, logically, there should only ever be one instance of each unique addr:housenumber (and addr:housename?) per addr:street?
Would that the conflict with @LordGarySugar’s suggestion to tag both Tudor Court’s apartment buildings within the complex with addr:housename=Tudor Court? Or, does the logic change when they’re marked with addr:flats instead of addr:housenumber?
I really appreciate this discussion, it’s helping me understand the whole breadth of address tagging best practices for OSM in the UK
You’re suggesting it’s unnecessary for the landuse=residential area to be split as all of the tags already exist on the buildings themselves AND the car park with access=private, situated next to the apartment buildings, is implied to be part of the complex without an overlapping, named residential=apartments area? Despite being an apartment complex, it is redundant tagging without representing an actual change in land use.
I likely won’t but, if I wanted to, I could add an operator=Tudor Court tag to the car park to make explicit that the car park is private for residents..?
There’s also the rarely used but documented provides feature relation type which allows parking to be explicitly linked to the object it exists to serve.
I’ve decided, for now, the solution to my query is the below schema. Despite the supposed break in addr:housenumber–addr:street logic, this schema provides the most consistent and correct address according the validation tools made available to me.
I really wouldn’t use that, if the address is not the same as “4, Wellfield Road”. addr:substreet is not a universally recognised tag, and any data users that don’t understand it, will incorrectly associate the address you’ve entered as being “4, Wellfield Road”. To avoid ambiguity, and ensure backwards compatibility, I think addr:substreet should only be used when there’s a named collection of properties that uses the same numbering as the main street.
If you don’t want to use housename+unit, then it would be better to use
instead. Data users that don’t understand addr:parentstreet will then still get the number being associated with the correct ‘container’ of Wellfield Bungalows.
Though in this case, I think the arrangement is a much better fit for
addr:housename doesn’t have to be a single building. Blocks of flats will sometimes have two or more buildings that share the same name.
You mean that if ‘Wellfield Road’ comprised of eight Houses numbered 1–8, but four (1–4) of those Houses were named ‘Wellfield Bungalows’, you’d use addr:substreet? Whereas, the
tags mean that the addr:housenumber’s are being duplicated on ‘Wellfield Road’ and should not be used. OK.
Do you mind looking at another example; how would you tag the address for Goodchilds Garage (and other businesses), on Norths Estate industrial estate on Old Oxford Road?
Currently, I’ve decided on addr:substreet here too. Its address is 1a, Norths Estate, Old Oxford Road, but there also exists a 1a Old Oxford Road. So addr:substreet can’t work here either, can it?
As you are discovering, there is no consensus across mappers. Had we developed address tagging from scratch we would likely have gone with street and substreet as this most closely matches the official address standard in the UK (thoroughfare and dependent thoroughfare). But as with many things the schema has evolved gradually and parentstreet came into being. I personally don’t care for Robert’s logic here as it is the equivalent of “tagging for the renderer” (tagging for the search engine in this case).
It’s all set out on the wiki page if you want a long read!
Basically:
(1) any tagging that works on the envelope map is better than none; and
(2) make up your own mind which of the multiple choices you prefer / don’t feel you have to listen to whoever speaks last or shouts the loudest
) and, while searching for tagging help, came across 