[Talk-GB] UPRN Locations Map

SK53 sk53.osm at gmail.com
Thu Jul 2 18:52:02 UTC 2020

I've also had a quick look, and your hints are quite useful.

One thing I've noticed is UPRNs which I suspect are for building shells. So
there's a house which has recently been converted into student housing with
a hair transplant surgery on the ground floor. This has 3 UPRNs. The next
property down is a small parade of 4 shops with 3 office units above and
this has a total of 10 UPRNs. The next parade of 6 shops has 7 UPRNs. At
least one is the shell, or possibly the land parcel on which the property

One of the university buildings has a number of self-contained flats, but
there are no UPRNs for these (presumably postal mail goes to the tenants
through the internal mail).

Post boxes, substations, patches of grass (I presume), and bus stops are
things I've spotted. The oddity is a great forest of UPRNs over a hospital
building <https://osm.mathmos.net/addresses/uprn/#19/52.9423/-1.1827>.
Student residences ('villages') seem to have one UPRN per flat/studio + 1
for the building itself. What is interesting is that quite a few shops (on
university premises) which pay business rates do not seem to appear, and
conversely ATMs which are ratable do not appear at all. Street lights, man
hole covers, street cabinets, gullies etc do not appear, and I think are
not captured for MasterMap. It does not resolve a local mapping conundrum,
which is what happened to the rifle range underneath a former canal bridge.


On Thu, 2 Jul 2020 at 17:38, Robert Whittaker (OSM lists) <
robert.whittaker+osm at gmail.com> wrote:

> I'm not completely sure if/how we can best make use of the new OS
> OpenData (UPRNs, USRNs and related links) in OpenStreetMap, but as a
> first step I've set up a quick slippy map with the UPRN locations
> shown:
> https://osm.mathmos.net/addresses/uprn/ (zoom in to level 16 to show the
> data)
> The UPRN dataset literally just contains the UPRN number and its
> coordinates (both OS National Grid and WGS lat/lon). There are some
> additional linking datasets that link these ids to other ids (e.g.
> USRNs, TOIDs). But no address information is available directly. (You
> may be able to get street names by matching to OS Open Roads via TOIDs
> though. Coupled with Code-Point Open, you might be able to assign
> quite a few postcodes in cases where there's only one unit for a whole
> street.)
> The UPRN data has already helped me find a mapping error I made
> locally though -- it looks like I'd accidentally missed drawing a
> house outline from aerial imagery, and also classified a large garage
> a few doors down as a house. The two errors cancelled out when the
> houses were numbered sequentially, so I didn't notice until now. Today
> though I spotted a UPRN marker over some blank space on the map, and
> no marker over the mapped house that's probably a garage.
> Now a few initial thoughts on the data that I've explored so far:
> I believe that the UPRNs are assigned by local authorities, so
> conventions may vary from place to place. I don't know who actually
> assigns the coordinates (authority or OS). Looking at those for rows
> of houses around me, they don't seem to have been automatically given
> coordinates from the house footprint, it looks more like someone
> manually clicking on a map.
> The UPRN dataset should include all addressable properties. It is also
> ahead of reality in some places, as it includes locations for houses
> on a new development near me that have yet to be built yet. For blocks
> of apartments/flats, the UPRN nodes may all have the same coordinates
> or may be displaced from each other, possibly in an artificial manner.
> Other objects also appear to have UPRNs. Likely things I've noticed so
> far include: car parks, post boxes, telephone boxes (even after
> they've been removed), electricity sub-stations, roads and recorded
> footpaths (the UPRN locations seem to be at one end of the street, so
> usually lie at a junction), recreation grounds / play areas,
> floodlight poles (around sports pitches), and allotments. There's no
> information about the object type in the UPRN data unfortunately.
> Anyway, I hope some of this is useful / interesting. I hope to be on
> the OSMUK call on Saturday to discuss things further. Best wishes,
> Robert.
> --
> Robert Whittaker
> https://osm.mathmos.net/
> _______________________________________________
> Talk-GB mailing list
> Talk-GB at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-gb
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20200702/84477262/attachment.htm>

More information about the Talk-GB mailing list