SotM Meeting 2020-07-04

2020, Minutes

As part of our participation in 2020’s online State of the Map conference we held an hour long virtual meeting to discuss new OS Open Data and how it might be used by OpenStreetMap mappers.

Download the slides:
Meeting notes

New uk open data:

  • OS Open USRN – Unique street name references
  • OS Open UPRN – Unique Property Reference Numbers. Just a CSV of locations in lat/lon or grid ref, no addresses.
  • OS Open Linked Identifies – cross references a bunch of the open OS databases
  • Changes of their terms opens up INSPIRE polygons – Now available under the OGL.

Nick – LGA planning service, UPRN location can move. Isn’t immutable. Importing it wholesale is not good

Main uprn/usrn comment – use for quality flagging. Indication of missing elements?

UPRN data concern – It’s not just properties. Includes streets (often at a junction). Other objects bus shelters/postboxes substations. Challenge is classifying into property/street/something else. This isn’t necessarily possible automatically yet.

ndrw6 – adding uprn/usrn is not practical to add by hand. Should be straight import. The question is how useful it is as it’s not verifiable on the ground like postcodes. Suggest an overlay that shows the data.

Andrew – going from UPRN to location. There’s no point duplicating UPRN to points. How can we make UPRN to address possible?

Tony – real benefit putting UPRN on properties, can we use the limited OS mastermap usage to fill in the blanks? It’s a useful linking node.

Rob N – free monthly allowance to OS mastermap usage licensing is unsure, further investigation required. Suspect it is for non-commercial use only.

Gareth – overlay v added values into OSM shows ‘completeness’ of map.

Rob W – Not sure if IDs used in UPRN are re-used, deleted, etc when objects are demolished.

Might be worth collating questions and OSMUK can forward to OS as queries.

Nick – LA using corporate address gazetteer. Querying if LGA would be able to release full data meaning we get addresses + referenced to UPRN from them directly.

Rob W – Suffolk County Council already include URSNs on their List of Streets, which provides a direct mapping from USRN to Street Name. (I’m not 100% sure whether they’re allowed to release this under the OGL though).

Rob W – I think someone said on talk-GB that the USRN associated with each UPRN doesn’t necessarily correspond to the street in the postal address.

Gareth – not sure if cut-throughs will have access implied.

Rob W – there might be indicative data on the USRN/UPRN that shows if it’s publicly maintained.

Nick – The data is quite large, would it be helpful if it could be split into LGA areas to make it more accessible.

Rob W – wary to extend my website to offer small downloads up to LGA level due to my server limits.

ndrw6 – It would be good to have imagery with USRN/UPRNs like the one we have for postcodes:{zoom}/{x}/{y}.png

Robert W – There was OS openroads dataset layer which could be useful to have an equivalent from this data.

Justin – Maybe create a lookup table that links UPRN to OSM for anyone who may need it.

Robert W – There are streets missing from OS open roads for some reason. They do have USRNs, so geometry but no name.

ndrw6 – I can do what I’ve done with postcodes – filter USRN/UPRNs against nearby objects (for example missing streets or streets with missing names) and make the available for download in jOSM.

ndrw6 – Does osmuk have computing resources to do tile serving etc? [Answer: No].

Nick – (on using ref:GB:) consistency is the main part. changing the existing UPRN tags for birmingham trees to :GB:

Robert W – Proposal on wiki and vote on wiki for the tag seems popular. Should define consensus.

ndrw6 – osmuk could help coordinate the tagging agreement.

Nick – if we go ref:GB:USRN/UPRN, this may impact other tags which would also need a similar process to update

Robert W – (USRN and UPRN on single element) possibly use addr: as it should refer directly to the feature.

ndrw6 – without extra data we only have USRN and UPRN so cannot conclude a lot of info just from this.