[Talk-GB] A quirk of UPRNs

[Mark Goodge]

On 20/08/2021 10:45, SK53 wrote:
> The other day whilst processing the UPRN table to find locations with 
> more than 1 UPRN I happened to notice an odd thing: some UPRNs have 
> integral eastings & northings, whereas others have these fields to 2 
> decimal places. OS Master Map has a horizontal RMS error of around 25 
> cm, so the precision of these suggested that they had not been derived 
> directly from Master Map. My suspicion is that these are converted WGS84 
> values.

[snip]

> I've been exploring this data within CH45 (New Brighton & Wallasey 
> Village) and my tentative conclusion is that the latter class represent 
> new URPNs after some unknown cut-off date. 
[snip]

That's an interesting theory, but possibly not entirely supported by 
other evidence. There are some very low numbered UPRNs that have 
non-integer OSGB coordinates, and, since UPRNS are, broadly speaking, 
issued sequentially, they're unlikely to be new.

What's possibly more plausible is that non-integer coordinates are found 
where the coordinates have been updated after a certain cut-off. That 
would, of course, include all new ones, but it could also include older 
ones that have been moved.

Mark