The Land Registry has released various data sets as recorded in this blog post, but there is a serious issue with some of their data, the polygons that show the cadastre, or property bounds, data. As you might expect, the Land Registry use the UK's national mapping agency, Ordnance Survey, to base any mapping on. So when they reproduce any GIS-based data, it is based on OS maps and, as usual, OS claims that their viral licensing clause applies. So even though the Land Registry maintain the cadastre data and want to publish it under the liberal Open Government licence, OS prevent that by adding a personal-only, no publishing clause. This is a further example of how OS viral licensing is harming innovation and thus damaging business and potential open data users.
I have been using the Land Registry polygons to research the accuracy of creating Openstreetmap cadastre data simply by using surveys, permitted aerial imagery and local knowledge. I have decided to publish an example of what I have found under fair dealing of OS copyright and database rights. If OS are not happy about this I'm sure they will contact me and we could discuss it.
Some time ago I added all of the buildings in the postcode area HU14, Melton, North Ferriby and Swanland and added their address. In North Ferriby I also added all of the hedges, fences etc around properties. It took a lot of doing, using aerial imagery and some local knowledge. Having done it I decided that I wouldn't do any more places like that - I would only add buildings and address at most.
Now I have the means to check how good a job I managed to do, by comparing my boundary lines with the cadastre polygons from the Land Registry. I extracted a small area from their data, reprojected it to Mercator and used TileMill to render a simple view of part of North Ferriby with the cadastre polygons laid over it. Here are the three images:
OpenStreetMap data showing gardens and buildings |
Land Registry cadastre polygons |
OpenStreetmap data overlaid with Land Registry cadastre polygons |
I am very pleased with the way the two datasets coincide. The biggest area of error in creating the Openstreetmap boundaries is the alignment of the aerial imagery - that seems to be very good in this instance. I did use multiple GPS traces to create an average trace and aligned the imagery to that. That seems to work well.
The Openstreetmap images © Chris Hill, CC-BY SA 2.0 which is based on Openstreetmap data © Openstreetmap contributors ODbL 1.0
The Land Registry data require the following statements:
This information is subject to Crown copyright and is reproduced with the permission of Land Registry
© Crown copyright and database rights [2013] Ordnance Survey 100026316