The Office of National Statistics is consulting on a new statistical boundary for the rural-urban definition.
Although the amount of coverage of urban areas was quoted at the highest levels in the su7mmer planning debates we have not had an accurate delineation since 2001 – when it was 10.6 % – so no idea where the Prime Minister gets 9% from.
The ONS is looking whether to go for a manual delineation based on polygons as below – accurate but time consuming,
Automatic based on polygons- very inaccurate will require manual tidying up or automatic based on 50m grid squares, as below (small holes would be filled in).
On a national scale the three produce identical results. The issue is cost and usability. Personally from a GIS point of view you will need both polygon and raster versions of the dataset, polygon as an aid to planning, raster for analysis, and I can see no reason why the initial grid square results can be used with some simple programming to refine a polygon boundary. It just needs a good recursion based algorithm. I cant see why ONS needs to consult on their results of them not have solved a simple programming issue.