Thanks for the interest.
I haven't tried to collect parish boundaries yet, but it's an obvious step forward. The first thing is to say that I wouldn't want a load of effort going into supporting measuring stuff that is better spent mapping real stuff. So I wouldn't want to encourage anything that didn't have another purpose as well. Secondly the whole concept of this depends on some consistency between different areas. Big inconsistencies come from "hinterland" i.e. a settlement where there is a built-up core, surrounded by an area of low population. That's why some of the existing areas produce strange results. So without trying a bunch of parishes I can't be certain whether they would all produce a useful result. On past experience, most will, but some won't. Oddly, in some cases guessing at the built-up core from the population seems to work better than the actual boundary. So we will need to try this to see what happens. I'm happy to do that, but not just yet because of real-life pressures, so there is plenty time to add more parish boundaries before this gets redrawn. So here are my suggestions..... What the process of analysis needs is a recognisable boundary, a name to match things up and label things, and a population figure. Everything else comes from the map database itself. I've got some boundaries from the map, but not yet parishes. I should get population figures from elsewhere. I think the published 2001 census data goes down to parish level, and although its a bit out of date, it shouldn't be a million miles out. I believe there are crown copyright issues with adding these figures to the map (?) but presumably not when the processing is off-line. So on the map we basically need parish boundaries, with a name. For anyone who hasn't done boundaries before, see local authority boundaries as a model. Basically the clearest of these treat each segment of the boundary as a way tagged "boundary administrative" and "admin_level-x" where x = 10 for a parish (other numbers for higher admin levels). See http://wiki.openstreetmap.org/wiki/Key:admin_level#admin_level Then group all the segments with a relation, which is tagged as boundary, administrative and admin_level 10 with the name of the parish. It's important to do it this way with relations, not just with ways, because boundaries at different levels overlap each other. To process it, I will need the relation to be closed. This is where many of the local authority boundaries needed a lot of work in the past. The best way I know to check closure is with OSM Relation analyser - http://betaplace.emaitie.de/webapps.relation-analyzer/ which will tell you whether the boundary is complete (i.e. one joined up line) and closed (a closed loop). Without a closed loop I cannot manipulate the intersections between roads and boundaries. So "complete" is necessary but insufficient. And that's about it. I need to put this aside for a while because of other things, but will return to it in a week or two, and try to incorporate any parish boundaries that are there. On the overlapping areas in Medway, I'll see if increasing the transparency a bit can improve visibility - it's not difficult to adjust, so take another look tomorrow. One a slightly different but related subject, with a general election coming up, does anyone know how things stand with constituency boundaries?
_______________________________________________ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb