The entire landcover tag discussion on the wiki is a huge distraction,and not based on any objective criteria, let alone an attempt to see if what we have works.
I, on the other hand, gave a paper at SotM-Eu in 2011 which showed that use of existing tags could provide a level of land-use/land-cover mapping comparable to the European Environment Agency's Urban Atlas. The slides are here <https://sotm-eu.org/slides/38_JerryClough_UrbanAtlas_SK53.pdf>. I think there are enough details in the methodology for anyone to replicate this for other places. Unfortunately a lot of Europe has imported Corine data and therefore it is not possible to assess the practicability of OSM as a general source for land-use/land-cover. Furthermore the main dataset I used, Nottingham, was based on a snapshot so that I was not tempted to add tags to correct. I actually ran the data for several european cities, and for two square degrees of the UK. The main problem in matching OSM to Urban Atlas was not tagging, but the absence of mapping of landuse. Other discrepancies were due to change in landuse in the timeframe since the UA data had been compiled, and faulty interpretation in the UA data set (faculty buildings at Nottingham University classed as residential, for instance). Whereas there are some issues with landuse tags, they will not be fixed by inventing another category which will be beset by the same problems over time. It's much better to try and persuade people that things like landuse=grass for farmland pasture is a bad idea. Jerry PS. Can I just echo what Richard said in a recent e-mail: you are taking the wiki far too seriously. It really is a curates egg. On Thu, Aug 29, 2013 at 8:25 AM, OpenStreetmap HADW <osmh...@gmail.com>wrote: > On 28 August 2013 23:15, Dudley Ibbett <dudleyibb...@hotmail.com> wrote: > > > > This would perhaps suggest they should be marked as ways with > barrier=hedge > > and hedge=line_of_trees or perhaps just the latter. > > > > An alternative might be to use natural=tree_row which is defined in the > wiki > > but the examples seem more to related to trees that have been planted at > > regular intervals and where there isn't generally an overlap in the > canopy. > > I have used this a few times but I'm not convinced it is the right way to > > tag this feature given that it seems they are a type of hedgerow. > > > > This sort of micro-woodland feature is a case where it would be nice > if Mapnik supported the proposed landcover key. I found a whole > forest growing in the central reservation of a short section of dual > carriageway, but had to leave it that way as it no longer rendered > when converted to landcover (in most areas, people wouldn't have > bothered to try to make it render, but since someone had done so, > they would reasonably object if it stopped rendering. > > _______________________________________________ > Talk-GB mailing list > Talk-GB@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-gb >
_______________________________________________ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
