>Surely population density could be accessed more accurately by numbers of buildings than by sizes of "blobs", which are outlines very inconsistently ?
I've seen people take the square area of all the buildings and apply a formula to it that includes the number of storeys. Not as good as a census but not bad. If the buildings are mapped with the JOSM building_tool plugin you stand a chance. I know it is possible to map them accurately with iD, I've seen it done but I do more validation than mapping these days and to be honest yes there is a building roughly there but the area is quite often twice what it should be and no two sides are parallel. Quite often the settlement is only partially mapped ie half the buildings are missing. Also its a question of resources. Mapping all the buildings in a settlement takes time much more time than just mapping a landuse=residential for the settlement. Now look at the number of projects we have that have been open more than five years. Some haven't been touched for years. We do have a limited number of mappers, is the best use of them to map buildings in one small area? If you're a project manager in that small area the answer is probably yes but from the big picture its probably better to get a few highways and villages in first. Then the locals can start to add to the map. You need boots on the ground to add detail anyway and hopefully that will give a few more mappers to add buildings. Cheerio John On 29 March 2017 at 18:03, Kretzer <kret...@gmx.net> wrote: > Hi, I was also bothered by the coexistence of very small and very large > residential area in many loosely populated African areas. > I agree that the tag seems to make most sense in densely built-up areas - > to differentiate it from other uses of built-up land, not from open areas. > > But I'm not sure about "boundary" either. In Johns original example it was > suggested to circle clusters of two or three buildings, and I'm not sure > about the benefits of this. In these cases I think mapping buildings alone > would be sufficient. Or am I missing something? In most renderings you can > see the residential areas in smaller scales as the buildings - so you could > identify populated areas more easily. Is that a benfit? Or could that be > solved by different rendering? > > Surely population density could be accessed more accurately by numbers of > buildings than by sizes of "blobs", which are outlines very inconsistently ? > > > *Gesendet:* Mittwoch, 29. März 2017 um 10:33 Uhr > *Von:* majka <majka.zem+t...@gmail.com> > *An:* "HOT@openstreetmap.org" <hot@openstreetmap.org> > *Betreff:* Re: [HOT] Fwd: Re: landuse=residential within > landuse=residential > > First, overlapping landuse areas (even different ones) should *always* be > corrected. It brings problems with the map data, I have seen and corrected > areas where the overlapping did hide ponds from the rendered map. The > *same* overlapping area masks some of the problems but should be > corrected as well - either by deleting of one of the areas or by merging > both together. > > The next question is the landuse *size* in the mapped area. > > From the view of the mapper in Europe, the landuse=residential in HOT is > problematic. The residential area should be only where the region is used > *above > all* for housing people. The HOT use is to mark areas where there are > *some* houses, depending on the project instructions. This ends with a > very problematic rendering of some areas. Visually, you get one big blob of > something most people understand as a town, not the reality of fields and > farms. The very loose residential areas shouldn’t be there at all, IMHO. > Villages/towns boundaries have their own tag, *boundary*. Usually, this > is paired with boundary=administrative which is mostly unusable for HOT > distance mapping because the information isn't available to the mapper. But > nothing speaks against own tag - see here > <http://wiki.openstreetmap.org/wiki/Key:boundary>. > > IMHO, the ideal solution would be to change the HOT practice of mapping > residential areas. Leave landuse=residential only to the areas, where the > buildings are densely packed together (even in a village, where there is > *real* street there might be a residential area) - keeping the common > interpretation. Give the residential area a lower importance than it has > now, and start using the boundary instead, for example boundary=residential > to mark the areas with buildings. A later mapping on the ground or use of > governmental data if available could then change this in real > administrative areas marking the hamlets, villages, and towns where > appropriate and leaving the *residential* boundaries to the rural farm > areas. > > Ideally, such change would be preceded by discussing on the HOT and > tagging list and followed by updating the wiki definition of a boundary, > and by updating the HOT materials for users. It would need a slight change > in JOSM HOT presets and in the iD editor as well, probably. However, it > shouldn’t be very difficult to do so. > > I understand the residential areas are used for getting population density > in the HOT projects. The use of both tags together would be a better > choice, getting the information about sparsely and densely populated areas > at the same time. > > Majka > > On 29 March 2017 at 08:10, Vao Matua vaoma...@gmail.com > <http://mailto:vaoma...@gmail.com> wrote: > > > > Nick & John, > > Determining where to draw the edge of landuse=residential can be difficult. > Here in Ethiopia most of the population lives in a rural setting where > they farm areas of 1 to 10 hectares in size. > http://www.openstreetmap.org/#map=16/6.9634/38.4408 > There are places where people live in villages, but often dwellings are > quite dispersed. > > > _______________________________________________ HOT mailing list > HOT@openstreetmap.org https://lists.openstreetmap.org/listinfo/hot > > _______________________________________________ > HOT mailing list > HOT@openstreetmap.org > https://lists.openstreetmap.org/listinfo/hot > >
_______________________________________________ HOT mailing list HOT@openstreetmap.org https://lists.openstreetmap.org/listinfo/hot
