This discussion has convinced me not to use landuse=reservoir. It sounds like the only benefit is its historical use, whereas I've personally seen benefits of the natural=water approach.
I've mapped quite a number of farm dams as natural=water without being sure what subtag to use. I now think that's because there isn't an appropriate subtag. I definitely don't want to tag it as a pond. While a farm dam is structurally and functionally a reservoir, there are clear differences with large reservoirs. Already now, farm dams tend to be mapped more prominently than I'd expect. The dominant feature of these grazing landscapes is fencing, and I'd therefore expect farm dams to appear on a similar scale to fences. water=reservoir and landuse=reservoir wouldn't do that. One of the things I love about OSM is the ability to map incrementally, which by definition results in incomplete, lower quality maps that are constantly improving. If the priority was a high quality map, we'd map systematically (like Missing maps, but for everything that will appear on a render) and not release an area until it was done. I wouldn't be mapping. On Thu, 17 Dec 2020, 1:26 am Tomas Straupis, <tomasstrau...@gmail.com> wrote: > 2020-12-16, tr, 16:01 Mateusz Konieczny rašė: > > > https://wiki.openstreetmap.org/wiki/Tag:landuse%3Dreservoir#water.3Dreservoir > > (just added) > > Thank you. Maybe it is better to discuss here before adding to wiki? > My arguments on the points you've added: > > 1. Regarding benefit of having a combining level/tag natural=water. > If today you would query all data with natural=water - you will get > not only lakes and reservoirs grouped, but also riverbank polygons > (totally different beast) and micro elements like water=pond. This > could only be partly useful in the largest scale maps and only if you > make very simple maps and for some reason use the same symbolisation > for such different water classes. For example ponds usually have less > complex and less prominent symbolisation because of their size and > importance. Riverbanks would not need polygon labelling, but rather > use river (central) line for label placement. Most of GIS/Cartography > work goes in middle/small scales and it will be impossible to use only > natural=water there, you would have to add "and water not in > ('riverbank', 'pond', ...)". This erodes the benefit of "one tag" and > makes it of the same complexity from coding perspective as original > water scheme. > > 2. Very important disadvantage of water=reservoir from > cartographic/gis perspective: it allows mappers to NOT differentiate > between natural lakes and man made reservoirs. If first point > describes how different classes are USED, this second point is about > how these classes are CAPTURED. > > Did I miss anything? > > _______________________________________________ > Tagging mailing list > Tagging@openstreetmap.org > https://lists.openstreetmap.org/listinfo/tagging >
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
