Re: [Tagging] Tagging National Forests
W dniu 17.08.2015 4:10, Martijn van Exel napisał(a): But after some discussion I realized that this may be a side effect of a different problem, namely how we tag national forests. In the US, these seem to be tagged as landuse=forest which is only partly true: within a National Forest, many different land uses can occur, only one of them being forest. We had the same problem with imports of national forests in Poland. It's exactly the counterintuitive problem you've mentioned: forest area is not always covered with trees! In our case that was probably just areas being property of the national forest operator "Lasy Państwowe" (which is the same as "National Forests" by coincidence =} ). So should we just not tag National Forests as landuse=forest? We started redrawing the boundaries, so the forest is just the ground truth (only the trees), but now I'm not sure that was the best action to take, even if simple and useful. Somebody lately said, that the forest area may include burned areas, young trees fields and other such things. I'm not into the forestry, but it looks we have the opportunity to redefine our trees/forest tags, starting from general understanding what the forest really is and what parts it consists of. While discussions about landcover=trees are useful, they are way too narrow. I feel we need to rethink the whole tree tagging in OSM, because we have no general agreement on the subject: http://wiki.openstreetmap.org/wiki/Forest The whole issue is not as straightforward as one can reasonably expect. According to Wikipedia: "A forest is a large area of land covered with trees or other woody vegetation.[1] Hundreds of more precise definitions of forest are used throughout the world, incorporating factors such as tree density, tree height, land use, legal standing and ecological function.[2][3][4] According to the widely-used[5][6] United Nations Food and Agriculture Organization definition, forests covered an area of four billion hectares (15 million square miles) or approximately 30 percent of the world's land area in 2006.[4]" The FAO definition is linked: http://www.fao.org/docrep/005/y4171e/y4171e10.htm and it has about 17 pages on my screen. Actually it's rather good that it is so comprehensive, because it may be a good base for understanding the background and to identify parts we may be interested in. Another idea is to research common GIS practices regarding trees. Anybody willing to get deeper into the subject? -- "The train is always on time / The trick is to be ready to put your bags down" [A. Cohen] ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Tagging National Forests
For me, forestry is the production of wood, using trees. So a 'forestry area' would include mature trees, young trees, saplings, fresh plantings and places where the trees have been removed. I think that is what is meant by landuse = forest On the other hand there are areas that are covered in trees .. that are not intended to be used for wood products, so natural=wood (or landcover=trees) is more appropriate. On 17/08/2015 7:45 PM, Daniel Koć wrote: W dniu 17.08.2015 4:10, Martijn van Exel napisał(a): But after some discussion I realized that this may be a side effect of a different problem, namely how we tag national forests. In the US, these seem to be tagged as landuse=forest which is only partly true: within a National Forest, many different land uses can occur, only one of them being forest. We had the same problem with imports of national forests in Poland. It's exactly the counterintuitive problem you've mentioned: forest area is not always covered with trees! In our case that was probably just areas being property of the national forest operator "Lasy Państwowe" (which is the same as "National Forests" by coincidence =} ). So should we just not tag National Forests as landuse=forest? We started redrawing the boundaries, so the forest is just the ground truth (only the trees), but now I'm not sure that was the best action to take, even if simple and useful. Somebody lately said, that the forest area may include burned areas, young trees fields and other such things. I'm not into the forestry, but it looks we have the opportunity to redefine our trees/forest tags, starting from general understanding what the forest really is and what parts it consists of. While discussions about landcover=trees are useful, they are way too narrow. I feel we need to rethink the whole tree tagging in OSM, because we have no general agreement on the subject: http://wiki.openstreetmap.org/wiki/Forest The whole issue is not as straightforward as one can reasonably expect. According to Wikipedia: "A forest is a large area of land covered with trees or other woody vegetation.[1] Hundreds of more precise definitions of forest are used throughout the world, incorporating factors such as tree density, tree height, land use, legal standing and ecological function.[2][3][4] According to the widely-used[5][6] United Nations Food and Agriculture Organization definition, forests covered an area of four billion hectares (15 million square miles) or approximately 30 percent of the world's land area in 2006.[4]" The FAO definition is linked: http://www.fao.org/docrep/005/y4171e/y4171e10.htm and it has about 17 pages on my screen. Actually it's rather good that it is so comprehensive, because it may be a good base for understanding the background and to identify parts we may be interested in. Another idea is to research common GIS practices regarding trees. Anybody willing to get deeper into the subject? ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Tagging National Forests
On Mon, Aug 17, 2015 at 11:45:23AM +0200, Daniel Koć wrote: > > http://wiki.openstreetmap.org/wiki/Forest > > The whole issue is not as straightforward as one can reasonably expect. definitely. > According to Wikipedia: > > "A forest is a large area of land covered with trees or other woody > vegetation.[1] Hundreds of more precise definitions of forest are used > throughout the world, incorporating factors such as tree density, tree > height, land use, legal standing and ecological function.[2][3][4] According > to the widely-used[5][6] United Nations Food and Agriculture Organization > definition, forests covered an area of four billion hectares (15 million > square miles) or approximately 30 percent of the world's land area in > 2006.[4]" which only scratches the surface. We want to map ground level, shrub level and tree level vegetation etc. Also the soil and some geology bellow. Those are genrally overlapping areas with densitiy maps. Richard ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] Tagging National Forests
On Sun, Aug 16, 2015 at 08:10:17PM -0600, Martijn van Exel wrote: > Hi, > > The new rendering of forests broke cases where a lake is inside a forest > and the lake is not mapped as an inner section of the surrounding forest > polygon. it also "broke" the case of parking lots. > But after some discussion I realized that this may be a side effect of a > different problem, namely how we tag national forests. In the US, these > seem to be tagged as landuse=forest which is only partly true: within a > National Forest, many different land uses can occur, only one of them being > forest. > > So should we just not tag National Forests as landuse=forest? for the administrative entity of the national forrerst you definitely need a different tagging scheme. I have seen several "National Forests"s which I would call deserts or praries. Nevertheless closing the bug report was a bit premature, the developpers should have their word whether this was an intended change or not. It should be also considered that many other renderers render those lakes and parking lots without the trees. Richard ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] landcover=trees definition
On 17.08.2015 00:29, John Willis wrote: > This is the crux of the landcover argument. > > Because landuse=* implies what the land is used for - therefore man-altered > and decided usefulness. natural=* was then interpreted by taggers to be the > opposite - the "natural" state of the land which was heavily influenced by > the landuse=forest /natural=wood debacle. > > Landcover=* just says "this is here" , without adding implications as to its > use or origin. I know what you mean, but you are missing the point that landcover is layered. "This his here" applies to bedrock, ground water, soil, surface water, vegetation (root layer, moss layer, herb layer, shrubs layer, tree layer), and air. So we need multiple keys to specify them all. Or we just consider one of these layers, but this needs to be clearly defined. We already have tags for certain layers, such as surface=*. Unfortunately, that key is spoiled by surface=grass which means another layer. This would better go to a vegetation related tag. The most common tag for vegetation is natural=* - which in turn is even less clean because it covers surface, water and landforms as well. Let's not make the same mistake again with landcover=*. One solution could be a landcover:* scheme instead of a single key. Say, landcover:surface=* for earth/sand/mud/rock/concrete/asphalt/etc. Then some vegetation tags: landcover:vegetation:moss=yes/no/percentage landcover:vegetation:herbs=yes/no/percentage landcover:vegetation:herbs=yes/no/percentage landcover:vegetation:shrubs:=yes/no/percentage landcover:vegetation:trees=yes/no/percentage (with percentage = 100 * covered area / total area, so the sum of the percentages possibly exceeds 100) as well as landcover:vegetation:herbs:height=0.2 landcover:vegetation:shrubs:height=1.5 landcover:vegetation:trees:height=10 This would enable nice 3D rendering. > This also would allow for some man-made landcovers; as several times i am > dealing with a place where concrete or asphalt is covering the ground, but > not as road or path or building. This is a weaker use case, but it would be > nice to say "here is 2000sqm of concrete. It is the remnant of an old > airport. The airport is gone, it is not a road, a building or a structure. It > is now a (currently) purposeless expanse of concrete. Currently I have to map > it as the negative space surrounded by other things (meadow) to leave the > impression something is there (NAS Alameda in San Francisco is a perfect > example: > https://www.google.com/maps/@37.7813303,-122.3170894,16z/data=!3m1!1e3 part > of it is now roads, tracks, or other facilities, but it is an abandoned > airport where most of the feature has no use nor is natural). We can map the area of a highway as either highway=xxx + area=yes, or area:highway=xxx. If it is no more in use, we can add disused=yes or abandoned=yes. We can use a similar approach for abandoned airports. I've also seen some abandoned primary highways tagged as highway=track, because they can still be used as tracks. This would also work for abandoned airport runways. All in all, we've got plenty of possibilities. Of course, if you just want to store the information that there is an area sealed with a layer of concrete, some simple surface=concrete would be more to the point. > Grass along the sides of manicured roads (like on a cutting or separation for > safety or noise control), which are part of the roadway's land, but not part > of the road - nearby residential houses, but not part of a residence nor used > as a park - its there just to be grass. We've got landuse=grass for that. > Landcover=iceplant would be brilliant for southern California freeway mapping. > > Its not used for anything other than being "iceplant"- occasionally a car > will go in it, but it's job os just to be "there" so the ground isn't dirt or > dead meadow grass. Sounds like a landcover to me. Or landuse=flowerbed and possibly species=Mesembryanthemum crystallinum. > If I had landcover=trees with a boundary line like nature reserve, I > wouldn't have to decide between wood and forest, when it is a bit of both. I agree that the forest/wood distinction causes headaches, yet both are more than just a cluster of trees. I wouldn't oppose landcover=* as much if the suggested tag für forests/woods were landcover=wood. -- Friedrich K. Volkmann http://www.volki.at/ Adr.: Davidgasse 76-80/14/10, 1100 Wien, Austria ___ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
Re: [Tagging] landcover=trees definition
Then we can create some biome tags to handle more complex tagging, but being able to define commonly encountered landcovers is necessary. My city has huge flood control embankmnets along the natural river in certain places. There is abandoned sections of asphalt and concrete in patches in odd places (abandoned places) There are also huge expanses of (planted) grass kept on other sections of the flood control levees which are just there - not a park, not a meadow - just grass. There are natural=scrub sections too. I know this issue started with forest/wood/trees, but there are other simple tags that can benefit from just "landcover" - and we can figure out more complicated "biome" tags as we go. Sometimes it is just cedar trees, pine trees, bamboo, eucalyptus, or California Oak, - and the ground level biome is sparse/not interesting/not worthy enough to bother mapping. Maybe that can be extended with simple subtags (like forests) or some "OpenBotanyMap" kind of additional tags. Javbw > On Aug 18, 2015, at 1:50 AM, Friedrich Volkmann wrote: > >> On 17.08.2015 00:29, John Willis wrote: >> This is the crux of the landcover argument. >> >> Because landuse=* implies what the land is used for - therefore man-altered >> and decided usefulness. natural=* was then interpreted by taggers to be the >> opposite - the "natural" state of the land which was heavily influenced by >> the landuse=forest /natural=wood debacle. >> >> Landcover=* just says "this is here" , without adding implications as to its >> use or origin. > > I know what you mean, but you are missing the point that landcover is > layered. "This his here" applies to bedrock, ground water, soil, surface > water, vegetation (root layer, moss layer, herb layer, shrubs layer, tree > layer), and air. So we need multiple keys to specify them all. Or we just > consider one of these layers, but this needs to be clearly defined. > > We already have tags for certain layers, such as surface=*. Unfortunately, > that key is spoiled by surface=grass which means another layer. This would > better go to a vegetation related tag. The most common tag for vegetation is > natural=* - which in turn is even less clean because it covers surface, > water and landforms as well. > > Let's not make the same mistake again with landcover=*. > > One solution could be a landcover:* scheme instead of a single key. Say, > landcover:surface=* for earth/sand/mud/rock/concrete/asphalt/etc. Then some > vegetation tags: > landcover:vegetation:moss=yes/no/percentage > landcover:vegetation:herbs=yes/no/percentage > landcover:vegetation:herbs=yes/no/percentage > landcover:vegetation:shrubs:=yes/no/percentage > landcover:vegetation:trees=yes/no/percentage > (with percentage = 100 * covered area / total area, so the sum of the > percentages possibly exceeds 100) > > as well as > landcover:vegetation:herbs:height=0.2 > landcover:vegetation:shrubs:height=1.5 > landcover:vegetation:trees:height=10 > > This would enable nice 3D rendering. > >> This also would allow for some man-made landcovers; as several times i am >> dealing with a place where concrete or asphalt is covering the ground, but >> not as road or path or building. This is a weaker use case, but it would be >> nice to say "here is 2000sqm of concrete. It is the remnant of an old >> airport. The airport is gone, it is not a road, a building or a structure. >> It is now a (currently) purposeless expanse of concrete. Currently I have to >> map it as the negative space surrounded by other things (meadow) to leave >> the impression something is there (NAS Alameda in San Francisco is a perfect >> example: >> https://www.google.com/maps/@37.7813303,-122.3170894,16z/data=!3m1!1e3 part >> of it is now roads, tracks, or other facilities, but it is an abandoned >> airport where most of the feature has no use nor is natural). > > We can map the area of a highway as either highway=xxx + area=yes, or > area:highway=xxx. If it is no more in use, we can add disused=yes or > abandoned=yes. We can use a similar approach for abandoned airports. I've > also seen some abandoned primary highways tagged as highway=track, because > they can still be used as tracks. This would also work for abandoned airport > runways. All in all, we've got plenty of possibilities. > > Of course, if you just want to store the information that there is an area > sealed with a layer of concrete, some simple surface=concrete would be more > to the point. > >> Grass along the sides of manicured roads (like on a cutting or separation >> for safety or noise control), which are part of the roadway's land, but not >> part of the road - nearby residential houses, but not part of a residence >> nor used as a park - its there just to be grass. > > We've got landuse=grass for that. > >> Landcover=iceplant would be brilliant for southern California freeway >> mapping. >> >> Its not used for anything other than being "iceplant"- occas