Rendering of natural=ridge #1148
Comments
|
How ridges should be rendered? Just label? Label + something else?
It is unlikely that on low zoom levels displaying ridges will be feasible - natural=ridge may be applied to both major mountain range and for series of hills.
I think that intended meaning was "at higher zoom levels than 10" (as in - start displaying on zoom level 10 and continue to display it as map is zoomed in).
This is not going to work, ridge may be represented by mutiple separate natural=ridge ways, joined together (for example - different name values across ridge - or other tags that change value). |
|
Ridges could be shown at zoom levels and in a style similar to natural=cliff. This would be good to reduce the widespread abuse of natural=cliff for ridges just to make them show up on the map, like: http://www.openstreetmap.org/#map=13/52.8631/93.4115 |
But it should not be too close - many ridges are gentle and completely unlike cliffs. |
Actually, in the example you gave, the editor intended to tag natural=arete instead of natural=ridge, see the "note" tag he added which, allbeit with a typing error ("aerte" instead of "arete"), says this: http://www.openstreetmap.org/way/218061863 So unless natural=arete is rendered too, I don't think the rendering of natural=ridge is useful by itself, there would still be a lot missed. If implemented, both need to be supported. And it is likely just natural=arete needs a line symbol, the natural=ridge could simply be labeled only, like @mkoniecz already suggested. Many aretes aren't labeled because they don't have a name necessarily, ridges most likely do have a name. |
|
As discussed in #1097 natural=arete is actually a special case of natural=ridge. Using natural=ridge instead of natural=arete is never wrong, using natural=cliff is. As far as rendering is concerned - it would make sense to render ridges as a plain line, styled of course in a way that makes them easy to distinguish from other line features like tracks and possibly render aretes with an additional line signature similar to natural=cliff but symmetric. Rendering just labels would IMO not be a good idea, it would be extremely prone to tagging for the renderer - people would widely use it for just placing labels. |
I understand, and agree about the observation regarding the difference with cliff.
However, I don't agree with this. Most true topographics maps don't seem to use any real symbol for showing ridges. A ridge, often being far less prominent and "sharp edged" than an arete, just isn't easily defined in terms of its shape and extent, as @mkoniecz justly pointed out, also with his reference to mountain ranges. Hence a label showing the approximate location is a more truthful and better cartographic solution. Of course, the label can't just be placed "horizontal", it needs aligning parallel with the ridge, like road labels. I am also wondering if some of the "natural=mountain_range"s and "natural=ridge"s haven't been defined as polygons. Since nothing prevents this, I think, even if they are currently only defined as line, it needs to be taken into account that some ridges and mountain ranges will start being tagged as polygons (and maybe even points), and that these need support too (a label could be placed in the direction of biggest width). Rendering aretes with a line symbol thus seem reasonable though.
I think that is a risk one needs to take. Actually, there is already a "generic" tag for placing labels "anywhere": place=locality (http://wiki.openstreetmap.org/wiki/Tag:place%3Dlocality) As the Wiki says: And yes, this tag does already render, see this one in London for example: |
Actually this is mostly a matter of map style and fashion - these days the prevailing methods of relief depiction in maps are shading and contour lines but ridge lines have a long tradition as well. One of the major disadvantages of using contour lines is that ridges as the major structuring elements of the terrain are not well visible. But my argument is not that the map has to depict ridges, it is that doing so would give a much better incentive to tagging them correctly than now. And evidently there are a lot of people who think mapping ridges is important.
Yes - and look how frequently this is abused - including the example you gave. Anyway i think this part of the discussion is fairly pointless - of the 17416 ways tagged natural=ridge only 2314 have a name tag at the moment. |
I wasn't arguing this, I agree ridges may be useful on the map. I just think labels, and no line symbol, like @mkoniecz suggested, is the more appropriate way to depict them. The same for mountain ranges: only label. Aretes with line symbol on the other hand, does seems logical. |
If that is the case, I think many of these unnamed ones are either redundant, as covered by contour lines, or need to have their locally known names added. The ones that do have a geographical, and locally known name, are probably the ones that matter. Contours can cover the rest of unnamed stuff. Many of the others may have just been "made up" by looking at contour lines anyway, with no real local (historic) significance to the people living there, or to anyone visiting trying to find its way in unknown terrain. I don't think we will be missing out on much if ridges without name don't show up on the map, just like an unnamed mountain range is of little use in orienteering, and isn't orientation the main reason for the desire to have these rendered? By the way, I now also see this is well argueed in the proposal page for ridges (https://wiki.openstreetmap.org/wiki/Proposed_features/ridge), as it says there: Ridges also have names, which are used colloquially. This is relevant geographical information which lets you talk in terms of ridges names rather than of "that hill behind this village". An unnamed ridge or mountain range in OSM is unfortunately again "that hill behind the village"... |
|
I agree that ridge geometry data in OSM usually adds little original information that is not available elsewhere but I don't think this is the place to tell people what features are map-worthy. And if ridges are mapped from accurate high resolution imagery the data is also more accurate than what can be derived from freely available elevation data. If ridges have names depends on the cultural background of the area in question (if it is even populated). A ridge is a ridge even if it does not have a name and it can be an important feature for orientation even then. I think the discussion here should focus on if and how the >17000 ways with this tag in the database should be rendered. And the examples i gave using the cliff tag to me seem to indicate that mappers do not only think ridges are map-worthy, they at least partly also think these should show up in the map. |
|
Side step, but out of curiosity I now also looked up the few natural=mountain_range features on overpass turbo. Most seem to be tagged as nodes and are "smaller", lesser known ranges, and some justly, because there is no real orientation in the "range", like here: http://www.openstreetmap.org/#map=13/39.2164/-121.8046&layers=Q registered as node http://www.openstreetmap.org/node/2267790077 |
I think we are having this discussion right now... but let others step in. I am just joining because I am developing my own style, and need to make my own decisions on the same topic. |
|
Another point is that rendering of ridges and cliffs is most useful if also height contours are available, something unfortunately not available in the standard rendering. Look at the image below, depicting Snowdon mountain in North Wales, UK. Left OSM Standard, right is from my OSM Renderer for ArcGIS that I have in development. I have enhanced the data with open data contours of Ordnance Survey UK. In OSM Standard, the natural=cliff, seem kind of "misplaced" and non-functional with the lack of contours and no artificial relief shading, you don't really get a feeling of the landscape. Contrast that with the right part of the image, where the "cliffs" kind of nicely enhance and complement the contours, accentuating the most prominent "ridges" and "cliffs".
|
|
My rendering suggestion based on all of this and the observations I made, and that I will implement in my personal renderer (with appropriate display scales - hopefully ;) ): As nodes / points:
As non-closed ways / lines:
As closed ways / polygons or multipolygon relation:
|
|
I have now made an attempt to render natural=ridge. This actually turns out to be quite difficult. A solid line like for the one sided natural=cliff representing a steep decline, isn't a suitable representation. Neither is a simple dashed line, as they look to much like paths. The image below represents about the best I could come up with, trying to represent the not so prominent nature of this feature (click to enlarge). The image also contains another novelty, the rendering of the natural=valley tag, which really adds something in combination with contour lines.
|
|
I think your first attempt looks very promising, but the ridges shown here seem to be quite small. How would larger structures be rendered? For example Serra da Moeda in Brazil: or an even larger one like Serra Geral de Goiás I've added several new ridges in my region over the past weeks, don't know how many, but certainly over a hundred, and I am not remotely done. It would seem than ridges are currently severely under-represented in OSM database. One thing I noticed is that many ridges define boundaries, in a very similar way as some rivers do, Serra Geral de Goiás is one of those. Therefore one need to keep in mind that in many cases the rendering of ridges and boundaries will overlap. I don't know what kind of challenge that represents for rendering, but I think it is worthwhile to mention it. I like your representation of valleys quite a lot, in fact it is closer to what I had in mind initially for ridges. |
@drgweber, to be honest, all of the examples you linked here in the this issue thread are in my opinion natural=cliff, not natural=ridge... All of them seem to represent nearly impassable steep descends coming from relatively flat high areas, as can clearly be seen if you look at the data using the OpenCycleMap view and its great relief view using shading and contours: E.g. A natural=ridge is something that should have two clearly sloping sides, not one: http://wiki.openstreetmap.org/wiki/Tag:natural%3Dridge and compare that with cliff: Very generally speaking, cliffs are usually a consequence of erosion processes, e.g. water of the sea or a river cutting through some soft rock or sediment formation, while ridges may be results of uplifting and folding, or deposition processes (ice-sheet, wind). But of course there can be exceptions here, and this is just a rough guide. I would also strongly recommend you to digitize the course of the natural=cliff (ridges) more accurately, following the top of the cliffs, as represented e.g. by the contours in OpenCycleMap. This will make an ultimate rendering of the cliffs or ridges together with contours or shading much more useful. E.g. look at the accurately drawn cliffs / ridges in the first image I posted with the comparison of Snowdon Mountain in North Wales in both OpenStreetMap Carto, and my own renderer:
(I am assuming here the location of the contours is accurate, if your GPS/GPX trace says something different when walking past the edge of a cliff, it may be wise to use that GPS data in some cases). |
Dear mboeringa Sorry, no they are not cliffs. Please trust me, viewing this from OpenCycleMap is misleading you. I mean, I go to Serra da Moeda frequently, and I can assure you it is a ridge which is what the name "Serra" actually means. In fact "Serra" literally means a "saw" and probably got this name because the sequence of peaks resemble a saw when facing upwards. The name has nothing in common with the Portuguese name for "cliff" which is "penhasco". I mean, I know what a cliff is, here is an example which I mapped Serra Geral de Goiás is bordered by a lower plateau and some parts are indeed cliff-like, but no, not the whole structure. I mean tagging the whole 200km (or more) of that Serra as a cliff because some parts are cliffs sounds strange to me. What you could argue though is that they are perhaps better described as "Mountain Ranges", but this tagging is not covered by OSM rules. Probably that is the reason why neither Andes nor Alps show up in nominatim. Therefore, the closest match at present for serra is ridge and not cliff. Here is another example of a large structure if you like, Serra do Espinhaço
Of course, given enough time, I would be more than happy to get a more accurately mapping of ridges, but we are very short of mappers in our region. At present I am trying to get as many ridges into the map as possible, even with an just approximate location. Why? Because it looks very stupid when a local user searches OSM for the name of a famous ridge and nothing at all shows up. Therefore I am afraid that accuracy will have to wait a bit. Anyway, thanks a lot for your comments. |
Not to say that they are definately ridges, but they've got tertiary roads crossing them (e.g. http://www.openstreetmap.org/way/318209935#map=15/-20.2713/-43.9533&layers=C). Taking a look at photos of that road, it's not a cliff. I had a look on MQ to see if we're doing anything with ridges or similar features, and all I spotted were some very large-scale labels like Rocky Mountains |
|
In any case, alignment with contours doesn't matter here - we don't render contours, and they are examples of long natural=ridge features, which do exist. |
OK, I agree your local knowledge is more valuable than my looking at OpenCycleMap. So if you know they are better classified as ridges, than that is fine with me.
_However_, if the general consensus in your country is, and I can well imagine this considering the size of the features you are digitizing, that this is a "mountain range", than please add a natural=mountain_range tag instead of declassifying them to natural=ridges just for rendering... Of course the distinction between these two is a bit vague, but I would rather think that features this big (hundreds of kilometres), would generally be better served with natural=mountain_range. The tag is already unofficially documented on this proposal page: http://wiki.openstreetmap.org/wiki/Proposed_features/Mountains Although OpenStreetMap Carto doesn't yet support this tag in rendering, you can always post another issue here on the bug-tracker with a request to do so. In addition, my OSM Renderer for ArcGIS that I have in development, already supports rendering of this tag. In the below image, starting at the red arrow, you see the name of the Great Dividing Range in Australia, tagged as natural=mountain_range, as just letters spread out along its major length, as most topographical maps display this type of feature:
|
I wonder where this data is coming from? Running an overpass turbo query for areas, nodes, ways, relations using natural=ridge, natural=mountain_range and natural=massif, doesn't return anything named Rocky Mountains in the western US... (it does return other named features though). Is it Natural Earth? natural=ridge natural=mountain_range natural=massif |
|
To hopefully clear up confusion a bit: http://www.openstreetmap.org/way/318565962 is an escarpment - in large parts forming cliffs, in very small parts possibly also ridges. Neither of them are continuous along the whole thing. http://www.openstreetmap.org/way/318209935 is mostly ridges, but with several interruptions, a ridge is supposed to be a continuously convex structure. http://www.openstreetmap.org/way/248919543 is neither - it is a mountain range, strongly structured so ridges and possibly cliffs - if they exist - only extend a few kilometers.
There are taggings in use for mountain ranges, some maps also render them. Like: http://www.openstreetmap.org/relation/2127978 As far as rendering goes - i still think a fairly thick gray line would be a good option for this style. The examples mboeringa showed are not that useful i think since they render contours and showing ridge lines together with contours is problematic at best. Of course ideally ridge lines should be interrupted by other features along their path (roads, peaks etc.) - not sure if there is a good way to do this in mapnik. It does not have to look perfect, the most important thing is to properly show what mappers have mapped without disturbing the map too much. |
The two examples you show actually use another tagging scheme than natural=mountain_range. They use: This is probably justified, since these are subdivisions of the bigger Alps and Himalaya natural=mountain_range, so the tag region:type=mountain_area seems suitable.
I am not to sure if that is a good basis or justification for any render rule development, if i take your assumption to the extreme, I could say: drop _all_ rendering using carefully crafted symbols, colors and labels, just show nodes and ways as in _ID's or JOSM's editor window_ because that shows every last node and way ever digitized in OSM even if it is not rendered on any style yet... I guess we don't want to go there... I do think careful thought should be given to any type of rendering. I am not yet sure if I will keep the ridge rendering as I currently have it, or only start rendering named ridges, by showing their label stretched along the digitized line, as in valleys. |
I did not mean this as a general rule for anything mapped but specifically for natural=ridge which is (a) widely mapped - >15000 times (b) is not at all uncommon to be shown in maps in general (c) other tags, i.e. natural=cliff, are widely abused to make ridges show up in the map and (d) it is repeatedly being requested to be added to the style. |
I am not disputing the sense of rendering ridges at all..., that's why I am investing time in this, I just think it needs careful consideration how. And if it needs just a label, or that a "physical" rendering in the form of a line is necessary... Thanks for the clarifications about the map features @drgweber pointed out, and your opinions about them. As you already more or less pointed out in your clarifications, considering the size of the features, is probably vital in terms of calling something a natural=mountain_range or a natural=ridge. To my feeling, natural=mountain_range should be used for features in the minimum 5-10 km wide , to 100's of kilometres length ranges. So this includes the: Serra Geral de Goiás which clearly extends for 100's of km. natural=ridge is probably best suited for anything in the range of a minimum 100 meter wide to 5-10 km length range or so. Ridge is a kind of "micro-structure" feature of mountain_range, just like valleys probably are. Maybe an image explains it better, at least how I currently interpret these tags, or how they possibly should be interpreted, @imagico, let me know what you think...:
|
|
Well, to get a clearer picture of this, I decided to try and locate this place of the above image - Serra de Moeda - on Google. This gives the image below. As @drgweber pointed out, the "contours" were slightly confusing. It does look like the original line digitized by @drgweber could indeed be interpreted as a rather big natural=ridge, albeit probably an eroded escarpment... The other ridges stuff I marked in the above image, is in reality probably to insignificant to bother to map it (except maybe the right side). Of course, this is always a question, small ridges do exist, and are they worth mapping? That is a question that probably only locals can answer, do they have significance and possibly names locally...
|
|
When I started to map the Serras in this region my understanding was that "Serra" translates best as "ridge". The official IBGE maps which we are allowed to use for OSM simply show in capital letters the approximate location of these Serras, they do not go into any geological details. At the same time, the only documented/approved tag I found on OSM's wiki was natural=ridge. So it was a natural step to map all the Serras as ridges. It is now clear to me that it best to revise this. After your comments, I am realizing that English word "ridge" is being used in this discussion in a very strict context, while the Portuguese "Serra" seems to be applied in a very broad sense. (Please correct me if I am wrong, English is not my first language.) I don't know if the related Spanish "Sierra" is used in a similar broad context, but I suspect it is. On the other hand this wiki page http://wiki.openstreetmap.org/wiki/Proposed_features/Mountains as well as the http://wiki.openstreetmap.org/wiki/Proposed_features/ridge do not give me the impression of such a strict context for ridges. So clearly, there seems to be the need for a revision of these tags to get some consistent definitions. Still, there are numerous Serras that fit the strict definition of ridges, one of them is Serra do Curral in which defines de border between Belo Horizonte (where I live) and Nova Lima. Another point that I am understanding now is how difficult it is to pin-point Serras from aerial images. Historically of course these Serras were named by looking at them from the ground. What you see from the ground are seemingly continuous stretches of peaks running along the horizon which completely dominate the landscape. For example the interesting picture @mboeringa found of Serra da Moeda doesn't make justice to the view you get when driving alongside. Clearly, it is not at all straightforward to correlate the known Serras to what is shown on Bing images or altimetric contour lines. This is one of the reasons why I have not yet been able to map several other Serras in my region, such as Serra da Gandarela. |
|
In Spanish and Catalan there's the same meaning about ridge .In Catalan also it sounds and have to be written like Portuguese : SERRA. Ridges are very important for our territories.There are studied at the schools and they are one of the most important part of our environmental joyness. I think about Serra de Collserola , near Barcelona city or Serra del Cadí near Pyrenees. |
|
OK. |
|
Reopened, since |
|
Closed by #3767 |
Ridges are important geographical objects which, to my knowledge, are not currently rendered in carto style
They can range in length from 1 km to over 100 km and are commonly shown on printed maps in large upper case letters (see public domain image from IBGE, Serra=Ridge). Current usage is around 17000 http://taginfo.openstreetmap.org/tags/?key=natural&value=ridge#overview, but I guess usage would be higher if rendered. It also seems that natural=ridge was used in many cases instead of natural=mountain_range because the key ridge is formally defined while mountain_range is not.
In South America ridges are very important reference points defining whole regions, for example the Serra da Mantiqueira http://www.openstreetmap.org/way/266727831. Frequently, when planning trips the lack of rendering can be an important drawback since passing over ridges can be difficult or impossible.
While important ridges are commonly shown on printed maps, commercial digital maps (e.g. Bing, Google) usually do not show them. Therefore, this could be a nice opportunity to add a distinguishing feature to OSMs main rendering style.
My suggestion is to render at zoom levels smaller than 10, but perhaps this would depend on the length of the ridges.
Similar issue seems to be #1097
The text was updated successfully, but these errors were encountered: