OpenStreetMap

Towers and Masts

Posted by BushmanK on 29 November 2016 in English (English)

Browsing through the issues at Openstreetmap-carto (also known as OSM Standard style or "Mapnik" style) tracker on GitHub, I came across several issues, both open and closed, touching the topic of rendering vertical man-made structures such as poles, masts, and towers.

Communication engineering was my thing for awhile, so it always strikes me when at least two of these terms - mast and tower - are used in an uncertain manner. In one of the discussions on GitHub, the difference between masts and towers was called "philosophical". Actually, there is no philosophy (at least if you don't look at wrong and misleading examples in OSM Wiki). Because of that, I've added an engineering definition to pages of man_made=tower and man_made=mast both in English and Russian because what is in the first section of those pages makes zero sense and contradicts the basic principles of tagging, because it uses comparative terms such as "bigger" and "smaller" to distinguish between these structures. Tags tower:construction=guyed* are obviously redundant because if you need that, it means that object must be tagged as a mast, not as a tower.

I didn't want to rewrite the whole "definition" without discussing it, while I don't really believe that discussion could be successful, so I just added clear definition in case if someone would prefer it. Just for the reference:

Mast is a vertical man-made structure, supported by the guy lines and the anchoring system.

Tower is a vertical free-standing man-made structure, supported by its own foundation only.

(Anyone can find it even in Wikipedia, so it makes me wondering, how ignorant an author of these OSM Wiki articles was to write that.)

And it doesn't matter, that some contractors (and regular people after them) calling cellular communication towers "masts". It is not only wrong as it is to use "transistor" to call a radio receiver or "Xerox" to call a copy machine (which is common in some languages), but it makes it impossible to actually distinguish masts from towers for mapping purposes.

So, getting back to rendering, both "inverted T" and "inverted Y" are completely appropriate for tower symbols. Inverted T looks like a tower with a single stem or column, standing on its foundation, tower:construction=freestanding. Inverted Y looks more like a rough outline of a steel lattice tower (more strokes could be added to make it look fancier), tower:construction=lattice.

Masts are a bit more tricky, but just a bit. The most obvious symbol is an "inverted bird foot" symbol, similar to inverted Y with the central stroke, extended all the way to the bottom. It also looks like an inverted antenna symbol used for circuit diagrams. Central stroke represents the mast itself, diagonal strokes represent guy lines.

As a bottom line, rendering of masts and towers is not solely a question of style and preferred icons, it's also a question of using proper definitions. If definitions will get clarified one day, no philosophy will be involved in rendering and tagging anymore. (Personally, I really doubt that it will happen.)

Added from comments: These tags currently do not have "OSM-specific meaning", they are completely mixed into one mess - it is technically impossible to be sure if an object, tagged with man_made=mast is a mast and vice versa. So, changing anything can't do any harm, because it can't be messed up more than it currently is.

People are arguing about that only because almost every person has an own tradition of tagging and thinks that all others have a similar one. But it's not true - different objects are tagged similarly by different people as well as similar objects are tagged differently by them. Belief, that there is any global consistency in tagging masts and towers is just a fallacy.

Comment from TomH on 29 November 2016 at 09:09

Tags should mostly be view as opaque keys to be looked up in a reference source to determine their meaning rather than expecting them to be technically accurate on their own.

Although tags often use words which have a very specific meaning in a particular domain they were often created without full understanding of the details and have therefore simply become placeholders and we normally avoid going back and changing them to reflect a correct understanding and instead treat them as having OSM specific meaning.

Hide this comment

Comment from BushmanK on 29 November 2016 at 18:36

@TomH, I'm very well aware of the abstract nature of tags. I even explained different issues linked to this concept in several diary entries.

However, this particular case is a kind of different. Usually, uncertain tags are reflecting uncertain terms of natural language, just like in the case of amenity=fast_food and amenity=restaurant. There is no any clear definition of a difference between these two amenities. While masts and towers do have clear and simple distinctive features while corresponding tags not only have the unclear original definition but this definition also contradicts basic OSM principles (no relative values for tags, verifiable values).

These tags currently do not have "OSM-specific meaning" as you call it, they are completely mixed into one mess - it is technically impossible to be sure if man_made=mast is a mast and vice versa. So, changing anything can't do any harm, because it can't be messed up more than it currently is.

People are arguing about that only because almost every person has an own tradition of tagging and thinks that all others have a similar one. But it's not true - different objects are tagged similarly by different people as well as similar objects are tagged differently by them. Belief, that there is any global consistency in tagging masts and towers is just a fallacy.

Here, I'm saying that proper rendering of malformed data might promote fixing it with proper tagging. And I have an example of a similar situation. In Russian, "kiosk" refers to a small shop in general, not to specific product range of a small shop (as it is in many European languages). Right after shop=kiosk got its own newspaper icon in OSM Standard style, Russian mappers started fixing hundreds of improperly mapped amenities by setting proper values such as ice cream, tobacco and so on. I don't see any reason why it can't work the same way with masts and towers.

Hide this comment

Comment from BushmanK on 29 November 2016 at 18:46

By the way, I can easily draw SVG icons for these objects to use them in OSM Standard, I just don't want to do any unwanted work if this idea will get negative reception.

Hide this comment

Comment from GinaroZ on 29 November 2016 at 21:51

I agree it is confusing, especially when the wiki's "Is it a mast or a tower?" examples all look very similar yet are defined in three different ways. Then you have osm-carto which only renders a mast, so many people will just add a mast to the map so that it gets rendered when it might indeed be a tower.

For example in the UK - this is mapped as a tower when it should be a mast as it is guyed: https://www.openstreetmap.org/node/872106782 - https://en.wikipedia.org/wiki/Winter_Hill_transmitting_station

In the UK cell site are normally called "mobile phone masts" when technically they are not masts but rather small towers. But then if you add them as a tower/communication_tower they don't get rendered on the map.

Perhaps you could find a number of worldwide examples, with both the technically correct, colloquial word for each structure and OSM definition so we can see how big the problem is?

Hide this comment

Comment from BushmanK on 30 November 2016 at 00:23

@GinaroZ,

Right, there are several factors affecting the situation in addition to the unclear definition in Wiki, including common "natural language factor" (or "tag as you call it") and mapping for the renderer (specifically, OSM Standard style).

I definitely can do some analysis of this problem, however, it is impossible to tell, how many objects are tagged according to the engineering definition, because there is no reliable reference data source I can use for comparison (or I'm just not aware of one). So, it is only possible to compare OSM objects with a street view photos or something like that. An educated guess is impossible because we don't have reliable information about "mapping traditions" either.

For now, I can tell that there are 196363 "towers" and 35422 "masts", and nobody can actually tell anything clear about two hundred thousand objects, are these really towers or masts.

Hide this comment

Comment from dieterdreist on 30 November 2016 at 08:36

Actually there is more to towers and masts than just looking whether they are freestanding or guyed. There are indeed "guyed towers". There are several definitions, and there are subtypes. For example a "power tower" is quite different from a man_made=tower. For towers, the key "tower:type" is quite important.

Another distinction for example is whether the structure is made for people to access (and maybe remain as opposed to maintenance only). Masts are not buildings (in a technical way of "building") while towers are (in OSM "building" is also often used for what technically are structures and not buildings).

This being said, I do agree that the difference between a mast and a tower is not the size. I would use a criterion like: can you (people) enter the structure, and would also look at freestanding vs. guyed, but accept exceptionally towers to be guyed, and generally masts to also be not guyed, for instance to me these are all masts: http://www.windturbinestar.com/uploads/images/Type%20of%20towers.png This one could be seen as tower: http://media05.myheimat.de/2014/04/26/3077610_web.jpg?1398514834

Basically, accessibility (for more than maintenance) should be the main distinctive criterion.

Hide this comment

Comment from dieterdreist on 30 November 2016 at 08:40

@TomH while I agree it can be seen like this, (and is actually the case with some tags), it is kind of risky, because most people do not look up definitions, and editors presets present the tags mostly with a one-word description that repeats the tag name, so the actual word used for the tag IS important. It's almost impossible to establish a tag with a different meaning than the word that is use to describe it.

Hide this comment

Comment from GITNE on 30 November 2016 at 14:42

@BushmanK Thank you for shedding light on and explaining a very important topic in your diary entry, which basically comes down to definitions. I am all for clear definitions.

People are arguing about that only because almost every person has an own tradition of tagging and thinks that all others have a similar one.

This is true. But traditions and thinking to oneself have never brought any progress nor have been ever good advice in the long run. Now, I am not saying that traditions are a bad thing per se, since every progress which has been proven to be correct and effective can or should become a tradition, yet every tradition should constantly be under scruteny for improvement, sometimes even for abolition or revolution.

Belief, that there is any global consistency in tagging masts and towers is just a fallacy.

This is also true. However, it is never to late to start a standardization or term (tag) definition process. This is exactly what organizations like ISO, W3C, IEC, or national agencies like ANSI or DIN, or even professional associations like the IEEE do. All of these organizations not only work on standards but first and perhaps foremost on definitions, mainly to establish a common effective and efficient language for communication between members and implementers of standards. Basically, the same applies to definitions of tags in OSM. If you have good thoughts on definitions and the capacity to produce definitions (or propositions) accepted by a lot of people (mappers) then spending time on this is as helpful as mapping itself. Sometimes just clarifying definitions to accommodate for the ever evolving changes in use of language may be as helpful as developing a new definition. So, please do not stop here or let anybody tell you that things cannot be improved. ;-)

because most people do not look up definitions

@dieterdreist There is no excuse for ignorance or laziness, although I am aware that you probably did not intend to excuse this sort of behavior or had that in mind while making your statement—you've simply been describing what uninformed masses do. I have found that learning definitions and pondering on dissociation of definitions is key to enable anybody to become an expert in any field or to simply learn about the world which surround us, and finally to enable one to efficiently communicate with others. So yes, I would love and urge everybody to actually learn definitions (in every field or subject) instead of just relying on vague self indulged feelings about terms and their definitions. Learning definitions is indeed a tedious task but in the end it is very rewarding for better understanding the world, of course including OSM.

It's almost impossible to establish a tag with a different meaning than the word that is use to describe it.

I did not get the impression that this is what @BushmanK had in mind. @BushmanK's conclusion seems to be rather that the description of definitions of man_made=tower and man_made=mast should be made more clear first, and only then new tags could be introduced to make further, more precise distinctions in types of towers and masts.

Hide this comment

Comment from GinaroZ on 30 November 2016 at 15:07

@BushmanK didn't realise there were so many towers mapped - even more reason why they should be rendered on osm-carto! I like your inverted T/Y idea, and I think for prominent masts/towers (height>100m?) they should be rendered at a higher zoom level than the z17 currently for masts.

Reading the wiki it does make me wonder, what is the point of having a third tag - man_made=communications_tower? It's apparently for a "huge tower" but in the UK it's being used both for mobile phone masts and some, but not all, large towers. So should this tag be discouraged, given it's only been used ~ 3,800 times? Surely tower+tower:type+height is better?

The wiki's photo of the "mobile phone mast" for a mast is also confusing. That should probably be changed, and a proper description of how to tag a "cell site" or "base station" added.

Editors could be improved as well - iD has man_made=mast described as Radio Mast, and other presets could be added.

Here's some examples to consider, from a good site for UK/Ireland photos:

http://www.geograph.org.uk/photo/310144 - mast, but not shown anywhere in the wiki

http://www.geograph.org.uk/photo/4952269 - probably a tower?

http://www.geograph.org.uk/photo/3383904 - again a tower, but a mast in the wiki?

http://www.geograph.org.uk/photo/5198501 - tower

http://www.geograph.org.uk/photo/1613094 - it is guyed so should be a mast, but needs additional tags to add detail. A good example of why man_made=mast shouldn't be rendered on its own!

http://www.geograph.org.uk/photo/803411 - towers

Hide this comment

Comment from BushmanK on 30 November 2016 at 16:42

@dieterdreist, I disagree with the most of the arguments you've presented (and some of them are targeted to statements I didn't really make and had no intention to make).

Indeed, there are borderline and mixed cases in the real life, however, these are exceptional, rare cases. I'm talking about guyed towers. The vast majority of vertical structures can easily be classified using a simple rule about guyed/cantilevered (freestanding) design, while tiny fraction (my rough estimation is one per ten thousand or even less) should be examined more thoroughly. This doesn't make the engineering definition unacceptable for the OSM.

Second thing, I didn't say that tower:type=* is redundant. But tower:type=guyed* is, because it means that in the vast majority of cases it's a mast, not a tower. This is a bit different, right? tower:type=* should be used to describe the design of the main stem without mixing it with other features such as support method. So, for the very rare case of a guyed tower, extra separate tag similar to tower:guyed=yes could be used. But let's not mix it with a stem design.

We are going away from the main topic of tower/mast classification, while I feel like I need to address the whole scheme to convince readers that my vision only makes things more clear, not more complicated.

While many towers have room inside it and could be also classified as buildings, we must not mix these features or make one a requirement for another. The presence of room inside a structure must only affect classifying it as a building.

Many high (about 50m and higher) radio transmission steel lattice masts have cross-section large enough (to withstand the near-axial load) to have a system of built-in ladders inside. And these still are masts. In the same time, many lattice towers (similar to common power towers by design) have only bracket-shaped steps welded to it, so there is no room or even an access space inside it. And these are, by all means, towers. Another thing is that means of access are completely secondary to the whole design of a structure and independent of it. Therefore, accessibility or means of access can not and should not be used for classification.

So, looking at this picture you've provided, I can clearly say that only the left one is a mast, not all of them, as you said: wind turbine installations The one on the left definitely can not stand straight by itself and it is supported by an elaborate system of guy lines attached to it at three levels, which tells us that the stem is flexible enough to break if it will be supported at one point only. Two others don't have any guy lines and they obviously don't need ones, because they are rigid enough to withstand lateral force (bending - because of a larger progressive cross-section, and falling - because of massive bolted flange in its foot). So, these are definitely two towers.

The second picture you provided represents a structure with a central supporting mast: Aussichtsturm Wismar. And it is pretty clear too because it has another massive guy line system. However, from the engineer's point of view, the reason for adding that system is not the same as usual - it helps to withstand uneven vertical load caused by people walking by that fancy stair, not uneven lateral load caused by a strong wind. By the way, Aussichtsturm Wismar is usually described as the structure with a central supporting mast.

Speaking of people who don't read Wiki - that's a factor I'm definitely taking into a consideration, but I have the completely different conclusion. Having a proper map symbol and editor preset icon could give them a hint. Simpler shorter definition used in editor such as "guyed mast" also helps. Unlike that nonsense currently describing these structures in OSM Wiki that brings only confusion and nothing else even to those who want to learn a good definition.

Hide this comment

Comment from BushmanK on 30 November 2016 at 17:06

@GITNE, thank you for your support and understanding.

Indeed, my original intention is very simple:

  • to finally clarify these two definitions (which I already partially made a year ago by adding engineering definitions to OSM Wiki);
  • to promote it by introducing consistent symbols for OSM Standard style (which I promise to draw).

This by itself could improve the whole situation about vertical structures. Additional things I've mentioned, such as tower:type=* usage, are secondary to it and, therefore, optional.

I'll repeat that for the third time: currently, more than 200000 objects are tagged "by feel" because definitions do not make any practical sense. So, by sticking to an engineering definition, we can't make things worse because information stored in these two tags is unusable.

Hide this comment

Comment from BushmanK on 30 November 2016 at 17:23

@GinaroZ, by going deeper into other related issues such as large towers like ones historically used for the analog TV broadcasting, we increasing a chance to spread the effort, which is something I don't want to cause. I agree with you that there are more or less significant mess and inconsistency in several schemes, describing vertical structures and communication infrastructure, but let's just focus on the mast/tower issue since it seems independent enough to be eventually resolved in a positive way.

Hide this comment

Comment from BushmanK on 30 November 2016 at 20:39

@dieterdreist, I believe, you've replied to me via private message accidentally, so I'll continue here:

By the way, Aussichtsturm Wismar is usually described as the structure with a central supporting mast.

yes, there's no doubt the core is a mast, still the structure altogether (mast, stairs, platform) qualifies as tower if you look at the criterion made for people (unlike those masts you wrote about above, which have ladders in the inside, i.e. are not intended to be climbed besides for maintenance work).

If you think that vertical structure should be tagged as man_made=tower in case if it's "made for people to be in there", then you should be aware of contradiction (redundancy, actually) with tower:type=* key. This key indicates a function ("type" is a very bad choice, by the way) of a tower, almost directly telling that certain types are "for people to be there" (observation, bell tower and some others). So, this information is already carried by tower:type=* key, therefore, it would be wrong to use it for something else, like distinguishing between tower and mast.

And I still can't see any reason why very simple widely accepted engineering definition that doesn't require any special knowledge is so bad that something vague should be used instead of it.

Hide this comment

Comment from GinaroZ on 1 December 2016 at 13:04

but let's just focus on the mast/tower issue since it seems independent enough to be eventually resolved in a positive way.

Well that's why I posted the example images, to see if you thought they were a mast or tower.

The point I was making with man_made=communications_tower is, if it is just the same as man_made=tower + tower:type=communication then surely the single tag can be eliminated? Having a third tag just adds confusion, results in inconsistent tagging and means there's another tag to render.

Hide this comment

Comment from BushmanK on 1 December 2016 at 16:50

@GinaroZ, while I agree with you that since there is no definite criterion to distinguish man_made=communication_tower from man_made=tower tower:type=communication except some nonsense like "Really big ..." (which is an explicitly discouraged practice), I don't want to say anything about elimination of the whole tag, because, unfortunately, OSM often works by the same rules as politics - it could be enough to mention something that makes certain people uncomfortable to be labeled as a revolutioner in a bad meaning of this term, so the whole idea will be discredited.

As a gesture towards those who can't just accept the idea of "really big" communication tower being equal to any other (communication) tower, I don't see any problem in keeping their cherished tag, if they'd be able to come up with a real and usable distinctive property like certain size (which is usually known for "really big" remarkable towers) or a presence of working space, intended for a permanent use by communication personnel, inside the structure.

Regarding of pictures you've posted, my opinion on it doesn't matter: engineering definition is simple, anyone can apply it and you are most likely right in conclusions. Don't take it as something impolite - I just want to avoid being referred to as an expert - anyone can be an expert, using simple definitions.

Hide this comment

Comment from BushmanK on 1 December 2016 at 17:13

@dieterdreist, I'm pulling another reply from PM:

If you think that vertical structure should be tagged as man_made=tower in case if it's "made for people to be in there", then you should be aware of contradiction (redundancy, actually) with tower:type=*

there's no contradiction or redundancy, tower:type is a subkey to further specify of which type a tower is.

tower:type=* indicates a function, but often, function presumes a space for people by its nature.

Anyway, you haven't answered my last question, why exactly structural characteristics are less acceptable than a presence of space for people. This is actually pretty important question because there could be any number of views and opinions, but only the direct comparison can help selecting the most appropriate one. I have presented both my idea and arguments in favor of it (as well as certain arguments against your idea), but I haven't seen any arguments in favor of your idea, just explanation of it and arguments against my idea (which I a kind of managed to logically dismiss). If it's a formal discussion, it makes perfect sense to tell, why your method should work better (or to try dismissing the whole issue like non-existent, unimportant or something).

Hide this comment

Leave a comment

Parsed with Markdown

  • Headings

    # Heading
    ## Subheading

  • Unordered list

    * First item
    * Second item

  • Ordered list

    1. First item
    2. Second item

  • Link

    [Text](URL)
  • Image

    ![Alt text](URL)

Login to leave a comment