Proposal talk:Peak

From OpenStreetMap Wiki
Jump to navigation Jump to search


I have a small proposal regarding rendering of the tags peak=hill and peak=knoll/hillock. Stereotypical (in reality it varies) mountain have a very distinct peak, while hills and knolls/hillocks (especially the latter) usually don't. If we're going to use different symbols for different tags, maybe it's a good idea to reflect this fact in them, and use more rounded icon for hills and (or?) knolls/hillocks.

For example:

  • peak=mountain and current natural=peak: Peak-8.svg (current peak symbol)
  • peak=hill and peak=hillock/knoll: Hillock mapnik.png

Live example:

--Psadk (talk) 18:00, 6 July 2014 (UTC)

  • I redo symbol for hill (Hill symbol.png) (because, of course, I forgot to save the xcf file...) and made a new one for hillock/knoll (Hillock symbol.png). Live example: --Psadk (talk) 07:39, 7 July 2014 (UTC)
The more I look at them, the more I like them! -- Kocio (talk) 18:07, 16 July 2014 (UTC)

General issues

For me, idea of these tags is not clearly described. The first thing I'd like to understand is what exactly should these tags reflect: height, shape, origin (geological processes), surface, or composite of these properties. In case of more than one property at once (for example: height and shape or height and surface), I'll strongly oppose this, because OSM is already full of tags, reflecting several properties in the same time because of historical reasons.

--BushmanK (talk) 23:33, 6 July 2014 (UTC)

I guess this is not a historical reason at all - people use that distinction contemporary. Relative height is the most important property here, but it can be more complex, of course - as with any definition. -- Kocio (talk) 17:30, 16 July 2014 (UTC)
Please, read carefully my statement above. I'm talking about lack of clarity in proposed terms. And this problem seems to have the same impact as tags, introduced in the beginning of OSM history, because these tags are reflecting equally non-clear entities. Currently, we have to do our best to avoid introduction of any non-clear tags.--BushmanK (talk) 23:24, 2 August 2014 (UTC)

not useful

too vague

The proposal starts with the hint that "the distinction between those land forms is more or less subjective", which already contradicts the subsequent hint to randomly set the limit to 300 m. This is like person=young/old with a random limit of 40 years. If the age is of any interest, then a number will definitely be more helpful than such a random classification. There's already ele=* for elevation above sea level, and you may use height=* or something like prominence=* for relative height.

1. The limit is the hint, because while I like to explicitly trust people to make intelligent choices (as they do all the time), I also like them to have something more concrete (our definition to be overpassed if needed). Definition can be made without mentioning trust for overpassing it, just like many others, I just want the trust to be seen.
It remains unclear in which cases an intelligent choice should overrule the definition. --Fkv (talk) 12:18, 31 July 2014 (UTC)
2.Classification, however, is not entirely subjective - there are just different criteria we can use (see [1]). That being said, look at the definition of "highway:primary" - doesn't "A major highway linking large towns, normally with 2 lanes" sound too vague, too (what is "large"/"small" town and what is sure way of recognizing the primary from secondary)? Do you propose to stick with less problematic "highway=yes" only?
Most highway=* tags were invented at a time when OSM was new and immature. They were adopted from paper maps of the UK. Other countries were not considered (look at the original definition for trunk roads), and other tags such as lanes=*, surface=* and access=* did not exist. As unclear definitions are horrific for both mappers and application developers, people began refining the definitions for each country. For example, highway=primary has been defined to mean B roads in Austria, and the tag highway=trunc is used on motorroads in many countries. These highway tags now also act as shortcuts to add implied default access and maxspeed values. There's no such a benefit for peak=* values. According to the proposal, it's merely a classification by one phisical quantity, which is the relative height. Nothing implied, no use as a shortcut. --Fkv (talk) 12:18, 31 July 2014 (UTC)
3.Height tags are welcome, of course, but how do you suggest we can use them? -- Kocio (talk) 17:00, 16 July 2014 (UTC)
See height=*. The value is the relative height in meters. --Fkv (talk) 12:18, 31 July 2014 (UTC)


Given the 100m/300m criteria, the Kleinglockner which is known as the second highest mountain in Austria, would only be a hillock because the col to the Großglockner is only 17 m lower.

It is debatable whether Kleinglockner is a mountain. Some claims it's subpeak. And author of this proposal in his rationale wrote: "Especially the micromapping phenomenon makes it important to separate some smaller, but still visible and useful land forms, from the typical alpinist/hiking aims." So if Kleinglockner is a well-known peak for hiking, it should be mapped as natural=peak and without peak=* subtag. Rafmar (talk) 22:08, 11 July 2014 (UTC)
A subkey should be applicable to all features with the parent tag. (This does not say that all possible values need to be documented beforehand.) --Fkv (talk) 12:18, 31 July 2014 (UTC)


More and more exact DEMs get publicly available. Starting tagging topographic prominence for millions of peaks worldwide would be wasted effort as these data will be able to be computed sooner or later. Derivable information like this does not even belong in the OSM database. It's the same as with contour lines. Note that ele=* is an exception, as the elevation of peaks is often surveyed even more exactly (e.g. by theodolite) than a laser scan could currently do.

1. If so, why do we really tag the peaks, while they can already be "computed" - especially mountain/high?
To map their names. Also, a surveyed position or elevation may be more accurate than the values derived from a laser scan. --Fkv (talk) 12:18, 31 July 2014 (UTC)
2. Not all peaks are made equal, we tag only those, which are of some importance for the people, not every local maximum. Some of them have names, some other have some other special meaning.
3. Extracting such informations would be computational intensive task, from the pure technical point of view.
I don't think so. It's just 2-dimensional data. And computer powers keep getting boosted infinitely over time. --Fkv (talk) 12:18, 31 July 2014 (UTC)
4. Existing (usable for OSM) terrain data are poor and are useful only for high mountains. However if we can use them for maps, I would be happy, but only as a shading/layers background, not the substitute for peak tagging. -- Kocio (talk) 17:11, 16 July 2014 (UTC)
It's not about substitution but completion. Say, if a peak has an ele=* tag, that value can be used. If it hasn't (or if the difference exceeds a configurable threshold), the value can be taken from the DEM. --Fkv (talk) 12:18, 31 July 2014 (UTC)

natural/man_made headaches

The dictinction between landuse=forest and natural=wood caused so many problems, why should we repeat them for peaks? If something looks like a peak, tag it as natural=peak, otherwise man_made=heap or barrier=debris might be more suitable. One famous example for an artificial peak is Monte Müllo, which - as the name indicates - is generally referred to as a mountain since it developped a crust of soil and vegetation.

I thought about this problem and the real solution would be to make special "terrain=*" namespace, because while many such objects are natural, some notable examples are not, and that would cut the knot. Yet I don't think such a big change is a realistic scenario for this moment and treat man_made=peak as a half step in the right direcrion. -- Kocio (talk) 17:25, 16 July 2014 (UTC)

existing tagging for archaeological sites

Mounds use to be mapped as historic=archaeological_site + site_type=tumulus. is man-made, not a tumulus, and it is mapped as natural=peak. However, I'd rather use man_made=mound as a second tag than man_made=peak Rafmar (talk) 22:08, 11 July 2014 (UTC)

--Fkv (talk) 09:15, 9 July 2014 (UTC)


Prominence and isolation are two well known data points for most named ‘peaks’. Does it make sense to work those data points into OSM so renderers can judge labeling by zoom level based on those factors? Wikidata has a fields for prominence and isolation, but prominence seems to be more widely documented. Personally, I think isolation is more useful when choosing what to label but I have no doubt that others may disagree. A Hall (talk) 13:43, 25 May 2022 (UTC)