[talk-au] highway=track update

Michael Collinson mike at ayeltd.biz
Tue Feb 23 17:38:20 UTC 2021


 > Any OSM old-timers recall enough to comment?

As the oldest-timer in Aus mapping (2005), I'll take that one.

TL;DR:  Looks like a non-native English speaker individual created a 
continuum of of sand - fine_gravel - gravel in 2008 when sand - gravel - 
rough_gravel or just sand - gravel would have fitted the vernacular 
better.  Never discussed or challenged. Probably because the Brits and 
perhaps other Europeans with a mostly sealed public road network use the 
word "gravel" too broadly to be useful elsewhere.

The broad answer is that we just didn't define terms in the early days 
... and were very UK-centric, i.e. assumed a term was self-evident and 
true everywhere. The next wave was the nascent map features page 
https://wiki.openstreetmap.org/wiki/Map_features and then individual 
pages - where terms suddenly got definitions but by individuals such as 
myself with thought but not discussion. Often that worked but it just 
needs one person with some local bias to push it off. And again they 
tended to be dense population Euro-centric.

surface= values suddenly went from the undefined to very detailed in 
August 2008 thanks to the work of German (so presumably non-native 
English speaking) mapper Joerg:
https://wiki.openstreetmap.org/w/index.php?title=Key:surface&oldid=138642

Reviewing it now (finally!), I think he did an excellent job overall. 
There is a continuum of sand -*fine_**gravel* - *gravel* which kinda 
fits the different subjective experience of walking, riding, biking, 
driving. And "compacted" has its own definition.

In hindsight, I see two issues:

1) sand - *gravel - **rough_**gravel* would better have fitted the 
vernacular or perhaps sand -*gravel - **aggregate* (though an engineer 
or scientist might argue with aggregate).  As a trained geologist, rock 
chips between about 0.4 cm and 6 cm largest dimension are formally 
"gravel" but if you want to subdivide, hey, why not.

2) I blame my fellow Brits (in and outside OSM)! Virtually all public 
roads there are now sealed so they don't have the experience and just 
call anything unsealed with a few or many rock chips of any size 
"gravel".  This is a recurring problem when defining semantic 
ontologies. A word can have both a very broad meaning (sensu latu) and a 
very tight, precise meaning (sensu stricto). And meaning can also be 
different according to discipline, a road engineer may have a different 
understanding to me as a geologist.

In summary, (1) is too late to fix, so we should live with it. That 
probably means that many true gravel roads should be reclassified to 
surface=fine_gravel. And then my personal bugbear (2), a huge number of 
gravel roads just aren't and should be classified "compacted" - would 
very much appreciate other thoughts, it is how I map but I've never had 
a chance to converse about it.

 > Is there actually anywhere in the world where roads are commonly done 
this way?
Never seen it in common usage. In Sweden, and so perhaps other countries 
with winter ice, I do sometimes railway ballast size gravel used for 
logging roads for heavy, and I mean heavy, machinery and trucks.

Mike


On 2021-02-23 11:59, Josh Marshall wrote:
> This raises the question: how did the surface=gravel tag end up 
> getting defined as large aggregate/railway ballast anyway, given it 
> appears at odds with almost everyone’s usage of it, including other 
> significant online references such as: 
> https://en.wikipedia.org/wiki/Gravel_road 
> <https://en.wikipedia.org/wiki/Gravel_road> (which matches the 
> vernacular perfectly)?? Any OSM old-timers recall enough to comment? 
> Is there actually anywhere in the world where roads are commonly done 
> this way?
>
> With regard to:
>> Hi Josh and co, I ride a “gravel bike” on dirt roads that are 
>> signposted as “gravel road”but definitely don’t fit the OSM 
>> definition of gravel = railway ballast.
> and Michael’s
>> I don't map much in the US but do in Australia and Sweden. In both 
>> countries, I have rarely come across what I consider to be gravel 
>> roads, instead consider most unpaved roads and tracks to be 'dirt' or 
>> 'compacted':
>
> Same here. I might provide a single counter-example; the major through 
> road in the Watagans near me was actually lined with this large 
> ballast last time I rode through; an absolute nightmare to ride on, 
> and I can’t imagine it’s too kind on vehicles either. Presumably an 
> initial step before further surfacing? Has anyone else seen this surface?
>
>
>
>> On 23 Feb 2021, at 8:44 pm, Little Maps <mapslittle at gmail.com 
>> <mailto:mapslittle at gmail.com>> wrote:
>>
>> Hi Josh and co, I ride a “gravel bike” on dirt roads that are 
>> signposted as “gravel road”but definitely don’t fit the OSM 
>> definition of gravel = railway ballast. Because of the common usage 
>> of gravel as a variably textured dirt road in Australia, we face a 
>> massive uphill battle to get accurate, specific unpaved road surfaces 
>> in OSM. Here’s some data from Overpass Turbo queries of all unpaved 
>> highway surfaces in Victoria. This includes all highway tags (inc 
>> roads and paths) not just tracks:
>>
>> Surface 	       Number 	           Percent
>> unpaved 	48664 	80
>> gravel 	6159 	10
>> dirt 	4559 	8
>> compacted 	642 	1.1
>> sand 	406 	1
>> fine_gravel 	230 	0.4
>> earth 	46 	0
>> Total 	60706 	100
>>
>>
>> In case that’s illegible, if you add all of these unpaved/dirt/gravel 
>> ways, 80% are tagged with a generic unpaved tag (which is entirely 
>> accurate if not especially precise). Gravel is the next most common 
>> category, accounting for 10% of ways. Apart from dirt at 8%, the rest 
>> are used very rarely.
>>
>> My guess from tagging surfaces on a lot of unpaved roads is that 
>> perhaps 80% of the roads tagged as gravel do not satisfy the OSM wiki 
>> definition and should be tagged as something else. Interestingly, the 
>> two most relevant tags for formed, unpaved surfaces - compacted and 
>> fine_gravel - are very rarely used (around 1% each). There are 
>> probably more ways that have fence-sitting tags like “dirt; sand; 
>> gravel” that end up being pretty meaningless.
>>
>> Adding precise surface tags may be simple on roads that are freshly 
>> maintained but on roads that haven’t been maintained for a while 
>> they’re often pretty difficult to assess anyway.
>>
>> Personally, I feel that there’s often too much emphasis in OSM on 
>> precision (i.e. use detailed sub-tags) at the expense of accuracy. I 
>> believe most of the generic unpaved tags are accurate. I wish I 
>> could, but unfortunately I don’t believe many of the specific 
>> sub-tags are especially useful. (Sand is a goody though!). Cheers Ian
>>
>>> On 23 Feb 2021, at 5:22 pm, Josh Marshall <josh.p.marshall at gmail.com 
>>> <mailto:josh.p.marshall at gmail.com>> wrote:
>>>
>>> 
>>>
>>>     The approved OSM tag for surface=gravel
>>>     <https://wiki.openstreetmap.org/wiki/Key:surface> refers to
>>>     railway ballast, not the fine crushed rock or natural surface
>>>     that usually occurs on unpaved roads in Australia. However we
>>>     call the fine unpaved surface "gravel" in common parlance, and
>>>     many unpaved roads that don't constitute gravel as described in
>>>     the OSM wiki have been tagged as gravel here, erroneously
>>>     depending on your point of view.
>>>
>>>
>>> This is a matter of interest to me too. I spend a substantial amount 
>>> of time running+riding on fire trails in NSW (all highway=track), 
>>> and the surface type is useful and indeed used in a number of the 
>>> route planners I use. I have changed a few roads back to 'unpaved' 
>>> from 'gravel' due to the rule of following the description in the 
>>> surface= guidelines rather than the name.
>>>
>>> My question then however, is exactly what to tag the tracks beyond 
>>> "unpaved".
>>>
>>> There are definitely sections that are somewhat regularly graded and 
>>> appear to have extra aggregate/fine gravel added. From the surface= 
>>> wiki, these most closely align with surface=compacted. But 
>>> fine_gravel is potentially an option too. Many of these are 2wd 
>>> accessible when it is dry. (Typically smoothness=bad.)
>>>
>>> There are also others, usually less travelled, which are bare rock, 
>>> clay, dirt, sand, whatever was there. Is it best just to leave these 
>>> as surface=unpaved, and add a smoothness=very_bad or horrible tag? 
>>> None of the surface= tags really seem to apply.
>>>
>>>
>>> On Tue, 23 Feb 2021 at 16:45, Little Maps <mapslittle at gmail.com 
>>> <mailto:mapslittle at gmail.com>> wrote:
>>>
>>>     Hi Brian and co, in Victoria and southern NSW where I've edited
>>>     a lot of roads, highway=track is nearly totally confined to dirt
>>>     roads in forested areas, as described in the Aus tagging
>>>     guidelines, viz: " highway=track Gravel fire trails, forest
>>>     drives, 4WD trails and similar roads. Gravel roads connecting
>>>     towns etc. should be tagged as appropriate (secondary, tertiary
>>>     or unclassified), along with the surface=unpaved or more
>>>     specific surface=* tag."
>>>
>>>     In your US-chat someone wrote, "...in the USA, "most" roads that
>>>     "most" people encounter (around here, in my experience, YMMV...)
>>>     are surface=paved. Gravel or dirt roads are certainly found, but
>>>     they are less and less common." By contrast, in regional
>>>     Australia, most small roads are unpaved/dirt/gravel.
>>>
>>>     In SE Australia, public roads in agricultural areas that are
>>>     unpaved/dirt/gravel/etc are usually tagged as
>>>     highway=unclassified (or tertiary etc), not highway=track. There
>>>     are some exceptions in some small regions (for example in the
>>>     Rutherglen area in NE Victoria) where really poor, rough 'double
>>>     track' tracks on public road easements have systematically been
>>>     tagged with highway=track rather than highway=unclassified. See
>>>     here for example:
>>>     https://www.openstreetmap.org/#map=13/-36.1424/146.3683
>>>     <https://www.openstreetmap.org/#map=13/-36.1424/146.3683>.
>>>     However, this is not the norm in SE Australia and across the
>>>     border in southern NSW, this type of road is nearly always
>>>     tagged as unclassified, as it is elsewhere in Victoria. In SE
>>>     Australia, my experience is that tracks are tagged in the more
>>>     traditional way, and not as has been done in the USA.
>>>
>>>     If I could ask you a related question, what do you US mappers
>>>     call "gravel"? The approved OSM tag for surface=gravel
>>>     <https://wiki.openstreetmap.org/wiki/Key:surface> refers to
>>>     railway ballast, not the fine crushed rock or natural surface
>>>     that usually occurs on unpaved roads in Australia. However we
>>>     call the fine unpaved surface "gravel" in common parlance, and
>>>     many unpaved roads that don't constitute gravel as described in
>>>     the OSM wiki have been tagged as gravel here, erroneously
>>>     depending on your point of view. How do you use the
>>>     surface=gravel tag in the USA? Cheers Ian
>>>
>>>     On Tue, Feb 23, 2021 at 2:49 PM Brian M. Sperlongano
>>>     <zelonewolf at gmail.com <mailto:zelonewolf at gmail.com>> wrote:
>>>
>>>         Hello all,
>>>
>>>         Recently, there was a discussion on the talk-us list
>>>         regarding how we use the tag highway=track.  That discussion
>>>         begins here:
>>>         https://lists.openstreetmap.org/pipermail/talk-us/2021-February/020878.html
>>>         <https://lists.openstreetmap.org/pipermail/talk-us/2021-February/020878.html>
>>>
>>>         During that discussion, someone suggested that Australian
>>>         mappers may also be using the highway=track tag in a similar
>>>         way to US mappers.  Hence this message :)
>>>
>>>         I've recently made edits to the wiki page for highway=track
>>>         describing how the tag is used in the USA:
>>>
>>>         https://wiki.openstreetmap.org/wiki/Tag:highway%3Dtrack#Usage_in_the_United_States
>>>         <https://wiki.openstreetmap.org/wiki/Tag:highway%3Dtrack#Usage_in_the_United_States>
>>>
>>>         If there is similarly a local variation in how this tag is
>>>         used, I would encourage the Australian community to document
>>>         their usage as well.
>>>
>>>         Brian Sperlongano
>>>         Rhode Island, USA
>>>         _______________________________________________
>>>         Talk-au mailing list
>>>         Talk-au at openstreetmap.org <mailto:Talk-au at openstreetmap.org>
>>>         https://lists.openstreetmap.org/listinfo/talk-au
>>>         <https://lists.openstreetmap.org/listinfo/talk-au>
>>>
>>>     _______________________________________________
>>>     Talk-au mailing list
>>>     Talk-au at openstreetmap.org <mailto:Talk-au at openstreetmap.org>
>>>     https://lists.openstreetmap.org/listinfo/talk-au
>>>     <https://lists.openstreetmap.org/listinfo/talk-au>
>>>
>
>
> _______________________________________________
> Talk-au mailing list
> Talk-au at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-au

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-au/attachments/20210223/2d7520eb/attachment-0001.htm>


More information about the Talk-au mailing list