[talk-au] Aus tagging guidelines on highway surface tags

[Bob Cameron]

Ian/all

I have been boldly marking paved and no-tag as asphalt, 
causeways/fords/bridges as concrete (etc) as a result of examining my 
own Mapillary imagery, sometimes dovetailing that with the DCS data. 
These are not only regional highways, but backroads and most of small 
towns.

I actually use the overpass query to help route plan, deliberately 
checking those with no tag. The query also showed a number of really 
strange errors, like only the ends of a rural road are paved in real 
life, yet an entire road was paved (with a surface tag) on OSM (West 
Wilcannia Rd from Menindee to Wilcannia for example)

https://www.openstreetmap.org/changeset/98337907

My Garmin GPS unit with an OSM based map tries to route me down unpaved 
roads too.. Good surface information will give credibility to OSM map 
data for general navigating.

I'll admit I have concerns that my surface tags for a paved highway 
might be removed, so I welcome the change.

Bob

On 31/1/21 3:13 pm, Little Maps wrote:
> Hi folks, wondering if I can promote some discussion about the section 
> of the Aus tagging guidelines on adding surface tags on roads. The 
> text currently reads,
>
> “For most types of highway=* tags you don't need to specify the 
> surface=paved key/value pair as this is assumed, however make sure you 
> tag the road surface when it isn't a paved road.”
>
> This assumption is fine in large cities but is problematic in rural 
> and regional Australia. Can I suggest that it is replaced by something 
> like the following...
>
> "Surface tags should be added to roads wherever possible, especially 
> in regional areas. This advice differs from that on the international 
> key:surface wiki page, which states that, 'there is normally an 
> assumption that the surface is surface=paved unless otherwise stated.' 
> However this assumption is not valid across regional Australia as: (1) 
> most roads, including many major roads, are unpaved, and (2) mapping 
> intensity varies greatly among regions. Many roads that do not have a 
> surface tag may not have been examined by mappers. Adding a surface 
> tag will assist data users and help mappers to further refine the 
> regional road network."
>
> Long rationale (not for posting on the oz tagging guidelines page)...
>
> Surface tags have been added to relatively few rural roads in many 
> regions. Hence, the most prudent assumption is that the absence of a 
> surface tag means that the road surface has not received attention 
> from mappers. A default assumption that any road without a surface tag 
> is actually paved is most likely wrong.
>
> Efficiency of mapping. Even if one has no interest in adding tags to 
> paved roads, the most efficient way to refine surface tags is to 
> interrogate untagged roads and tag them (e.g. by using an overpass 
> query that distinguishes untagged, paved and unpaved roads, and 
> variants thereof. Untagged roads can be inspected and tagged as 
> appropriate.) However, if mappers are advised to not tag paved roads, 
> then every paved road that is untagged needs to be re-examined each 
> time this is attempted. This wastes a lot of effort.
>
> Some apps — especially routing and cycling apps (e.g. Osmand and 
> Komoot) — allow users to request paved or unpaved routes. Regardless 
> of the (unknown) assumptions that routers make about road surfaces 
> when creating routes, apps like Osmand present the data back to users. 
> The suggested route may be X% paved, Y% unpaved and Z% unknown. In 
> many regions, Unknown is the largest category. This doesn’t inspire 
> confidence in the route or underlying data.
>
> Some assumptions about road surfaces can obviously be made. For 
> example, a primary road is more likely to be paved than an 
> unclassified road. However, most roads in rural areas are tertiary or 
> unclassified. Some are paved, many not; the ratio varies unpredictably 
> across regions and it is impossible to predict which roads are paved 
> unless they are tagged.
>
> Perhaps not surprisingly, the OSM wiki on key:surface gives 
> conflicting advice, beginning with the (European?) position that 
> “there is normally an assumption that the surface is surface=paved 
> unless otherwise stated” and later adding an (American?) view that, 
> “There are no default values for surface, it is generally considered 
> as OK and desirable to tag it explicitly for all roads.” The latter 
> approach seems most appropriate in regional Australia.
>
> Adding surface tags to both paved and unpaved ways is the most 
> efficient method to: (1) allow data users to accurately predict road 
> conditions (this benefits users) and (2) improve the rate at which 
> unpaved roads can be reliably distinguished from paved roads (this 
> helps future mappers). They may be redundant on motorways, trunk and 
> primary roads, but these make up a tiny proportion of roads in 
> regional Australia and can all be coded with a minimum of effort.
>
> Advising mappers to not add a meaningful tag would appear to be 
> counter to the goals of accurate tagging. Can we change our advice to 
> encourage mappers to add a surface tag wherever possible?
>
> Thanks for your time, I'm keen to hear your thoughts. Best wishes, Ian
>
>
> _______________________________________________
> 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/20210131/b944d408/attachment-0001.htm>