But now we have smoothness as the indicator of road qualityâŚ
I think the essential service that a map should offer is that it helps the user decide how to go from A to B in the most efficient way. For this, the user needs to decide which roads to take, and the map should give him information to help in that decision. In order of priority, I think these should be:
Usability of the road: can the user use this road with the vehicle he intends to use, or would it be preferable to take another better road that is longer? smoothness serves this need for route planning.
Recognition of the road: once the user arrives at the road junction, he has to be able to recognise the road he planned to take by comparing it to information that the map is giving. surface serves this need by describing the appearance of the road surface. Thanks to the availability of GPS positioning, this aspect has become less essential.
Risk evaluation that the chosen road may be less usable than since it was last surveyed. tracktype serves this need (as it is now, in my interpretation of the wiki), by giving some information on how the surface of the track might have changed because of wet weather and use by heavy vehicles: how likely is it that because of wet weather, the surface has become so soft that the userâs vehicle might get stuck, or that it has been deformed by heavy vehicle traffic so that ruts have formed and the ground clearance of the userâs vehicle will not be enough to pass without damage? It gives an impression of seasonal variations in smoothness.
The photo in this post shows an example of a track that has a mixture of many surfaces, with none of them dominating. So instead of tagging it as surface=rock;gravel;fine_gravel;sand;earth, I think it is best to tag it as surface=ground. The
+0.5, this is what I think is done most, albeit it is hard to evaluate if you are not aquainted with the area (it can mean earth in some areas and rock in others), so I am not sure if it is recommendable tagging. Clearly it is used orders or magnitude more: https://taginfo.openstreetmap.org/tags/surface=ground (2,8 million)
Thanks, I see where youâre coming from. Iâd still probably mark that as surface=rock as it dominates the potential use of path (even if perhaps it does not have the supermajority percentage or surfaces). i.e. if someone were considering whether to use that path, they would likely not worry whether there is some sand and earth, but about those rocksâŚ
There seem to be few hundred uses of surface=bare_rock. Itâs not documented on the wiki, though.
Is that typo or not? If not: no, I would not mark surface of highway=path with natural=bare_rock, as that one is documented to applying only to areas, not ways.
If it was a typo, and you meant surface=bare_rock, I would consider it once it is documented on the wiki, and its description matches what I see. But as that one is currently undocumented, I would currently avoid it and choose best matching documented alternative (likely surface=rock).
I concur, the amount of loose material is just too much. A grade2 has to have at least a flat/crowned surface.
Maybe you have local knowledge, but the photo does not indicate bedrock just below the surface, to the opposite, to me it looks like quite some gravel all below.
@yvecai@Hungerburg what do you base your judgement on? Is (was) there a consensus that driveability with family cars, the quality of the track, the amount of loose material and the shape of the surface should be taken into account when evaluating track type? What is the documentation for this? The wiki starts that âTracktype is a measure of how well-maintained a track or other road isâ but then mentions quality only once at the end of the text, which in my opinion is all about firmness.
(I took the photo in this thread, so I have local knowledge)
Another factor that plays a role in my evaluation of what to tag for surface is that Iâm a lazy mapper who doesnât like micromapping (itâs often not useful) but prefers to map the big picture with usefulness in mind (Iâm an engineerâŚ). Indeed the largest fraction of the materials on the photo are large stones, but just around the corner it may be gravel. Instead of mapping every 10 m of that track with the tag for the largest fraction of material on it, I prefer to tag the big picture and use ground, which is a more general tag that I prefer to use when a more detailed tag that changes frequently over the length of the way is not useful.
surface=rock is an ambiguous tag at the moment, because it seems to cover 2 cases:
âBig pieces of rock used to improve path qualityâ and
âexposed bare rockâ
Maybe it would be best to deprecate surface=rock and replace it with surface=bare_rock for ways that pass through areas of natural=bare_rock (assuming that describes their surface well) and with surface=unhewn_stone which is like surface=unhewn_cobblestone (i.e. it is a kind of paving âused to improve path qualityâ) but with stones that are not rounded. Maybe an additional tag value is needed for surfaces mostly consisting of stones larger than track ballast (given as a limit for gravel), for instance for paths that go across areas tagged as natural=scree.
In my early days in OSM, a track large enough for a car was highway=track. Given the various quality that can be expected for them, grading them from 1 to 5 for 4 wheels vehicles made sense.
Now I may be wrong from the beginning, but counting and sizing gravel on a wiki picture or pinpointing every word in the wiki doesnât make sense either.
Thatâs a good example of many roads in the western US. Tracktype is about firmness, & Iâve quit using tracktype for these types of roads since itâs very firm, but very rough. tracktype is not very useful for these types of roads since it doesnât capture the nastiness of it. Smoothness is better as defined in the wiki. Iâd tag it as surface=ground, smoothness=horrible or very_bad. Itâs tough to determine from the pic how deep that erosion is, that would determine if its very_bad (high clearance), or horrible (4wd)
There was another topic recently dealing with surfaces made up of loose stone and I proposed to use surface=gravel for those as main tag and specify more details by subtags:
gravel = a surface more or less completey created by or covered with loose gravel of different size from 2 up to some 80 mm, either intentionally build like that (cheaper than compacted) or emerged out of deterioration of a former compacted surface due to missing service. Rough and shaking bike ride. Gravel can be naturally rounded material (natural gravel) or sharp edged (crushed stone). For more detailed tagging, use the appropriate subtags.
gravel=natural_gravel (if it is definitely rounded material)
gravel=lag_gravel or cobblestone (if it is rounded material with sizes bigger then some 60 mm)
gravel=crushed_rock (if it is definitely sharp edged material)
gravel=riprap (if it is sharp edged material with sizes bigger then some 60 mm)
Note that riprap does not specify a size range of the stones, it is used for very rough material generally. In the wiki the term âtrack ballastâ is mentioned for rough size crushed rock. This material also does not have a general size specification but normally does not exceed 60 mm.
Even without local knowledge one can see that this is a track with a firm structure which does not necessarily require a base of bedrock. One can see that the tracks in both sample pics have developed by erosion from a stony ground - soft parts, sand and small stones have been washed away and the bigger stones have accumulated. I would place any bet you can drive a heavy truck there even under wet weather conditions. To me this remains a track of grade 2 definitely.
That is subject to surface/smoothness, not track type.
So I went to unroll the history of the tracktype key, starting from the original proposal: Reminds me of how roads in openstreetmap are categorized - by importance in the network, primary, secondary, etc.
In my local area, nearly two thirds of all roads are tracks. If that was similar in the area of the proposal author, some desire to distinguish may look plausible.
Later in history it became a measure for how well-maintained a track or minor road was perceived and later still, firmness was added, perhaps in a bet to have a so-called verifiable aspect in the mix.
Are we there yet to do away with maintainedness and make tracktype only map firmness? Meanwhile, I still give the track in the photo above a 4 on maintainedness, as long as this is still the documented meaning of the key.
PS: The local administration has OGD on tracks. They do not grade; Properties they collect: paved/unpaved (gravel/ground), suitability for tractor/lorry/lorry+trailer, minimal width, steepness, crowning, and how water gets removed.
From comparing known tracks with the one shown above, Iâd say, its unpaved ground, tractor only, no water removal. Certainly, some heroes will drive their truck up there, they may even have an extra item for billing.
Now that the discussion in this thread has stopped, I think it is time to draw some conclusions. I think 2 issues have appeared here:
There is disagreement of interpretation and thereby inconsistent mapping of the tracktype key. This seems to have a historic reason: the key was originally proposed to be a measure of quality of tracks, but later changed to a measure of the ratio of hard and soft materials in the surface and developed towards being a measure of track surface firmness. The wiki describing this has developed in this direction, but still contains references to the original idea. The smoothness key was developed later as another measure of surface quality.
A clear tag is missing to describe way surfaces consisting mostly of large stones of size between gravel (which according to the wiki is up to the size of track ballast) and bedrock, with the tag rock having 2 meanings according to the wiki.
I have some ideas about what to do about this, which I will post as a new topic in the coming days. Looking forward to more feedback here!
Seen big, inmovable rocks being placed on ways to block particular traffic. The rock(s) could be mapped with a tag maxwidth:physical to indicate what can pass, MTB, foot, trike. Add those and within 24 hours, GraphHopper will tell in the OSM routing function if bike, foot, motorcar can pass the obstackle points, albeit GH is a bit tight, as the other router simulators within OSM are more âunderstandingâ.
Of course thereâs the supplemental smoothness tag, of several, but how that/those weighs in with routing Iâve never dug in to understand.
| 
There was