Proposing surface=ballast for surfaces covered with large loose angular crushed rock where dominant stones are 63 mm or above (roughly tennis-ball size). It is an opt-in precision value within surface=gravel, mirroring surface=fine_gravel at the other end of the size scale.
The material is rare globally (mainly forest and logging tracks in Central/Eastern Europe, France, UK) but has qualitatively different routing behaviour: difficult for cyclists and motorcyclists, risky for horses and standard cars. surface=gravel gives no signal that the surface is at this extreme. 23 organic uses of surface=ballast and 24 of surface=track_ballast already exist without documentation.
Photos of ballast-surfaced tracks for Wikimedia Commons?
Is “ballast” the right name? Railway track ballast (EN 13450: 22–63 mm) is actually smaller than the 63 mm threshold. Alternatives cobble_gravel and riprap were both rejected (see proposal).
In the context of a road, ballast would be considered at technical term, that regular people don’t know. A more well-known Danish term would maybe be “skærver” which basically means “crushed rocks”. Railroad tracks usually have ballast made of crushed granite rocks i Denmark.
The term “ballast” seems to designate the purpose of the material (stabilising the road/railway just like ballast on ships are used to stabilise ships) rather than the actual material. But maybe this is fine? surface= != material= after all.
I think i’d prefer surface=crushed_rocks, but maybe ballast means exactly that in other languages than Danish. Agree that surface=gravel alone is not sufficient to describe the road in the picture. smoothness=very_bad?
I’ve always wondered how to tag surfaces with gravel sized larger than typical track ballast, so your proposal makes sense and fills a gap. If we have surface=fine_gravel and surface=gravel, then why not have surface=coarse_gravel? I’m also in favour of creating new surface values for surfaces that have particular usability difficulties for certain vehicles that smoothness doesn’t catch well, and this is one of them. This surface is particularly difficult to ride or walk on (2-wheelers, horses, pedestrians) because the stones shift under load. Cars with broad tyres have no problems with this behaviour: if the stones are well spread, normal cars should be able to drive on it with reduced speed (smoothness=bad), only when they are unevenly spread so some stick out and can damage the undercarriage of a normal car, SUV, etc, smoothness=very_bad or worse.
The wiki for smoothness=very_bad (high_clearance) and worse is now described in terms of clearance only.
Even when it is quite literally track ballast, a value emphasizing that would be very handy. I’ve resorted to surface=gravelgravel=track_ballast for this short gap in a sidewalk crossing railroad tracks, because otherwise another mapper will eventually change it to surface=fine_gravel in a less careful edit. After all, the vast majority of surface=gravel in this country is finer gravel because that’s what we call “gravel” in the local dialect, so that replacement is usually a safe bet. Even at this granularity, track ballast already isn’t suitable for wheelchairs, so I wanted to preserve that distinction.
Track ballast is mentioned in the wiki as the upper limit in size for surface=gravel, so for gravel of larger size we should not use (track) ballast in the name.
Yes, all I mean is that there are multiple things that need to be disambiguated from gravel these days. You’re focused on a size that falls outside of the definition, while I’m describing a size that falls inside the definition but is prone to confusion anyways. Some mappers are using gravel=* as a workaround.
From the photo I cannot tell size. All I know is: The photo in top post was introduced into the documentation as the one and true meaning of gravel.
There is a German term “Schroppen” for material where the smallest parts over 63mm. A search on the web finds some producers. They consider that size beyond “Schotter” (a.k.a. gravel). It is not marketed for road surfacing. Rounded or crushed also not specified.
Thank you, this is directly useful for Q1. I’ll add Schroppen to the proposal as a German producer/trade term for crushed rock above 63 mm, distinguishing it from Schotter.
On the photo size: you’re right that the exact fragment size isn’t readable from the image. The lead image point in the proposal rests on the gallery caption on the current surface=gravel page, which explicitly calls that end of the range “big chunks of ballast”, that is the community’s own word, not my interpretation.
Exactly, and that disagreement is itself an argument for the proposal . Both smoothness=very_bad and smoothness=horrible are documented around vehicle clearance, not around lateral wheel deflection and fall risk. Neither carries the material signal. I’ve added this point to the proposal FAQ.
Agreed. Standard railway track ballast (EN 13450: 22–63 mm) sits inside the surface=gravel range, while this proposal targets ≥ 63 mm. The overlap in terminology is a real problem and the name should change.
This proposal is not about coarse_gravel. In ISO 14688-1, coarse gravel (cGr) is 20–63 mm — the same range as EN track ballast. That gap (20–63 mm, no dedicated tag) is real and worth a separate proposal. This proposal targets the ≥ 63 mm range, which ISO calls cobble: a qualitative step above gravel, not a gradation within it.
I had previously referenced a German scale that supposedly extended gravel to 200 mm, I was wrong about that. Checking the Wikipedia grain size article, all major classification systems (ISO 14688-1, Wentworth, German DIN) consistently place cobble at ≥ 63 mm outside the gravel class. That actually strengthens the case for a hard line at 63 mm, and surface=gravel should not cover material above that threshold when the mapper knows the size.
“Skærver” points in the right direction. crushed_rock is my preferred rename: “crushed” signals the angular manufactured character, “rock” signals the size, distinguishing it from crushed_stone which spans all sizes (See Crushed stone).
cobble_rock was considered but per the Wikipedia cobble article, cobble is a size class only (64–256 mm) with no angularity requirement, so it would include large rounded water-worn stones too, broader than this proposal.
That raises an open question: do we need a separate tag for large rounded cobble-size surfaces? I have direct experience (Indonesia) with deteriorated roads where the concrete has worn away leaving the rounded pebble base exposed, horrible to ride on. Should that live under a shared cobble_rock, or is a separate proposal needed (parallel to surface=pebblestone at the smaller end)?
crushed_rock stays scoped to angular crushed material. A wiki note distinguishing it from surface=rock (natural bedrock) would be needed. Thoughts?
Update: this table answers part of my question, pebbles are within the 4–64 mm range, so any rock fragment larger than 64 mm is considered a cobble, or a boulder if it exceeds 256 mm.
I went through a dozen Finnish companies selling crushed rock, and the relevant available size ranges were 0-56, 0-100, 32-60/64, 90/100-150, 150-200 etc. with some variation between vendors.
I would not spec this to be overly reliant on standard definitions, or to rely on mappers measuring rocks at millimeter precision.
Agreed. The 63 mm figure is a scale reference, not a measurement requirement. The field guide in the proposal is a tennis ball (65-68 mm): dominant stones roughly that size or larger.
The vendor data is actually useful here, even if crushed rock is sold for many purposes beyond roads (drainage, embankment, landscaping). There is a natural break between the 0-56, 32-60/64 mm products and the 90/100-150 mm ones, which corresponds roughly to where routing behaviour changes: 32-64 mm is rough but rideable, 90+ mm is the material that deflects wheels. A mapper encountering 90-150 mm crushed rock on a forest track does not need a ruler.
In OSM, we can set our own standards, we don’t have to follow ISO or other standards. We just need a name for the tag value that works for us, i.e. causes as little confusion among mappers as possible. If we would call the size above 63 mm cobble like in the standards, it could cause confusion with surface=unhewn_cobblestone (anyway it would be good to explain in the wiki that the surface we’re discussing here is made by dumping stones on the ground and spreading them rouhgly where they are wanted (with a bulldozer for instance), while surface=unhewn_cobblestone is made by placing the stones individually in a single layer with the aim to make a more or less closed pavement).
I don’t think so: their behaviour when riding on them is similar to riding on crushed rock of similar size, so we can include them in this surface tag value.
I protest! You always think of cars? For a cyclist, this makes hell of a difference. Said that, I also think, surface=cobbles good enough for OSM. How many mappings are expected from that? Personally, I do not know a single location.
No, I’m first of all a cyclist and I know it’s a pain to ride on stones of such size. What I’m saying is that I don’t think it matters much if the stones are angular or round; rounded stones may be a bit more difficult because they move more easily under the pressure of the wheel and they are more slippery.
. Both 