OpenStreetMap

Smoothness and MTB_Scale tags on paths

Posted by John Stanworth on 12 September 2021 in English (English).

My mapping of the paths used by cyclists has evolved over the years. If there are other keen cyclist mappers out there I’d be grateful if you would have a look at this post and say if you feel it is reasonable. My cycling friends and family use apps like Strava and Komoot to log their rides and increasingly to suggest routes. These same friends often have more than one type of bike - maybe a road bike, a gravel bike, a trail bike and sometimes a specialised downhill rig. Some of the apps will suggest different routes depending on which bike you specify. I’m not sure how these apps judge which paths are suitable when they are creating, say, a ’gravel’ ride but I would think the surface, smoothness and mtb_scale tags would be important. I now try to give all the cycleable paths that I map a smoothness tag and if they aren’t asphalt or concrete I give them an mtb_scale tag as well. Studying the descriptors for the different smoothness values makes me think : If the trail was ‘excellent’, ‘good’ or ‘intermediate ’ I would be happy on my road bike. If it was ‘bad’ I would prefer a gravel bike. If it was ‘very bad’ I would prefer a trail bike. If it was ‘horrible’ I would still use a trail bike but I would be up on the pedals and concentrating! If it was ‘very horrible’ I might ride it, downhill only, probably on a full sus rig. Studying the descriptors for the mtb_scale values, both on the OSM wiki and elsewhere, I think I would use a gravel bike if it was 0, a trail bike if it was 1, a trail bike but up on the pedals if it was 2 and downhill only on a full sus bike if if it was 3. So although they are measuring slightly different things there is a sort of equivalence between the smoothness values and the mtb_scale values for the non-asphalt trails. And I am now using this when mapping. For example if I look at a path and think it’s too rough for my gravel bike but it would be straightforward on a trail bike I will map it as Smoothness=Very Bad and MTB_Scale=1. Does that seem reasonable? I’d be grateful for any thoughts from other mapping cyclists.

Comment from ChristianA on 13 September 2021 at 07:55

I am not sure have the different apps (Strava, Komoot, etc) calculate routes and which OSM-attributes they use. I usually add mtb:scale to the paths I ride, but it is quite difficult to be consistent when deciding on the mtb:scale attribute.

Have you looked at the OSM wiki for the description of smoothness and mtb:scale?

https://wiki.openstreetmap.org/wiki/Key:smoothness https://wiki.openstreetmap.org/wiki/Key:mtb:scale

Anyway, best of luck!

Comment from John Stanworth on 13 September 2021 at 08:43

Thanks Christiana :)

Comment from Richard on 13 September 2021 at 12:04

smoothness is really not a great tag because it’s so subjective and its application to different surfaces is unclear. For example, there’s no clarity or agreement on how it should be applied to surfaces such as crushed limestone - common on cyclable canal towpaths.

I would suggest concentrating on surface, tracktype, and mtb_scale tags, which are less ambiguous.

Richard cycle.travel

Comment from John Stanworth on 13 September 2021 at 21:18

Thanks for that Richard. It’s great to hear from someone who is writing software to use our mapping! I’ve just been looking at your cycle.travel site. It’s great! The routing works really well here in South Yorkshire. I’ve sent links to my family.
I sort of agree about smoothness. I do feel that it’s a great idea for a tag but having words for the values instead of numbers means that mappers who haven’t studied the wiki are, with the best intentions, putting very wrong values on paths. The crushed limestone you mention (‘inch to dust’ is what our local builders merchant calls it :) might be smoothness=excellent to a mountain biker but smoothness =horrible to a skateboarder.
You recommend including the tracktype tag. I do put it on tracks but have doubted it’s value for cyclists - a ‘solid’ surface might be smooth but it might be very very rough. But perhaps I’m overthinking it. If routing software like yours is using the tracktype values do you feel that it would be useful for me to put that tag on paths and bridleways as well? Cheers John

Comment from Supaplex030 on 19 September 2021 at 12:17

There are growing approaches to establish more objective criteria for smoothness tagging. If these are taken into account, it is a useful attribute with clear benefits. A detailed overview with pictures and semantic descriptions in different categories and for different vehicles can be found here: (in German language) – (translated and supplemented by an English-speaking mapper) – (see also this discussion with comments from other mappers).

In some nuances it is certainly debatable, but it is already very suitable for distinguishing “good” from “bad” surfaces. I don’t use Komoot, but I know from conversations with the developers that they consider “smoothness” and charge worse surfaces with routing panalties.

Comment from John Stanworth on 19 September 2021 at 12:41

Thanks Supaplex030. Yes I absolutely agree. Smoothness is a really useful tag. But I can understand Richards scepticism - a lot of the smoothness mapping here in Sheffield UK does not fit the information in your very helpful links. I wonder how many new mappers are aware of the wiki information on tags. I certainly wasn’t when I started :) I wonder if you have any thoughts about my ideas on mapping non-asphalt trails with ‘smoothness’ and ‘mtb_scale’ smoothness-bad = mtb_scale-0 smoothness-very bad=mtb_scale-1 smoothness-horrible = mtb_scale-2 Great to hear from you. John

Comment from westnordost on 19 September 2021 at 18:30

Based on the links posted by Supaplex030, work has started in StreetComplete to implement a UI for recording the smoothness information. Feel free to contribute. We are still looking for clear and representative pictures for different smoothness levels of asphalt surfaces, of sett surfaces, of paving stones surfaces and more, see

https://github.com/streetcomplete/StreetComplete/pull/3257

Comment from bradrh on 19 September 2021 at 23:09

HI John, I think you’re on the right track. I’m from Colorado, US, so my geography & road/trail conditions are quite a bit different from yours. So take my opinions with a grain of salt. I’ve been a mtber for a long time, also done some road touring here, Europe, & south east asia.

I disagree with Richard about smoothness, I think you should use it. I think smoothness is a really important tag for roads and bike paths, not so sure about mtb trails. Over here tracktype is pretty worthless. It is mainly about firmness. Most of the trails or roads here are rocky and very firm, but could vary from easy to extreme. Smoothness captures that better. Over there tracktype is probably more relevant.

Most of us over here use trailforks.com, mtbproject.com, or a local map to get an idea of what to expect. I’ve given up on OSM for this. The trail ratings used by those apps, and local maps are a green/blue/black rating similar to what is used on ski areas. Sure it’s subjective, but there is no other way to do it. It works pretty well. I think the green/blue/black rating system is used over there quite a bit too.

mtb:scale is so poorly written that I’ve quit using it. Max gradient for mtb:scale=1 is 40%?? 40% is virtually unrideable downhill or uphill. Others here use mtb:scale & mtb:scale:imba interchangeably. mtb:scale:imba is better, but unfortunately the :imba tag was incorrectly defined to be for bike parks (by OSM, not by IMBA). Someone tried to change it a while back & a couple of us supported that, but were met with illogical resistance.

Comment from John Stanworth on 20 September 2021 at 08:10

Westnordost - You are doing a great job. The extra pictures are already making the smoothness wiki a lot clearer. Thanks. I’ll start taking pictures of trails and see if I can get anything useful.

bradrh - Thanks for that. Trailforks is bigger than mtbproject over here. I think this is probably the future for downhill mtb (and some other sports like climbing) - the specialist app/site has the specialist info overlayed onto OSM which gives the access routes. I will carry on using smoothness and mtb_scale on the gentle trails that I tend to be mapping but will have a look at the IMBA scale as well. PS I had noticed the 40% in the wiki as well. That is very very steep!

Comment from ChrisHodgesUK on 20 September 2021 at 13:37

I agree broadly with your take. I might push things a little further, by taking my tourer down stuff better suited to a gravel bike, or even round the local blue MTB trail, but me being odd is an edge case.

I plan using Komoot, with OSM (mainly cycle mode but flipping to standard view) almost always open when routing in unfamiliar areas, along with other sources of info. Komoot certainly makes good use of OSM track quality data, though how it uses each field is a bit opaque. Of particular interest to me, “touring” mode routes down gravel far more willingly than “road cycling” mode, as well as towpaths etc as you’d expect. It can be rather optimistic where a RoW runs through a farm, so I’ve ended up on slippery lumpy grass on slick tyres more than once Often there’s a lack of data for it to go on, something we can help with.

My MTB (entry-level hardtail) planning tends to be less automated, and some hike-a-bike is to be expected, thus I’m less aware of what Komoot does in that mode.

One stretch I did recently was tracktype=grade2, surface=gravel (NCN 78/Great Glen Way through Clunes Forest) with no “smoothness”. Where they’d been logging “smoothness=very bad” would have been appropriate (just about got through on the tourer without walking but standing on the pedals); adjacent sections just “bad”. Other nearby gravel sections of NCN78 would have to be “intermediate” smoothness per the wiki though honestly you could go flat out on 23mm tyres if not for walkers and dogs, so “good” wouldn’t be wrong.

I’ve updated some sections (and may be able to do a few more before I forget my recent trip) but am wary of making a judgement on mtb:scale so tend to concentrate on surface material, tracktype and smoothness. Plus on long rides there’s a limit to how much surveying I can do - I often rely on remembering numbers from my tripmeter with a single keyword, until I stop anyway to write things down.

Regarding max gradient on mtb:scale: When I have used mtb:scale, I’ve tagged for the most severe features encountered - so plenty of mtb:scale=2 despite not going over 40% steepness. The steepness numbers do seem too high and inconsistent with mtb:scale:uphill; perhaps the description and examples came from an experienced downhiller.

One thing that would be nice to see is an (up-to-date) article on all the various surface type/quality fields for cyclable ways.

Comment from CycleStreets on 20 September 2021 at 16:46

We would agree that smoothness is subjective, and would prefer to see fuller mapping of surface and tracktype.

We would definitely recommend inclusion of these on both paths and bridleways - there is a big difference certainly around the UK between different qualities of bridleway, and far too many don’t have any surface indication tagged.

Martin, CycleStreets

Comment from bradrh on 20 September 2021 at 17:47

I don’t understand why folks think smoothness is subjective but tracktype is not?

Comment from bradrh on 20 September 2021 at 17:53

To be clear, I agree smoothness is subjective, but so is tracktype. Even surface is subjective at times.

Comment from John Stanworth on 20 September 2021 at 20:37

Thanks for that Martin. It’s really good to hear from users of the tags like you and Richard and get an insight into which tags you use. I will try to include tracktype in my future mapping of paths and bridleways because both you and Richard use it in cycle routing. I hesitate to ask but do you feel that some of us are taking the wiki too literally when we use tracktype as a measure of softness alone. Do you software writers use it as a general indication of the trail quality - tracktype=1 means great for wheeled vehicles, tracktype=4 means terrible for wheeled vehicles? It won’t be the first time that I’ve been told that I take the wiki too literally :)

Comment from John Stanworth on 20 September 2021 at 20:48

ChrisHodgesUK - great to hear about your experiences of the quality of mapping of cycle trails in the UK. It’s very variable isn’t it! But I’m as much to blame as anyone - the way I map them seems to be constantly changing! :)

Comment from ChrisHodgesUK on 21 September 2021 at 08:01

It’s always going to be subjective (bradrh) to some extent, partly because these surfaces are variable. There’s a byway near me that has seen some gravel in the past, and is fine for road bikes in dry weather. Even a summer downpour turns bits of it into a mudbath. It’s currently tagged “tracktype=grade3” though grade4 would be more appropriate in winter. I’ve changed “surface=unpaved” to “dirt”, that being the dominant surface from a user point of view (do we have too many surface types?) and set “smoothness=bad” based on the wiki.

When reviewing the smoothness levels I see a discrepancy from the point of view of someone on 2 wheels - “intermediate” includes some pretty awful asphalt, while good unpaved roads are supposed to be “bad” but can be far smoother than damaged paving


Login to leave a comment