How we use highway=track in Thailand

So you still haven’t clarified exactly what you meant here…

I will presume that you meant a way that joins a settlement to the road network can be tagged as Unclassified (or higher) unless it is only passable by an offroad vehicle, in which case it will be tagged as Track. As this is something that can only be accurately ascertained by actual surveying, armchair mappers can tag such a way as they see fit, unless the way has already been surveyed (source marked as GPS or similar) in which case they shall not change it from a Track if so tagged. And, if after surveying, a way is found by an OSM editor to be tagged as Unclassified when it should be tagged as Track (offroad vehicle required) then the tag should be changed to Track.

OK everyone?

You keep trying to say that these tracks into villages are being tagged as Track because of a want to ‘tag for the renderer’. That is not correct - and, unless you want to actually get out there and survey these ways, you are in no position to judge. You, seem to think it is acceptable to ‘tag for routing’. That is no better then tagging for the renderer. After surveying, a way should simply be tagged for what it is - ie. it’s physical attributes considered along with its function. Otherwise there are falsification and safety issues, etc.

Note, whatever happens with renderer support is irrelevant if the way has not physically changed.

Due to bad weather, I haven’t got round to visiting the access roads for the hamlets I said I would take pictures of - but I did go past the same village today and took a few more pics after the recent rain… Though, unfortunately, I was too busy keeping upright to stop on the steepest and most slippery parts.

And Dave, know you have a bike though don’t know what - but if you’re OK for riding offroad, and you want to come for a bit of a tour, let me know.

@Chris - I have a Honda CB500X but I’m strictly a road rider. Those slippery tracks you show me me shudder. Enjoy!

I still try to find a common understanding regarding the tagging of highway=track as the title says.

So let’s leave aside the roads of a bad quality like those shown by Chris.

  1. Can we agree, that in general roads suitable for a city car which link villages to the road network are at least unclassified and we add physical tags to describe surface/lanes? The physical tags only being added if checked by survey or recognizable from imagery without doubt. The tagging is based on the function of a road.

This seems to be something we all can agree, right?

  1. We seem to agree as well that a road without any special function being unpaved and in a bad condition can be very well be a highway=track. Could be roads from villages leading to farmland, forest or orchards. Typically roads not used for through-traffic or to reach a village.

This still leaves open the question regarding roads with a function and bad surface but I want to postpone that discussion to move forward to some agreement.

Leaving away that dirt road in the north, how about a highway with surface=concrete, lanes=2 going through a town, connecting a hospital, local houses, public buildings and businesses to the main network?

The following is a Streetview of that location.

streetview example

That road is currently tagged as a track. I don’t want to blame any specific user, so I won’t mention who did the tagging.

The purpose is still to figure out if we consider a highway=track appropriate for that type.

I might have tagged the road above as either highway=residential (as it is the road to a lot of houses) or maybe a highway=service if it is more catering for isolated businesses. Main point is whether we consider it highway=track.

Not a track - I’d say Service.

Re. the discussion you want to postpone - that already seems decided by a majority. What is there left to discuss?

Function should decide a way type as long as physical attributes fit. If not then a way type that does fit those attributes has to be selected.

@crsCR:
So we agree on the tagging of 1) and 2), do we?

I don’t want to go forward with the discussion of what I postponed. I understand that you have a different opinion on this. We even disagree on the level of “majority” which supports the one or the other position. But as this is leading nowhere I see no point in going on with this for another round.

Let’s not discuss to the point who is “right” but focus on what is the common understanding on tagging. I will draft common understanding to the wiki page.

Would you mind uploading some of your pictures (the 2nd and 3rd) in post #42 to wikimedia commons or the osm wiki? They could be useful in illustrating tagging.

Re. 1 & 2 - yes. After surveying, Way type should be checked and Surface should be tagged and possibly Smoothness. Not Lane as a necessity - too irregular on country roads and not so useful…

Re. pics - you can upload if you like.

Not one person has agreed with you. I call that a majority - and a common understanding. The discussion is not postponed, it is decided. Just you choose to ignore that. Does anyone else not see an agreement? You also ignore all arguments, evidence and opinion put against you - so this has not been much of a discussion…

I have answered your questions so maybe you would be so courteous as to answer one of mine. I will ask you again, for the 3rd time :

A village where access is only by path - would you tag that way as an Unclassified road or a Path? Function always first?

Great that we could find an agreement on the undisputed tagging cases.
As said before I will leave out the gray areas.

Rereading the thread I find a lot of other mappers consider such “gray” areas difficult to judge as it is more the exception than the norm. Not sort of a “majority” in a poll. As said before I understand the desire to support consumers unaware of surface tags.
So I see a majority agreeing to go forward with highway=unclassified for roads connecting villages. If you mean by “disagree” the point to tag your specific road example as a track, then yes: There is a majority tagging this as a track.
As there is always an exception to a rule I suggested above a wording to take this into account.

Tom said he tends to tag bad roads as tracks to prevent the router from using them
Bernhard suggests to use the function but might fall back for the renerer to catch consumers unavare of surface.
Paul sees a problem in defining when something leads to a village or just a single house
Russ agrees in principle but sees problems with unpaved/bad roads
Alaska agrees in principle but is undecided on the specific special case

So by leaving the disputed/uncertain/maybe an exception/… edge cases away we can make progress into defining a more general tagging scheme for Thailand.
It will match for 98% of all cases. You might (or certainly will) find cases where that tagging scheme just is not reasonable to apply. But neither the world is perfect nor OSM. So having a tagging scheme fitting for 98% is a great progress.

With us agreeing on 1) and 2) we are moving a bit step forward. I acknowledge your road might belong into that exception group.

A path is something not used by motor vehicles. So it is not a road (and as such neither unclassified nor track).
If there is only a path leading to a village then I believe it has to be tagged as highway=path.
For use by non motorized vehicles there might be classifications by functions. bicycle lanes or pedestrian areas (which you tag as such even if cars might be allowed on an exceptional basis). Others might be more experienced in tagging for non-motorized and suggest different tagging.

If you are talking about remote villages and the only access is by path then it is a path. If that “path” is also by offroad trucks to get to the village then we come back to the previous discussions about edge cases. Then it would need to be discussed whether unclassified and physical tags or a track.

I agree completely with the tagging of ways connecting villages as unclassified until such time as surface and other characteristics can be added (case 1). No argument about case 2 either. As for the example streetview, ways like this tagged as tracks are, in my opinion, completely incorrect. I change those whenever I come across them, either to residential or unclassified, depending on the particular situation. I try to add a lanes tag and a surface tag as well.

Because I don’t drive off road, I seldom encounter the borderline cases as shown in Chris’ post. I will probably never find myself traveling to a some village on something Chris might tag as a “path” either so I will leave the mapping of those to the rest of you.

Path is something that can be used by non-motorised - I checked: http://forum.openstreetmap.org/viewtopic.php?pid=572409#p572409 The OSM Path Wiki really needs correcting - it was either written with regard to a Western, developed countries perspective of a path - something that, typically, it is illegal to use a motor vehicle on, or a way physically too narrow for a motor vehicle (motorbikes were not considered when writing).

So, then we agree: Function should decide a way type as long as physical attributes fit. If not then a way type that does fit those attributes has to be selected.

Yes, I think you know I was referring to this example of where the only access to a settlement is by a way only passable by an offroad vehicle. I count 5 saying tag it Track with you saying Unclassified. (Not sure on Paul.) You can call it a 2% exception if you like, as long as it’s worded into the rule.

And so, it also seems you would agree with this:

A way that joins a settlement to the road network can be tagged as Unclassified (or other) unless it is only passable by an offroad vehicle, in which case it will be tagged as Track (if wide enough for a 2 track vehicle, Path if too narrow). As this is something that can only be accurately ascertained by actual surveying, armchair mappers can tag such a way as they see fit, unless the way has already been surveyed (source marked as GPS or similar) in which case they shall not change it from a Track if so tagged. And, if after surveying, a way is found by an OSM editor to be tagged as Unclassified when it should be tagged as Track (offroad vehicle required) then the tag should be changed to Track.

I that’s OK with you then, fine - we have an agreement. If you would like to word a rule differently, please show.

Note, Stephan - if you just include 1 & 2 to the Wiki - without this further statement (or similar) - I do not agree with 2.

‘Typically roads not used for through-traffic or to reach a village.’

In Thai countryside, this is not uncommon - where ways to and through are only passable by offroad vehicle. As Thailand is still largely rural/agricultural with many National Parks, forests, etc, this is a large % of total area. Possibly, as you and others do not venture into the countryside much, you lack awareness of this - but that is how it is. The 98% you stated is just pure guesswork on your behalf…

If your plan is just to put 1&2 to the Wiki - and forget about the rest… Think again.

And some more input form the Oz guys:

‘The Australian tagging guidelines do not suggest unclassified for rough tracks. I would use unclassified only for maintained (made) roads. I think you have a track.’

So, for me, only if a way is passable by a normal road vehicle (a non offroad vehicle) - and preferably at least graded - will I class it as a road (unclassified, service, residential).

Well this was a most interesting thread… !

I’m an Brit expat living in the Chonburi area. I ride a Triumph Tiger 800 and log my journeys and also do some armchair mapping over the last year or so that I’ve been using OSM, particularly making corrections that stop me from planning routes (Khao Yai Park north gate wouldn’t let me in - was tagged as a toll) and I map areas from satellite views that i plan to travel to and may have minor roads and features not yet mapped.

I’m certainly not as experienced as some of you guys so i guess i see this whole topic as a something of a layman user.

I use OSM on Basecamp and my Zumo 390 and when route planning i normally select my routing preferences to not select tracks. By doing this i trust that my routing will still select interesting but passable unpaved roads - i’m no off-roader!

I agree with the advice given above and will follow this when mapping.

Many thanks

Rob

Hi Robby, and welcome…
Nice bike, by the way.

We do have relatively small OSM community in Thailand, and new contributors are always welcome. We were all beginners once too, and I think I hold the dubious claim of preventing routing along Hwy 12 through the Namnao NP, simply because I added a lift gate at the agricultural inspection points on the way :slight_smile:
Fixed now of course, just buy adding access=yes to the tag but we live and learn.

I would just add that with the exception of correcting obvious mistakes, I think we do, out of courtesy, ask other users before any significant changes to data others have added, and would be great if u follow suit.
And if you note any issues like the one you found, feel free to publish so as we can learn from the errors.

I know that NP roads have been the source of some discussion as with a toll gate at entrance & exit, by definition, the road is toll … but not to everyone ! For now, I think we are just tagging the gates and not the roads.
I presume your GPS was not set to exclude toll roads, as that would have stopped routing you that way, for sure ?

Best, Russ.

Hi Russ

Thanks for the reply. I take your point about informing the user or others before making a change.

Is there a way to determine who mapped a particular feature and how to contact them please - or do i just raise the point on this forum?

BR

Rob

Dear All

Following the main point of this thread i note to the north east end of the Srinagarin Dam Reservoir there are a number of roadways near to the reservoir that on there OSM edit view are listed as Unmaintained Track Road, with no other information listed in the description fields.

On Basecamp these are described as Unpaved Roads.

From the Bing Satellite image i can see that these roadways clearly lead to houses and hamlets.

Using your logic above would they be better described as Minor / Unclassified Roads?

BUT

I note on OSM edit view if you head further south there are some Minor / Unclassified Roads are shown as a line of small light brown dots and on Basecamp these are still called Unpaved Roads.

A little nearer to the SPH 4031, closer to Muang Thao, sections of the Minor / Unclassified Road are shown as a straight light brown line and on Basecamp are called a Road.

This therefore affects my route planning on Basecamp as i ‘turn off’ unpaved roads in preferences, for the reasons listed in the thread above, as i’m no off-road hero…

What do you guys think?

BR

Rob

Hi Rob,

This is one of those issues that really cant be explained in an email… takes too long.
We have some OSM guru’s that can explain better than me, but in principle, OSM is a set of data (based on a node with lat & long, and a tag that imparts information). All other programs, such as Basecamp, & the OSM rendering of Thailand in its many forms, take that data and display it any way they like.
A highway=track can be called many things depending on the program… Garmin calls it a Trail, and it can also be an unpaved road or a track etc etc.
And depending on the website that you take your Garmin readable OSM map from, that will vary too.

And then to open the other can of worms … many senior advocates on OSM state roads should be tagged as for their use, so a road through an orchard to a house should be residential. Tracks between two villages should be unpaved minor roads.

Others think that it should be a track, more based on its actual physical condition.
While a few might even consider it a driveway (or service road).
Consistency will always be an issue for crowd sourced data, which I agree, for your use, makes planning a nightmare.

And on a side note, I dont like the word “unmaintained” as its not used much, and all roads are maintained by someone, even if just the local village.

And finally, many roads have been armchair mapped in Thailand, and the view here is we dont add a surface tag unless we are sure we know. ie, fine to add “paved” if you can see the white centre lines in Bing, but if you cant identify the surface clearly, leave it out. I would say that up to a point in Thailand, the surface tag is fairly accurate, and conservative at the least. A track would be unpaved (dirt, gravel etc) but just might have become paved at some point. A minor or residential road could be anything in rural areas.

And be aware that if a road has potholes and someone has tagged as smoothness=bad, then Garmin OSM maps will render as unpaved, even though they might be quite ridable with just small areas of potholes.

I cud write for ever, but in summary, I dont think you can rely on route planning based on your surface requirements just yet, unless you are prepared for everything the road might throw at you. I wud say we tend to use highway=path for single track stuff, so keep clear of them as they are probably too tight for the Tiger !

Lastly, depending on your editor, you can easily view the history of nodes & ways, the shortcut key “H” in Potlatch2 does it for me. Then scroll down the changes and you can see which user changed what.

If you are ever in Chiang Mai, I’ll be happy to have a beer and chat more, but believe me, it can get quite overwhelming !

Best, Russ.

Rob,
I took a look at one of the areas you talked about … http://osm.org/go/4TmJKwUR-?m=

Yes, this was plotted by a well respected mapper about 4 years ago. And his base data came from a ride report & GPS track at the time. There is a clear Bing image available now, which was probably also available then.

It appears the middle section tagged as minor road, does have some sort of paved surface, and it changes to track at either end where the roads use is more of an unpaved agricultural nature.

I would probably have tagged the dominant access “track” (or both) as unpaved minor roads, based on our current thinking and influenced by the roads clear use to connect the village to the highway… this doesnt mean Beddist did wrong, but goes to show how we can all have different interpretations of the mapping guidelines.

It also goes to show what this original thread is about, and reinforces my statement about surface being too unpredictable to plan using the routing engine in Basecamp.

So while u are out, any road, especially one that is unpaved, that does not have a surface tag … then please put it in for us. :slight_smile: