I was thinking how to really differentiate those networks and I couldn’t find any wiki entry about bike route networks classification. ICN is rather obvious but lines between NCN, RCN, and LCN are quite blurry. Especially difference between LCN and RCN.
I asked claude ai. What do you think about this?
ICN - International Cycle Network
Value: network=icn
Used for major cross-border cycling routes
Examples: EuroVelo routes that span multiple countries
Typically well-maintained and signed
Often connects multiple national networks
NCN - National Cycle Network
Value: network=ncn
Major routes within a single country
Examples: The National Cycle Network (NCN) in the UK
Usually maintained by national cycling organizations
Connects major cities and regions
RCN - Regional Cycle Network
Value: network=rcn
Medium-distance routes within regions/states/provinces
Connects towns and cities within a region
Often managed by regional authorities
Links between national and local networks
LCN - Local Cycle Network
Value: network=lcn
Short-distance routes within towns/cities
Typically managed by local authorities
Connects neighborhoods, schools, shopping areas
Used primarily for daily commuting and local trips
Well, firstly I think that asking some “AI” is a mistake and is likely to repond with something that sounds nice, but in practice is completely unrelated to actual usage.
In practice, I suspect that there will be significant national differences. To take the UK as an example, what Metal Mickey has suggested would work quite well for RWN classifications, but less so for for RCN, where RCN usage is a bit of a special case. Outside of the UK, you’d need to ask the relevant community and ask them how they classify their routes.
No comment on the merits of the idea itself, but since the “CN” in all the values are cycle network, I think it be more readable to have network=international | national | regional | local
Instead of network=ICN | NCN | RCN | LCN
Thank you for the response. You make a valid point about national differences in route classification. This actually highlights why I believe we need better documentation about these network types.
I’d like to propose creating a dedicated wiki page that would:
Provide general guidelines for network classification (ICN/NCN/RCN/LCN) that could serve as a baseline for mappers worldwide
Include country-specific sections documenting local practices and exceptions
Link to existing country-specific wiki pages and documentation
The goal isn’t to enforce a one-size-fits-all approach, but rather to create some place where new mappers start.
Actually, as I’m writing this - I think instead of creating new page we could enhance existing documentation Cycle routes - OpenStreetMap Wiki
There is a section on another page Relation:route - OpenStreetMap Wiki which gives some overview. It’s nice base but we could use examples and info about country differences.
International routes typically well maintained and signed? I doubt it, but even if so, that is not a criterium. local, regional and national routes are often well signed and maintained as well, and international routes sometimes have no signage of their own, but use national signage for the national sections which are also ncn.
In Nederland, management of routes doesn’t say much about the geographical scope of the routes. E.g., many local and regional routes are managed by the national agency for forest maintenance, or by the province administration, or by a route maintenance bureau for a province or a natural region.
I sympathise with the thought, but the system of network values for routes also includes *hn for horse routes, *wn for walking routes, *in for inline skate routes, *pn for paddling routes, *mn for motorboat routes. I may have forgotten a transport method.
All lower case values, by the way.
The combination of geographical scope and transport method in one tag is not the best idea, but very well established so I would recommend changing it, certainly not just for cycling, since the tagging system of these (mainly recreational) routes is very much alike for all transport methods and geograpical scopes.
If I had to design it anew, I probably would split off the geographical scope in a scope key and the transport method in a transport key. In the other hand, handling a particular recreational route category now requires ony one tag, which is easier than two.
Ich beschäftige mich nicht mit Radrouten - jedoch gibt es das Gleiche mit Wanderrouten:
iwn, nwn, rwn und lwn.
Im Wiki habe ich unter DE:Wandern sogar Längenangaben gefunden.
iwn = International ( > 500 km, Europäischer Fernwanderweg)
nwn = nationaler Fernweg (200…500 km, mehrere Bundesländer/quer durch ein BL)
rwn = Regionalweg (50…200 km, mehrere Landkreise/Kantone)
lwn = Lokaler Weg ( < 50 km, in einem Landkreis, einer oder mehreren Gemeinden)
Das hilft aber nicht so recht weiter. Ich habe bei den Wanderwegen den Eindruck (ich habe das nicht statistisch überprüft), daß die Grenze zwischen lwn und rwn nicht wie im Wiki vorgeschlagen bei 50 km gemacht wird sondern eher bei 20.
Und wo ist die Grenze zwischen rwn und nwn? Ein Wanderweg, der Hessen von Nord nach Süd durchqueert (etwa 400 km)! ist etwas anderes als ein Wanderweg, der das Ganze mit dem Saarland tut. Wie wäre es mit Liechtenstein?
Hessen von Nord nach Süd könnte man als rwn einordnen. Die Strecke ist aber viel viel länger als ein Weg von Feldkirch(A) über Liechtenstein in die Schweiz - letzterer wäre ganz klar iwn…
tl;dr
Diese Einordnung ist äußerst vage und nur als SEHR grobe Orientierung zu gebrauchen…
edit: Kilometerangabe zur Länge von Hessen - muß man ja nicht wissen…
I like the idea of mentioning hints about typical distances. Do we have some statistics about that or could someone with the technical skills create them?
Having been a “major contributor” to *cn (especially network=ncn) in the USA, I can say that network=icn is most definitely appropriate for bicycle routes which cross an international border (though this EXCLUDES route=mtb routes).
In the USA, this has resulted in a consensus-based wiki (initially, and eventually “running on its own power” of having consensus-achieved, now let’s “simply implement what we’ve agreed upon”), United States/Bicycle Networks - OpenStreetMap Wiki . There is a “companion” wiki, United States Bicycle Route System - OpenStreetMap Wiki which denotes the specifics of our “national numbered bicycle routes” (and describes how there are OTHER, but un-numbered national-level, network=ncn routes. (We call these “quasi-national” and both denote them and define them carefully). There are also several state-level wiki which talk about those routes which are at a state-level (in the USA’s hierarchy) and denoted network=rcn.
I’d be wary about “distance-based” classifications with bicycle routes, although those can work for things in OSM like distinctions between route=train relations which cleave between, say, passenger=regional and passenger=suburban, those being “somewhat” distance-based. But for bicycle routes, these (again, in the USA) do “coalesce together into networks” based almost exclusively on the political jurisdiction which administer them. Only when they are “non-governmental” (for example, signed by a non-profit organization) do we (in the USA) allow distance to guide us in which network=*cn classification to choose.
But please explain it with better text. Crossing international borders doesn’t change a network=rcn to network=icn, i.e: https://www.openstreetmap.org/relation/107338 (Saar-Radweg / Vallée de la Sarre), the same is valid for network=ncn.
And there may be really big differences across different countries not to speak of different continents, see @stevea who was a bit faster than me.
The classification in Germany is the result of a discussion about 15 years ago. network=rcn should be about 150 km or longer.
We also (in the USA) find cycle_network=* to be a highly useful tag (communicative of the specific network). This has spread out among the world’s countries in a slow, but noticeable way over the last 15 years or so.
I have database with bike routes from Europe. I only imported relations. Some of the routes are super-relations (their members are other relations). I filter out those “child relations” and only counted super-relations.
SELECT
network,
round((AVG(ST_Length(geom::geography))/1000)::numeric, 2) as avg_distance_km,
round((percentile_cont(0.5) WITHIN GROUP (ORDER BY ST_Length(geom::geography)/1000))::numeric, 2) as median_distance_km,
COUNT(*) as route_count
FROM velo_routes
WHERE geom IS NOT NULL
AND network IN ('icn', 'ncn', 'rcn', 'lcn')
AND is_child_route IS NULL
GROUP BY network
ORDER BY avg_distance_km DESC;
I guess this could work but we should link to it from other pages. I was expecting that page about bike routes will have all the info but couldn’t find it initially.
The xcn tags seem to be intentional to represent a specific kind of route.
Of the walking tags (*wn), some seem to be actually hiking routes (so the error may be in the route tag), or an attempt to map a shared cycling and walking route, e.g. this one:
In case anyone is unaware, this example is no accident: network=ncn or uk_ncn used to literally represent the UK’s National Cycle Network, while network=lcn literally represented the London Cycle Network. By the time we started mapping cycling routes in other countries, these same tags were simply redefined to mean any national or local cycling network, and the uk_ prefix was dropped. That way, OpenCycleMap could continue associating the UK-specific tags with UK route colors without doing anything special for other countries (assuming everyone was OK with those colors). This is more or less how the highway=* values motorway, trunk, primary, secondary, and unclassified developed as well.
Later on, we realized that three networks isn’t enough for each country, so cycle_network=* was introduced as a workaround, using a more flexible format similar to network=* on other kinds of routes. In hindsight, we should’ve born the temporary pain of using more complex values in network=*, using some other key like network:scope=* to help renderers make generalizations without hard-coding long lists of networks.
Thank you yet again, Minh! I’ve very much got my listening ears on here. I’m still sometimes amazed *cn stays glued together as a wet, messy thing, but it works somehow.
Shoulda and coulda here and there? (For translators English, “should have” and “would have” as spoken-spelled). Sure. Hindsight can be a luxury afforded as long as we learn from it going forward.
There is clear road ahead to “make it shine,” I am listening. I remember some fairly vivid differences between the way things are done in parts of Europe vs. my USA perspective, and there may simply be a fundamental breakdown of semantics between here and there, to the point where we especially carefully document the differences. Indeed, wiki being somewhat regional with language, this is already true to some extent.
“Going by total length” is a certain way of looking at things, though to my (North American skewed) perspective it’s a way that isn’t “so” important (around here). Though, I can see how length dominates in visual ways. So, I listen. There’s a lot going on, visually, linguistically, right in the guts of this, and it’s a conversation OSM has now and again; I recall there being some difficulties. Totally solvable difficulties.
So, I listen. We’re talking. We’re displaying crisp, helpful charts. Good.
