Trying to make sense of it all

I am new to map editing. I started to address the issue when I was creating cycling routes using GPSies. There were places where the route (for no obvious reason) would just do a loop around surrounding streets and end up almost at the same place before continuing on along the selected track.

I discovered OpenStreetMap editing and found there were places where tracks were not quite joined up. So whenever this happened I logged in and corrected the issue. This led to finding a whole lot of issues: e.g. tracks and roads not lined up with the underlying satellite images, missing paths, features not included on maps or incorrectly located (like a public toilet a hundred metres away from where it was marked on the map and the other side of a busy highway.

But I soon came up against the fact that things (especially tracks and roads) are marked differently in different countries, states, regions and in some cases one side of town compared to another side. Major issues like sections of the Hume Hwy / M5 that are marked as “motorway” long after the motorway conditions officially end.

So my questions are:

  1. is there some compilation of decisions about (or standards) for markup in Australia?
  2. are there any areas of responsibility allocated to particular individuals / bodies? which the casual editor should not touch.

My particular interest is cycling routes - we have:

  1. on-road cycle lanes,
  2. road verges / breakdown lanes used by cyclists,
  3. dedicated cycle paths,
  4. long tracks between towns or suburban areas used mainly by cyclists
  5. shorter tracks within and between suburbs
  6. commuter paths joining residential areas to shops, schools, commercial areas, community centres etc.
  7. neighbourhood footpaths running beside roads, but separate from them
  8. and then there are ‘routes’:- any of the above joined together to form particular routes (e.g. the Canberra Centennial Trail has fire trails, MTB single track, minor suburban roads, main roads, sections of highway, etc., and how about the National Bicentennial Trail)

The above (1-7) are mainly sealed or concrete, but there are the off-bitumen equivalents on dirt and gravel roads, fire trails, right down to narrow single-track MTB routes, worn foot tracks.

There seems to be inconsistency in how these are marked up. This shows up quite distinctively in OpenCycleMap where for instance northern areas of Canberra at zoom level 5 appear as a dense network of blue mainly footpaths along and between suburban roads, while southern Canberra has just as dense a network of paths marked which do not show up the same way at that zoom level. There are mountain bike paths in Kowen Forest east of Canberra that show blue, yet those at Stromlo Forest on the west of Canberra show up orange. Most main bike paths are blue, but a few show purple.

Is there any suggestion on how to standardise this kind of markup? Maybe a set of example to judge against as a comparison?

Trying to make sense of it all.


You can also contact the Australian OSM community

There are a lot of issues there, but note that the gold standard for alignment is GPS measurements, taken on the ground, not aerial imagery (imagery precise enough to even compete with GPS is not going to be from satellites, but rather from aircraft).

Aerial imagery can be quite a long way from the true GPS position, especially if there are significant height changes, as the last mentioned can introduce parallax errors which may not be fully corrected. Different editions of even the same aerial imagery source can have significantly different offsets from reality.

Unless the map is wildly at variance from other sources, it is normal to accept that local map as being the true reference, and to move other sources so that they correctly overlay what is already mapped. JOSM certainly allows aerial imagery to be offset, and I seem to remember that the same is true for iD.

If you feel that the mapping really is a long way from true WGS-84 reality, and needs correcting, make sure that the corrections do not cause the true topological relationships between objects to be violated. Obviously your public toilet is currently violating the true topology and needs fixing.

Remember that consumer grade GPS measurements, especially ones made at only one time, may have errors of up to around 15m.

Congratulations on trying to mark up Australian cycling infrastructure. As a cyclist you would know that there are some bike lanes that are clearly poorer than others and minor differences in the way roads allow cyclists to interact. A seasoned road warrior has a different set of on-road requirements to a family group cycling together.

The key issue is that most roads are done by those looking at the satellite imagery and it does not give the nuances needed for cycling, nor is there a set list of statii. has some datasets but little to assist you.

The best bet is to make it your call. As a cyclist, you know better than the majority on what is and is not a cycle/shared path however try out Strava ID at (not currently supported), turn GPX tracking on and turn Mapillary on (as well as submitting images to Mapillary)

If you have any specifics, shoot me a dm. Good mapping and keep the rubber side down.

You need to be careful about subjective decisions. In general, you should record objective facts on which a router can make decisions about suitability, rather than decide whether something is suitable for yourself.

I am not going to map a bike route that is dangerous. City planners don’t ride these routes and they can be inherently dangerous in places so I will be subjective in places.

That is OK, but remember that it is also not OK to delete mapped bicycle lane that is dangerous if it exists (one may try to tag difference between poorly made and well made cyclelanes - width etc but it is tricky).

Also, you can now distinguish in tagging “proper” cycle lanes from those “dashed” cycle lanes with this tag: