OpenStreetMap
I hopped on the MTB today and surveyed North Lismore. There are a large number of unpaved roads in the area. Unfortunately the renderers don't show this but labeling them all highway=track would be inappropriate.
Wilsons River is proving annoying to geolocate, mainly because the Landsat imagery is so out of alignment and it's difficult to geolocate due to it's low resolution. There's an ABS boundary along the river but I don't know if it's meant to go down the middle or along one bank. Since the river is ~40m wide it kinda matters.
Short of renting a kayak I'm at a bit of a loss as to how to draw it accurately. I've taken some waypoints to represent the banks underneath the 2 main bridges, maybe them plus some interpolation is the best that's going to happen.
Oh, I also did a few more streets around Lismore Heights, so the list of suburbs which have complete street tracing are:
North Lismore
South Lismore
Lismore
Lismore Heights
East Lismore
Girards Hill
Goonellabah is to follow, with a small chance of Alstonville being traced before I return home early September.
Other partially traced areas in the region include:
Coraki
Casino
Ballina
Woodburn
Evans Head
Byron Bay
etc...
The rural road network to the NE of Lismore is surprisingly complete though :). There are some omissions (which will be filled in during endurance training rides) but they probably stand at ~70% complete, not including dead end roads.
Lastly, I started to trace out landuse=farm, produce=orchard areas last night from landsat/knowledge. There are lots of macadamia and sugar cane plantations in the area and marking them a) makes the map more useful and b) makes it more interesting to look at.
I should stop now and take a shower...
Discussion
Comment from wamble on 13 July 2009 at 07:08
Nice work
Comment from wallclimber21 on 13 July 2009 at 07:24
Why would it be inappropriate to label them track? That's basically what I do for all unpaved fire roads on which I mountain bike.
Comment from Biogenesis_ on 13 July 2009 at 07:36
Track is fine for fire trails, the problem is that these streets are very firmly residential. As in, they have houses on both sides at no more than 20-30m apart.
If there's a consensus that any unpaved surface could be marked as track I would be happy to follow that convention, but currently the OSM data convention is to label the way with the most appropriate purpose, leave a surface=unpaved tag then let the renderer use an unpaved linetype.
Unfortunately neither Mapnik or Osmarender use a different linetype to differentiate between paved and unpaved. No doubt it's been debated on the appropriate mailing list.
Perhaps we need an Australian convention as it may be possible that in developed parts of Europe an unpaved residential street is so uncommon that it's not considered important.
Anyway, whatever happens the data is there for anybody to interpret as they wish, so the hard work is done.
Comment from wallclimber21 on 13 July 2009 at 08:08
Ah, ok. In my case, those tracks are definitely not residential. In the US, the problem is usually just the opposite: that vast majority of data that was imported from the US census database has been marked as residential, even if it's smack in the middle of the Arizonan desert with no house in sight. :-)
Comment from Circeus on 13 July 2009 at 20:23
Well, I actually agree, for the most part. I don't think making a difference between "highway=unclassified" and "highway=residential" is a good idea. Having a "highway=residential" was as far as I'm concerned a bad idea from the start, in fact. We should have only one term fro the lowest "level" of street, and unless there is a functional difference besides what landuse surrounds it, separating these two "definitions" leads to ridiculous stuffs in either direction (why should a street in a commercial estate be marked as unclassified and suddenly become residential? Why should streets that are undifferentiated on the ground become so on the map??).
Comment from Biogenesis_ on 14 July 2009 at 01:26
So you think that it's a better idea to just have "highway=lowest" then have it defined as residential/unclassified/whatever by what landuse is marked around it?
Personally I think it's a bit of a moot point for the time being as the renderers don't seem to differentiate between residential and unclassified. Also, getting the data there in the first place is a much higher priority than nitpicking over exactly what it should be classes as.
Comment from Circeus on 18 July 2009 at 06:20
Actually although Mapnik doesn't make the difference, I'm fairly sure Osmarender does. AFAIK, unclassified should be officially considered deprecated, but getting something like that agreed is like herding cats, so I won't even try writing up a proposal. Heck, as far as I'm concerned the UK-centric approach to non-car routes has made a MESS of non-road stuff between access tags and path/track/footway/bridleway/byway/cycleway. And now they want to add an extra layer with an
officialtag!Comment from Biogenesis_ on 18 July 2009 at 07:39
Osmarender doesn't look to render anything differently with the surface=unpaved tag. Ref: Pete, Boorie and Currie Streets.
I might write a patch for it or something. At least then maps I render will be drawn more informative and if it doesn't get accepted then such is life.