LINZ/NZopenGPS to OSM conversion process
Joe Richards
I've had a go at mapping the runways and airports from LINZ for the
Chatham Islands.
We uploaded this file for the runways - http://gist.github.com/351267
Any feedback would be good.
Rob Coup has written a tool to handle the mapping of LINZ data to OSM
types - If anyone else is keen to have a go at mapping some of the
other LINZ layers then let me know and I can add you to the user list.
just drop me an email - glen [at] open [dot] org [dot] nz. I'm away
this weekend so won't have much internet access but can add you next
week.
Hamish
> It's great to see some progress on the LINZ->OSM import - good work! I'd
> like to see if I can understand what's happening so maybe you can fill me
> in...
(me too :)
>> I've had a go at mapping the runways and airports from LINZ for the
>> Chatham Islands.
Great! I did some research yesterday and then cleaned them up a bit.
LINZ provides runways as polygons, but OSM needs them tagged as lines.
So I converted the runway boundaries to be areoway=apron. The osm wiki
indicates that this is for airplane parking lot areas, but there is
nothing closer and it is what Heathrow uses for the runway verges, so
I figure it's good enough for Chatham Isl. airport too. Check it out,
I think it looks pretty good. I added by hand & a bit of paper on my
monitor (folded in half to get a center line) an actual runway line
and transfered some of the attributes across. (Pitt Isl. airstrip
seems to be missing any?) All three runways render a bit differently,
I'm not sure why or which is best. Pitt doesn't use a Relation, I
don't know why or if that matters or not.
For the rest of the country I or someone else can process the LINZ
runway polygon shapefile to hunt out rectangular-looking features and
create runway lines with the appropriate attributes copied to them. I
figure something like cycle through the vertices, and look for the
longest two diagonals >400m which are within 5% of each others length
and not within 5m of each other; then collapse those to midpoint
lines.
point being, we can place the aprons now and deal with the runways
sometime later.
> I might have missed it, but by what process/source/script etc was this data
> created and then uploaded?
the github link below says linz2osm v0.6. (or is that osm API v0.6?)
by all means, the shp2osm rules should be available & maintained in OSM svn.
>> We uploaded this file for the runways - http://gist.github.com/351267
>
> Did you upload it from JOSM? Just trying to catch up with the recent work...
>>
>> Any feedback would be good.
roads without name or LINZ ID (and thus access=unknown) were missing
highway=road.
as this came from the linz road layer, we can be pretty confident that
those should be tagged as such. I've gone through and edited them by
hand (I think I got them all) to add that. now they render (or at
least I hope will once the tiles@home download cache flushes through).
> Do we fix 'bugs' by modifying a conversion script, or basically by
> correcting the erroneous data that the uploader script/person has uploaded?
for the chathams I'd treat it as a learning exercise and don't mind
losing all hand edits as we figure out what the final output should
look like. for the mainland I have no wish to have to fix anything by
hand at all. the conversion scripts can of course highlight and
extract problem areas which need to be fixed. but tons of automation
is the only way we're going to be able to get through this.
>> Rob Coup has written a tool to handle the mapping of LINZ data to OSM
>> types - If anyone else is keen to have a go at mapping some of the
>> other LINZ layers then let me know and I can add you to the user list.
>> just drop me an email - glen [at] open [dot] org [dot] nz. I'm away
>> this weekend so won't have much internet access but can add you next
>> week.
>
> Could we share this tool? It seems more sensible to work on the tools first
> collaboratively and then only upload data when the tool has been tweaked
> sufficiently.
thanks, glad to see it happening..
Hamish