[OSM-talk-be] Fietsknooppuntennetwerken/Cycle node networks in Belgium

[Jo]

Hi,

As you probably already know I've been working very intensively on the rcn
routes for the past month. I brought together the routes into networks and
created a collection relation to refer to the network relations.

http://www.openstreetmap.org/browse/relation/1726882

I've been working on a Python script which runs inside JOSM to help me with
the chore of checking each and every node/knooppunt and the route relations
connecting them. The more important function of this script is to regularly
do quality control on the whole bunch of relations and nodes top down and
from the bottom up.

I also created a wiki page:

http://wiki.openstreetmap.org/wiki/WikiProject_Belgium/Cycle_Routes/Node_Network

This page summarizes the remaining obvious problems with the nodes and the
routes. It's easy to see where there are less than 3 rcn routes connecting
to a node, how many roads start from/end at/pass through a node and links
are provided to jump directly to the node on the map.
In the tables on the right side of the page the routes which still have
problems are shown. I check whether they are continuous in forward and
backward direction (I've been sorting all their members) and I've been
adding forward and backward roles on the members that needed them. I realise
now that I didn't do it as described here though:

http://wiki.openstreetmap.org/wiki/User:Ldp#.28C.29_Network_tagging

Instead I did it this way:

http://wiki.openstreetmap.org/wiki/User_talk:Ldp

So this might be(come) a problem. I'm still trying to come up with a way to
check forward/backward continuity of the route relations with the
'conventional' way of tagging roles.

Ldp also suggested splitting up the network relation for Limburg which
contains more than 800 members. Which brings me to my next point. When I
started this tour of Flanders, I was convinced I was going to discover some
sort of logic with regards to the numbering of the nodes. There seem to be
nodes which actually do belong together in a network, but there also many
for which it is hard to tell to which network they actually belong. I've
been trying to create networks where each node number appeared only once,
but I failed in the attempt. So at some point I created imaginary networks
in despair with ridiculous names like Dijlelandse Kouters. At other points I
simply added two nodes with the same number to the same network.

There are also nodes for which it is hard to pinpoint one OSM node, across
canals on roundabouts. These already share the same node number and I'd like
to have a way to distinguish them from the other nodes which happen to have
the same number

Anyway, do we throw all of them into bigger relations? We could also work
multilayered.

Now there is this:
collection relation (Belgium)
   network relations (39-40 of them)
     route relations and nodes

Which could become:
collection (Belgium)
   collection (Provinces W-Vl, O-Vl, Antw, Vl-Br, Limb., Liège) (In the
other French speaking provinces there are no node networks (yet))
      collection (Kust, Westhoek, Brugse Ommeland, Meetjesland, Leiestreek
(W.Vl & O.Vl) Vlaamse Ardennen, Scheldeland (Antw en O.Vl), Kempen (Antw. en
Limb.), Hageland, Pajottenland, Maasstreek, Midden-Limburg, Haspengouw
(Vl.Br en Limb.), Voerstreek, Höhes Venn)
        network (39-40 of them)
          route

The advantage would be that it would be easier to manage. When somebody
needs a GPX containing all the routes and nodes for only Hageland or just
one part of Kempen or all networks of the province of Antwerp.

Another issue is whether to include the nodes in the route relations, or
not. I took them all out except for the route relations in Limburg, as in
the rest of the country that seemed to be the convention. There is no real
need to include them, since they are already included through the ways and
now they are also members of the network relations.

And the last issue is, when one node is represented by several OSM nodes. I
created an extra relation to connect them (note=52-52) and the ways that are
members of this relation are not members of the route relations that
start/end at these nodes anymore. This simplifies matters a lot. So the
route relations start/end at the nodes and don't interconnect those OSM
nodes anymore.

Anyway, let me know what needs to be done differently and I'll try to
comply. It would, of course, also be nice if we could find ways to add the
nodes that are still missing and to fix the problems reported on the wiki
page I created. (But those are the harder problems, which need extra
surveying, the easy ones have already been dealt with by me, myself and I)

Kind regards,

Jo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-be/attachments/20110928/68bb9c5d/attachment.htm>