On 31/07/14 17:01, Nick Allen wrote:
actual stop was. As a result of this, it is quite possible that there are stops shown in the right place with the correct tags concerning shelters, benches, tactile paving, name of stop, timetables or bus route references displayed, etc, but added to a NaPTAN reference which new data would show to actually be elsewhere.
I've certainly found a lot of misplaced NaPTAN nodes, and ones where the logical stop has moved.
The only way of improving this would be a line by line comparison, with perhaps the transfer of a reference from new NaPTAN data, to the existing OSM data. If I appear in the history of the bus stop (Ctrl+h using JOSM), then there is a bus stop in that exact location. I suspect that this is true of many stops elsewhere, so some cautionary line by line merging is going to be needed.
The OSM history mechanism is flawed in that it can only track one side of a merge (more generally, detaching a node from a way without creating a new node can be difficult, although not many NaPTAN nodes have been accidentally attached to ways).
several stops which I have created, for which there is no NaPTAN data in the current database.
A small proportion of NaPTAN data has gone missing either through personal mapping (people don't want the stops cluttering their map) or because people decided that the stop wasn't present, but never read the bit about leaving the node, marking physically_present=no and not a bus_stop. I've definitely seen the former and there are cases where the latter is the most obvious explanation. One of the things I find missing from the process is actually an audit against the original import, to recover those cases - or to detect deliberate damage to the imported information.
Another issue with NaPTAN is stop area names. There is nothing on the ground to identify these, but the actual stop names change relatively frequently as landmarks come and go. Often stops within a stop area have different names. At the moment, this means that as you zoom out of the transport map, the name can change to one that is no longer/not used on the ground. I don't know if a re-merge would help with that, or whether there is a more fundamental problem with stop area names.
On the more general issue of maintaining bulk data, I've found I've inherited several high streets where someone takes a photograph, maps it six months later, and never returns to check changes, something which really needs doing at least every three months. I have taken on maintaining those within a couple of kilometres, but I'm aware of other ones which I don't visit often enough to recognize changes, or detect errors in the original mapping.
_______________________________________________ Talk-GB mailing list Talk-GB@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-gb
