[talk-au] Historical rail lines

[mick]

On Mon, 26 Nov 2012 13:18:30 +1100
Ian Sergeant <inas66+osm at gmail.com> wrote:

> On 26 November 2012 12:36, Alex Sims <alex at softgrow.com> wrote:
> 
> I don't think so.  Keeping historical data in OSM is going to require a
> more complex model.  Maybe a separate project, maybe layers, maybe
> something else.
> 
> There is a mailing list and a wiki page set up to gather ideas..
> 
> http://lists.openstreetmap.org/pipermail/historic

Thanks for the heads-up on the historical list, somehow I missed its birth notice.

> There are two main aspects to consider here.  
> Firstly, how you map that what no longer exists?

Most of my mapping interests relate to historical mapping, English parishes, Civil and Church, for my family history. Studying the Roman occupation of Britain, etc.

My usual methodology is to start with available, free to use, current maps; For UK OS Open Data, for Aust. Geoscience Australia TOPO250K series and for both OSM subsets; And clean them up.

Next, grab what I can find of earlier maps, georeference and trace them and then pull them into shape, I always seem to got plenty of distortion.

When the map layers are as clean as I can get them, I overlay them and copy the required details onto a fresh layer and work-up suitable tagging.

> Secondly, how you track changes made to OSM, so you
> can capture history within the OSM changesets.
> 
> The first one we have plenty of time.
> 
> The second we need right now, every addition I make it is impossible to
> tell whether I'm adding a new feature that didn't exist on the ground
> before, or just filling in a feature that has always existed but wasn't
> mapped.  And every feature I remove, it is impossible to tell if I'm
> removing it before it is wrong, or removing it because it has been
> demolished.  So, we're actually losing information as we go.
> 
I fear this can only be successfully done on the micro scale unless the tagging guidelines can be tightened. The free and easy approach to tagging makes "outside the box" application of the data a largely manual job.

I'm sure this is far from insurmountable, I just can't get my head around it.

mick