[Openstreetmap] UK railways
Andy_J_Robinson at blueyonder.co.uk
Wed Jan 18 18:06:28 GMT 2006
OSM has one dataset which has an unknown level of accuracy resolution, ie
accuracy of the data is not verified (except where it comes from verified
sources - TIGER?). It's up to users to edit the map using gps data and any
other data they care for (as long as it does not infringe copyright).
Having said that if I overlay my gps based mapping onto Google maps it's
very good, but zoom all the way in you still find there are subtle
differences between OSM's data and that used by Goggle (Teleatlas/Navtech
data). Doing this also shows up the obvious discrepancies caused by editing
OSM data using only the Landsat imagery.
The accuracy of OSM data will further improve with improvements in gps
technology - role on Galileo - anyone got a signal yet ;-) and more detailed
editing and local knowledge. However from my experience of what I have
produced to date, provided the streets are not buried in the urban jungle of
high buildings or where there is significant tree cover, the end result from
gps mapping at 1 sec intervals and good editing of this data is likely to
produce a map with more than sufficient accuracy for accurate navigational
use on the ground. It should produce some pretty maps too.
Take a look at the Sutton Coldfield area (7 miles to the northeast of
Birmingham, UK) using Just van den Broecke's excellent Google/OSM overlay
...and you will see that accuracy of final data is unlikely ever to be an
issue. It will just take time to get to that level of confidence in some
areas where gps tracking by contributors is less effective.
Andy_J_Robinson at blueyonder.co.uk
>From: Tyler Mitchell [mailto:tylermitchell at shaw.ca]
>Sent: 18 January 2006 17:06
>To: Andy Robinson
>Cc: openstreetmap at vr.ucl.ac.uk
>Subject: Re: RE: [Openstreetmap] UK railways
>I'm just curious, do you have multi-scale data representations? That is,
>do you maintain a set of data that is less accurate but suitable for, say,
>Of course if you have fine resolution information you can always dumb it
>down. But someitmes all you can get are interpretations off Landsat - have
>you considered throwing that data into a different "lo-res" layer instead?
>Just curious how you've thought through this in the past.
>----- Original Message -----
>From: Andy Robinson <Andy_J_Robinson at blueyonder.co.uk>
>Date: Wednesday, January 18, 2006 8:35 am
>Subject: RE: [Openstreetmap] UK railways
>> It's a mixture; a lot of the recent edits for the UK appear to be
>> basedloosely on a mixture of low density gps points and the
>> Landsat. Each
>> motorway trip I've done I've gone back over and re-edited the original
>> segment placement done by others. In the case of the M54 a couple
>> of weeks
>> ago I had to shift both lines a lot because it was originally
>> drawn with
>> reference to one low density gps track and the editor obviously
>> picked the
>> wrong direction for the carriageway assumption. The error is still
>> there to
>> the west of Telford because that's as far as I went on that trip.
>> Overall though, the error is relatively small (probably sub 50m in
>> mostcases) so in a global sense it's still pretty accurate.
>> Andy Robinson
>> Andy_J_Robinson at blueyonder.co.uk
>> >-----Original Message-----
>> >From: openstreetmap-bounces at vr.ucl.ac.uk [openstreetmap-
>> >bounces at vr.ucl.ac.uk] On Behalf Of Jonathan McDowell
>> >Sent: 18 January 2006 15:58
>> >To: openstreetmap at vr.ucl.ac.uk
>> >Subject: Re: [Openstreetmap] UK railways
>> >On Mon, Jan 16, 2006 at 01:15:28PM -0000, Jon Stockill wrote:
>> >> Inspired by the amount of work done creating motorways from the
>> lansdat>> data I set about doing the same for railways,
>> >Have the motorways really been added by landsat rather than
>> recorded GPS
>> >points? Doesn't that imply there could be a lot of error in them?
>> >I did a lot of the M11 and I used GPS points I'd taken myself.
>> I'd been
>> >assuming that's what others were doing as well.
>> >Where are we going? And WHY ARE WE IN THIS HANDBASKET?!!
>> >This .sig brought to you by the letter H and the number 29
>> >Product of the Republic of HuggieTag
>> >Openstreetmap mailing list
>> >Openstreetmap at vr.ucl.ac.uk
>> Openstreetmap mailing list
>> Openstreetmap at vr.ucl.ac.uk
More information about the talk