OpenStreetMap

Southwark buildings and Mulberry Bush mapping last week

Posted by Harry Wood on 7 June 2011 in English. Last updated on 8 June 2011.

We went to the Mulberry Bush pub again. Being disorganised I was in need of a pub decision at short notice, and when that happens I might as well pick somewhere easy because it's not going to be a big meet-up anyway. Sorry about that. But I do also like the Mulberry Bush for getting a bit more daylight than most pubs, and the weather was great in London last week.

But actually, looking at the building coverage situation, it turned out there was a couple of little gaps not a million miles from the pub. It was just over a year ago I drew some diagrams suggesting that we might aim to form the London building coverage into a blob around a central area. At that stage, the South Bank had a few buildings and a few gaps. Further South from there there was the usual odd sprinkling of uninteresting housing estates mapped with buildings (because that's only way to make sense of mapping a modern housing estate). ...but in 2011 we have a new triangular spear of very complete-looking building coverage extending down much of the London borough of Southwark. I believe this is mostly the work of Tom Chance vectorising StreetView (and other techniques described in the comment below) Tom's contributions shown here in red:

Tom Chance Mapping South London

So there was a little bit of building mapping to do because this red area was (and still is) making some of the northerly parts of Lambeth look gappy. This time I set up the cake diagram on MapCraft from the start. Slight glitch as it went offline the day before the mapping party, but apart from that it seemed to work, but we'll have to try it at a slightly bigger scale. It was just me and a new guy Matt Black doing some mapping. There was six or seven of us overall in the pub. A cosy little gathering.

Hydrogen bus! OpenStreetMap Mulberry bush OpenStreetMap Mulberry bush Shaun reading OpenLayers book

We talked about...

Emailing users about agreeing to OdBL and the contributor terms. A batch of emails have gone out to some users who didn't click Accept or Decline, to prompt them to do so. Do it here if you didn't already do so, but you probably did because otherwise you've been blocked from editing OpenStreetMap, which I'm sure you would have noticed!

We talked about train fares to airports. Gatwick express, Stansted express, Heathrow express. All a rip-off. Heathrow Connect is a lesser version of heathrow express which apparently was originally designed for getting the airport staff there. Less express, and less of a rip-off (but still a bit) At least you know where you are with rip-off airport trains. Try making sense of all the other train prices or the tangled mess of companies involved in UK trains.

We talked about plans for SOTM EU. I haven't signed up to go to it because I'm too disorganised and indecisive, but some of the other guys were getting together to share accommodation.

We mentioned Ollie's slippy map display of Muki's completeness analysis. See how this map has become more blue over time.

We talked about OpenLayers, probably because Shaun had a copy of the new OpenLayers book with him. Didn't get chance to look at it myself. We also talked about use of OpenLayers on the OpenStreetMap.org homepage. Firefishy told us, of all the bytes served from the www server, an amazing %37 of the traffic is javascript files. Not sure if that's before or after compression, but this is visitors newly downloading the OpenLayers library as they view the slippy map. TomH explained we actually serve up a reduced version of OpenLayers, with some unused features dropped from the javascript file. Despite this, we're dishing out a lot of javascript.

We talked about the definition of a tunnel versus a bridge. We were talking about the roads passing under many railway tracks at London Bridge. Most of us seemed happy with the rule of thumb: If the road passing under is longer (underneath) than the gap is wide, then it should be a tunnel on the under way, rather than a bridge on the over way. Dont know if anyone's attempted document such a rule.

Twain's done some development work on Nominatim which would allow the search to work better with UK postcodes (by supporting OS OpenData "CodePoint" I think we said) but he's not deployed this yet.

And finally I pondered whether we should have some sort of email subscription list for OSMLondon. @OSMLondon on twitter is probably the most reliable channel for getting nagging reminders about London events. I always update the wiki page too, but that's a bit less nag oriented. There's nothing quite so nagging as an email. I suppose ideally we'd have a system for emailing the text of tweets to people. I'm mainly thinking of enticing more people in, or making sure the vaguely interested (and not on twitter) people get a reminder.

So speaking of OSM London matters, hopefully we can resume our normal service after a rather manic period of geo-osmness (and I didn't even attend WhereCamp.EU) So I now need to decide where to do the next event in just over a week's time. Ideas please!

Location: South Bank, Waterloo, London Borough of Lambeth, London, Greater London, England, SE1 9NL, United Kingdom

Discussion

Comment from Rovastar on 7 June 2011 at 16:30

Thank you for the time in producing this updates of London mapping Harry. Always an interesting read.

Maybe one day I will come along to one often in London.

Recently finished of the City mapping so all the OS/itoworld roads are 100% now. And started doing the land use zoning there (well it is nearly the entire city area is bright pink:commerical) together with building mappings...London in general doesn't seem to have too much land use assigned......

Comment from smsm1 on 7 June 2011 at 18:57

That wasn't my OpenLayers book, i just happened to be taking a quick look at Ollie's copy.

Comment from c2r on 7 June 2011 at 20:42

Ooo, now the tunnel vs. bridge discussion is interesting... I've always used the rule of thumb that design of the bridge/tunnel is the primary consideration. For example, a road intersection over the top of a diagonal river/stream, like the one at New York http://osm.org/go/eu_9yw66e-- which on a site visit is clearly a sunk pipe to tunnel the stream under the road.

Whereas at other places, such as the Dog in a Doublet http://osm.org/go/eu74nFihD-- the design is clearly a bridge...

Now, in these examples a rule of thumb around wide vs. long could currently be applied... but conceivably in the first design the tunnel could be quite wide with just a single track lane going over it, and in the second design the road going across could be a ten lane motorway....

Just some thoughts
Chris (o:

Comment from Harry Wood on 7 June 2011 at 23:30

Hmmm yeah. I've noticed while doing motorway fixup in the U.S. (old TIGER fixup work), that waterways are more problematic when it comes to tunnels versus bridges for some reason. Maybe we're less likely to think of a waterway as being in a "tunnel", unless it's actually in a pipe.

This one near Finsbury Park might be thought of as a road tunnelling under the railway, but not really, because the tracks have separate metal bridges with daylight between.

Maybe we should try to expand the section Editing Standards and Conventions#Bridges

Comment from c2r on 8 June 2011 at 06:25

Agreed - I think waterways would tend to be bridged over (unless sunk into a pipe, which is clearly a tunnel; or physically tunnelled canal style through a hill, or something.

I know exactly what you mean at Finsbury Park there - I used to drive under it every day (shudder) - that's definitely a series of metal bridges that you travel under...

Now, looking at: http://wiki.openstreetmap.org/wiki/Key:bridge

we have the following text:
In case of an undercrossing/underpass you have to decide on the basis of the construction if you are over a tunnel or on a bridge.

as well as:

When there are several bridges close together it is sometimes acceptable not to tag them as bridges, but to tag the single road below as going through a tunnel.

which is helpfully ambiguous!

Comment from Tom Chance on 8 June 2011 at 09:39

Thanks for the update Harry, though I very rarely make them it's interesting to read what was discussed.

The buildings are a mix of sources/techniques. The middle section in north Peckham/Walworth was by automatically vectorising, fixing up and merging in OS StreetView data. The north section is a mix of StreetView, Bing aerial imagery and on a few roads ground surveys because the buildings are quite hard to understand from imagery alone. The southern section around south Peckham and East Dulwich is all ground surveys with Bing and StreetView to get good building outlines with full address info. That technique takes absolutely ages but it looks nice and is hopefully useful :)

Personally I'd like to see at least StreetView traced buildings across the whole of London, with improvements using Bing and ground surveys to gradually refine it. As with all the Yahoo! traced roads and landuse areas, I prefer lots of data that needs some attention than none at all, especially if we had a "no names" type view to show buildings that need to be checked.

Thanks for the bridge/tunnel distinction, I tried to make the London Bridge ones a bit more consistent a year or so ago. That definition makes a lot of sense, very practical!

Log in to leave a comment