[Talk-GB] Estimating coverage again
I've uncovered and fixed a few errors in my previous estimates of the % of UK roads included on OSM. I discovered a few days ago that there were a few areas where different boundaries overlapped, with the result that I was double counting some roads in more than one local authority. The problem areas included Rochdale, Thurrock, Stoke and Derby. I think I've fixed them now, and the updated map is here - http://www.reedhome.org.uk/Documents/OSMCover.png Unfortunately I extracted the revised data a day too early to pick up the new admin boundaries in Wales. But at least the data in map and the spreadsheet here - http://www.reedhome.org.uk/Documents/OSMCoverage.csv - is a little more accurate than it was. I've also been experimenting with ways to flag up residential areas that don't seem to have enough roads plotted, and to highlight roads that haven't been fully tagged. My first attempt is crude, but if anyone is interested, the map is here - http://www.reedhome.org.uk/Documents/OSMCover%20with%20extras.png (roads that are only classified as highway=road or highway=fixme are in plotted in black, and the little red splodges show area that are marked as landuse=residential but have a very low density of roads within them) I've not forgotten the helpful hints on improving the colour scheme, I still want to find a way to use the NUTS boundaries, I'd like to find a way of speeding the whole process up, and I'll pick up better data for Wales next time. I'd also like to find a way to trim each local authority at the coastline to make it look a bit better... ...but one thing at a time, and it seemed more sensible to fix blatant errors. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
[Talk-GB] Estimating coverage
I have now uploaded a summary of the figures for the UK coverage estimates as a CSV file here - http://www.reedhome.org.uk/Documents/OSMCoverage.csv Several people have expressed an interest in seeing the proportion of named roads by local authority area. Those figures are included in the table. There is also a map here - http://www.reedhome.org.uk/Documents/OSMNamed.png The other figure that may be of interest is the proportion of roads that have been plotted, but not completely tagged. See the column labelled Percent other in the CSV file. These roads are not included in my totals, because I can't tell what type they are. Broadly speaking these are roads marked as highway with a type of road or fixme. There are also a few that have been mis-tagged, for example as highway:some street name or with a combination of conflicting highway types. But these are a very small proportion of the total. Most are tagged highway:road which is normally intended to mean I know this road is here, but I have not yet decided what type it is. The overall proportion of these is quite low, but it is surprisingly high in some areas - notably Luton, and Northumberland for example, where almost a third of roads that have been plotted are not fully tagged with a type. Lincolnshire, Suffolk, Somerset, Trafford, Norfolk and Wiltshire also show a high proportion (10%) of roads that I can't classify. I thought that comparing the area enclosed by the admin boundary on OSM with the area published by national statistics would be a good indicator of whether the boundary was accurate. In practice it works sometimes, but not others. I think this is mainly because of how the coastline and estuaries are handled in different places. Bristol is the extreme -the government thinks it is about twice as big as the area included in the boundary on OSM. The CSV file also shows the breakdown between different types of road (motorways, primary, secondary, etc). I've not looked closely at this yet, but a quick scan suggest that the figures for primary roads and motorways could sometimes be a good indicator of how accurate a boundary is. For example, it looks as though the boundary for Luton on OSM includes a bigger chunk of the M1 than the DfT believes they are responsible for. Enjoy. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Hi Nice work! The area figures are obviously including the wet bits. Bristol is half water: http://news.bbc.co.uk/1/hi/england/bristol/7019663.stm I also notice the OSM motorway figures are generally a fair bit above the official figures - slip roads? Cheers Paul (southglos) -Original Message- Date: Tue, 11 Aug 2009 10:55:31 +0100 From: Peter Reed peter.r...@aligre.co.uk Subject: [Talk-GB] Estimating coverage I thought that comparing the area enclosed by the admin boundary on OSM with the area published by national statistics would be a good indicator of whether the boundary was accurate. In practice it works sometimes, but not others. I think this is mainly because of how the coastline and estuaries are handled in different places. Bristol is the extreme -the government thinks it is about twice as big as the area included in the boundary on OSM. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
The area figures are obviously including the wet bits. Bristol is half water: http://news.bbc.co.uk/1/hi/england/bristol/7019663.stm This article mentions Denny Island - which was absent from OSM. I've now added it from the NPE map, although I don't know whether its location has changed with the movement of the sands/mud. Does anyone have a more up to date source for its outline? Donald ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Interesting point from Paul (southglos) about slip roads. I've just worked the numbers slightly differently and he seems to be right. Adding up the total length of motorways in England according to DfT it comes to 6,021km. My total from OSM for England = 6,962km On face value we have found 900km of motorways that the DfT have lost. However if I break down the tags that make up my 6,962 km, it comes to 5,912 km tagged motorway and 1049 km tagged motorway_link. 5,912 km of motorway on OSM compared to 6,021km known to the DfT looks astonishingly close to me. What do you think - should I be ignoring motoway_link in the totals, counting it as something else (and if so what?), do we just put this down to the way DfT count the numbers, or is it just coincidence that the motorway numbers are so close? ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
On Tue, Aug 11, 2009 at 8:07 PM, Peter Reedpeter.r...@aligre.co.uk wrote: What do you think – should I be ignoring “motoway_link” in the totals, counting it as something else (and if so what?), do we just put this down to the way DfT count the numbers, or is it just coincidence that the “motorway” numbers are so close? There are marker posts every 100 yards alongside the hard shoulder of all motorways. I don't think slip roads have marker posts so if the DfT are calculating distance by counting the marker posts then excluding slip roads sounds a reasonable thing to do. Nick. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Nick Austin wrote: There are marker posts every 100 yards alongside the hard shoulder of all motorways. I don't think slip roads have marker posts so if the DfT are calculating distance by counting the marker posts then excluding slip roads sounds a reasonable thing to do. Nick Pedant mode on There are marker posts on (some) slip roads, but they use the same distance as on the main carriageway iirc, just a different letter, to distinguish them Pedant mode off Jeni ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Peter Reed wrote: Interesting point from Paul (southglos) about slip roads. Ive just worked the numbers slightly differently and he seems to be right. Adding up the total length of motorways in England according to DfT it comes to 6,021km. My total from OSM for England = 6,962km On face value we have found 900km of motorways that the DfT have lost. However if I break down the tags that make up my 6,962 km, it comes to 5,912 km tagged motorway and 1049 km tagged motorway_link. 5,912 km of motorway on OSM compared to 6,021km known to the DfT looks astonishingly close to me. What do you think should I be ignoring motoway_link in the totals, counting it as something else (and if so what?), do we just put this down to the way DfT count the numbers, or is it just coincidence that the motorway numbers are so close? Can you ask them? A FoIA request might clarify the situation. Cheers, Chris ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
2009/8/4 Peter Reed peter.r...@aligre.co.uk Meanwhile, there is a nice map on the French Openstreetmap page showing coverage of French Communes http://files.meurisse.org/osm/2009/communes-20090801.png I can read French a little, and I have tried to make out the background from the Talk FR pages. See “Avancement des communes” in http://lists.openstreetmap.org/pipermail/talk-fr/2009-April/thread.html As far as I can see, this map is plotting the extent to which boundaries have been plotted at the level of a French Commune (i.e- local government boundaries, not roads). My French really isn’t up to asking more in the TALK-FR lists, so I’d be interested in any more information from the UK. Can anyone help? What would like to know? The thread you are pointing is updated monthly with new statistics. The boundaries that we have is based from the Cadastre which is a database that is owned by the tax office. We cannot incorporate data directly, but we were given authorization to trace from it. It is giving us access to name of streets, municipal boundaries, and buildings. The data is not yet completely vectorized, and adding a new commune is not that easy even if we have programs to help. We know how much work is done because we also have access to administrative information from the INSEE. It is possible then to check which commune has been properly added against the list of the INSEE. It is not perfect but it works globally well. Emilie Laffray ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Peter Reed wrote: Thank you Emilie. When you talk of adding a commune, do you mean adding (vectorising) the boundary? It looks as though you have access to a lot more detailed information on boundaries in France than we have in the UK. The UK administrative boundaries on OSM have to be plotted from old maps that are now out of copyright. Yes, when we are talking of adding a commune, we are adding boundaries into OSM. We are also making sure that we create the proper relations so one way can be reused for a frontier between two communes. We have broad data on higher administrative boundaries that we are correcting as we progress in adding commune boundaries. We know in which higher level boundaries commune are in. There are currently talks about Spain and Italy boundaries on that ground. The data can be sometimes a bit incorrect so we are verifying everything carefully. Emilie Laffray signature.asc Description: OpenPGP digital signature ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Peter Reed wrote: There have been a number of attempts to estimate the level of UK coverage, of varying levels of sophistication, but I've not seen any that compare the length of roads mapped against actual road lengths. Over the last couple of weeks, I've had a first attempt at doing this for about 100 local authorities with decent boundaries. This looks like fantastic work. By way of helping to inform debate, with a real-world usage example: We have a need for a programmatic means to detect the very rough level of completeness of an area. For instance, we know from personal experience that http://cambridge.cyclestreets.net/ is very complete, but I can tell (to take a random example) that http://lichfield.cyclestreets.net/ is not very complete so far. As a result, people doing route-planning on cyclestreets.net will get poorer routes in the latter area because the data isn't there to plan over. However, we can't tell this programmatically. The benefit, if we could, would be that we can manage user expectations with a message that routing in this area will not yet work well because the map data is incomplete, and know better where to target a roll-out of the system out around the country in a way that gives us more confidence about the results people in each area will get. I'm not quite sure what the solution could be, particularly as the preloaded areas (map centre-points): http://www.cyclestreets.net/area/#england do not correspond to local authority areas as such. But I thought I'd throw in this real-world example to help inform debate.. Martin, ** CycleStreets - For Cyclists, By Cyclists Developer, CycleStreets ** http://www.cyclestreets.net/ ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
[Talk-GB] Estimating coverage
There have been a number of attempts to estimate the level of UK coverage, of varying levels of sophistication, but I've not seen any that compare the length of roads mapped against actual road lengths. The Department for Transport publishes statistics on actual road lengths by local authority here http://www.dft.gov.uk/pgr/statistics/datatablespublications/roadstraffic/roa dlengths/. The number of complete administrative boundaries plotted on OSM has shot up in the last few weeks, so it's now possible to compare actual road lengths (or at least DfT statistics) against the lengths of road that are in the map. Over the last couple of weeks, I've had a first attempt at doing this for about 100 local authorities with decent boundaries. Of the authorities I have managed to measure, the following all show more road mapped than the DfT believes exists: London Borough of Kingston upon Thames 111% Birmingham City Council 109% Rutland County Council 108% London Borough of Richmond upon Thames 107% Portsmouth City Council 106% City of London Corporation 105% London Borough of Waltham Forest 105% London Borough of Merton 105% London Borough of Redbridge 104% Solihull Metropolitan Borough Council 104% Reading Borough Council 104% London Borough of Hounslow 103% Kingston-upon-Hull City Council 103% London Borough of Sutton 103% Isle of Wight Council 103% London Borough of Barnet 102% London Borough of Islington 102% London Borough of Enfield 102% Southend-on-Sea Borough Council 102% London Borough of Brent 100% London Borough of Haringey 100% Given the scope for error in all this, the figures don't look too silly, and at least they suggests a pretty high level of coverage in these places. Eyeballing the map tends to confirm this. My measurements on the following authorities show the map holding less than half the roads that DfT believes exist: Ceredigion County Council 49% Gateshead Metropolitan Borough Council 48% Blackburn with Darwen Borough Council 45% Stoke-on-Trent City Council 45% Middlesbrough Borough Council 44% Rochdale Metropolitan Borough Council 42% Cornwall County Council 42% Bolton Metropolitan Borough Council 41% Sunderland City Council 41% Borough of Telford Wrekin 41% South Tyneside Council 40% Barnsley Metropolitan Borough Council 40% Darlington Borough Council 34% North East Lincolnshire Council 33% Again, there is considerable scope for error, but these at least suggest a low level of coverage in these places. For anyone interested in the technicalities, I am doing this by loading a Planet OSM extract into a Postgis database. There are about 100 authorities where I haven't yet managed to extract a useable boundary, and a number of Counties where the ceremonial boundary that is plotted doesn't match the administrative boundary used by DfT. I hope this proves useful - if only so to suggest where to holiday in order to make the biggest impact. Hint: Cornwall, Cumbria, Norfolk and N. Yorkshire all seem to have a lot of un-mapped roads. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
[Talk-GB] Estimating coverage
Of the authorities I have managed to measure, the following all show more road mapped than the DfT believes exists: There could be a number of reasons for this; 1.Our boundaries are plotted from old parish boundaries on NPE typically. I had to move the Trafford/Manchester boundary in a couple of places because it has obviously been changed when the Metropolitan Borough was set up. Also, NPE has variable accuracy. i.e. we could easily be including roads in OSM's count which actually are in different authorities on the ground. 2.I'm pretty sure the DfT list will include only roads maintainable at public expense or adopted roads. In some areas there are quite a few unadopted roads - which could easily be on OSM. Roads in newly constructed housing estates sometimes take a while for these to become adopted. There are many new housing estates on OSM. In addition, have you included highway=service in your tally? Roads in supermarket car parks, driveways etc. won't be on the DfT list. i.e. we could easily be including roads in OSM's count which are not going to be counted on the DfT list. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Peter Reed wrote: Of the authorities I have managed to measure, the following all show more road mapped than the DfT believes exists: Having mapped every road in Hull (Kingston-upon-Hull since today is a Sunday), some are fairly new and may not appear on the DfT figures. How did you account for dual carriageways? If you counted both carriageways and DfT only counted the road once that might account for some the difference. Some of the dual carriageways in Hull use the dual_carriageway relation, though not all - I confess that I gave up adding it when it seemed to be completely unused. Cheers, Chris ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
On 19 Jul 2009, at 11:54, Chris Hill wrote: Peter Reed wrote: Of the authorities I have managed to measure, the following all show more road mapped than the DfT believes exists: Having mapped every road in Hull (Kingston-upon-Hull since today is a Sunday), some are fairly new and may not appear on the DfT figures. How did you account for dual carriageways? If you counted both carriageways and DfT only counted the road once that might account for some the difference. Some of the dual carriageways in Hull use the dual_carriageway relation, though not all - I confess that I gave up adding it when it seemed to be completely unused. Firstly can I say thank you Peter! This is a great example of how OSM progresses with people popping up with new ideas and innovations where the first you hear of it is when the person has done it. I agree that dual carriageways are a potential source of over-counting. non-adopted roads might be another. Are you clipping roads at the boundary yet? If not there you may be including road sections that are only partly in the county. With regard to dual-carriageway relations, I think it is only a matter of time before they are taken up and then there will be a rush to add more as with the cycle routes and OpenCycleMap. I have added relations for dual-carriageways in my area as well. Could you publish a table of authority boundaries in the UK, their name, their admin-level and if you consider them to be complete or not? There are various manual boundary checks but none of them seem to work all the time and some say things are ok when other ones don't. We still don't understand why Hampshire is not recognised by Geofabrik boundaries viewer for example. I added some more boundaries to the England page today (ie some more I found on the map rather than ones that I added). http://wiki.openstreetmap.org/wiki/WikiProject_England Btw, are people ok if I go through the existing ceremonial boundaries (the ones that are only ceremonial and not administrative) and change there tagging to boundary-ceremonial (rather than boundary=administrative)? A final point. How does one create relations containing relations? There is a relation for 'London Boroughs'. I wondered if we should produced one for 'Regions of England', and 'ceremonial counties of England' and add the appropriate relations to them. Here is the 'London Boroughs' relation as an example. I like the map that is produced from it. http://www.openstreetmap.org/browse/relation/51908 Regards, Peter Cheers, Chris ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Hi, Peter Miller wrote: There is a relation for 'London Boroughs'. I wondered if we should produced one for 'Regions of England', and 'ceremonial counties of England' and add the appropriate relations to them. Generally, relations that just serve the purpose of collecting things are frowned upon. Relations are not meant to be a substitute for categories. http://wiki.openstreetmap.org/wiki/Relations/Relations_are_not_Categories For example, you would not do a relation buildings by Norman Foster because that can be simply done by adding a tag architect=Norman Foster to the buildings. If Regions of England is exactly a collection of relations with a certain admin level and location, then it carries no extra information and should not be created. (Rule of thumb: If you feel the desire to run a script that would automatically add and remove things to/from a relation based on their location and tagging then your relation is probably a collection relation that does not add value.) Having said that, it's all evolution, and if people really feel there are advantages to using relations as collections then there's probably nothing I can do against that ;-) Here is the 'London Boroughs' relation as an example. I like the map that is produced from it. Yes, I have the impression that people often do collection relations because they enjoy being able to simply request a relation/full OSM document from the API and retrieve all the objects, rather than having to find a working XAPI server and formulate a query. However this is *really* something that should be done at search time and not in the database - if we had grouping relations for everything that someone possibly wants so search for... hm, ok, the slippery slope argument doesn't help. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
On the problems with Hampshire - my only related experience is that I had problems with POSTGIS unable to process a number of admin boundaries, because they are plotted with loops in the boundary. I.e. the boundary crosses over itself. This mostly happened where the coast had been added to the relation ,and the coast was plotted at a very detailed level. Far too detailed to see on Potlatch. I solved it by getting POSTGIS to simplify the boundary, before doing the calculations. That effectively removed points that were nearer than about 1 metre apart, and seemed to get rid of the problem. At 1 metre distance I don't think it's losing any precision in the final result. I'm on a pretty steep learning curve here so it may not be the best way of doing it. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Frederik Ramm wrote: Interestingly I joined this list a while ago because I had got my hands on some admin boundary data for England and wanted to know if it was any good (the answer was no). I then forgot to unsubscribe. I'm still planning to extend the Geofabrik excerpts to cover all English counties individually once I have proper data. Unfortunately simply using OSM boundaries only works for landlocked counties; the coastal ones don't seem to include the coastline, and even if they did, a proper coastline is not what you want to use for the excerpt, instead you want to draw a line a few kilometres out to sea where the border meets the coastline, then up/down in a straight line, and back in - which saves computing time and also ensures that any pier etc. that crosses the coastline is also included. I have discovered recently that coastal counties end at the mean low water mark, with a few exceptions. I'm working on improving the boundary to be accurate, starting close to home in East Yorkshire. I need a few visits to the coast first. This will move the boundary, currently roughly along the line of the cliffs or sea wall further out to an estimate of the mean low water mark. It will allow a much needed tag for beaches, mud-flats etc. If you push the boundary further out to sea then any calculation of the area of the county will be unnecessarily large. I do agree that piers need some thought, and off-shore islands too. BTW Frederik, does the (very useful) Geofabrik download for East Yorks use the coastline as its edge? If so when I move the boundary I'll ask you to change the edge for the download. Cheers, Chris ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
FWIW I've now updated http://www.reedhome.org.uk/Documents/LGboundaries.csv to include a list of local authorities in Scotland, Wales and N. Ireland. ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
The way I have handled dual carriageways (and motorways) is to assume that both carriageways are plotted separately on OSM. So the figure that results should be twice the length estimated by DfT. However, for primary roads, DfT themselves show the total length of primary road, and the length of this which is dual carriageway. So by adding their two numbers, I effectively double up their figure for dual carriageways, to reach the same (in principle) as the OSM total. Hence I can ignore dual carriageway tags on the OSM stuff. I think. calculation. Generally the figure for motorway is pretty close. I suspect that significant differences in the motorway figure are down to errors in the boundary position. It should be possible to compare adjacent authorities to see where these have resulted in a motorway appearing in the wrong authority - but this is on my ToDo list for a later stage. The motorway figures aren't *that* close though. Your figures are almost all overstating the DfT's list. I count only a single authority that you've calculated the OSM length less than that of the DfT's. If it was purely due to boundary positions you'd expect that an overstatement in one authority would be balanced by an understating in the neighbouring authority. It does suggest that we're including things the DfT are not e.g. you mention sliproads. What would the figures be without highway=foo_link tags? Also, we might be double counting e.g. Shropshire. DfT list of motorways: 12.4km x 2 = 24.8km. OSM 55km That can't be right however we are counting, unless the OSM figures for Shropshire count that for Telford Wrekin Unitary as well (which you've done separate analysis for). If that's correct then we're still over in both Shropshire County Council area (OSM ~ 29km vs 24.8km), and Telford Wrekin (OSM 26km vs DfT 24.0km) - but it might be close if we remove sliproads? ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
2009/7/19 Frederik Ramm frede...@remote.org: Hi, Peter Miller wrote: There is a relation for 'London Boroughs'. I wondered if we should produced one for 'Regions of England', and 'ceremonial counties of England' and add the appropriate relations to them. Generally, relations that just serve the purpose of collecting things are frowned upon. Relations are not meant to be a substitute for categories. http://wiki.openstreetmap.org/wiki/Relations/Relations_are_not_Categories ... Here is the 'London Boroughs' relation as an example. I like the map that is produced from it. Yes, I have the impression that people often do collection relations because they enjoy being able to simply request a relation/full OSM document from the API and retrieve all the objects, rather than having to find a working XAPI server and formulate a query. However this is *really* something that should be done at search time and not in the database - if we had grouping relations for everything that someone possibly wants so search for... hm, ok, the slippery slope argument doesn't help. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb When I created the relation (end of 2008), I was doing a mass tidyup of London Borough boundaries, I primarily created the relation so I could quickly pull up a neighbouring boundary relation in JOSM when I found another section of it. I was idly wondering if it could be turned into an is_in relation for some point for the Greater London region, even though it is implicit through the Greater London polygon (which may or may not be complete). Thanks for raising my attention to this, since I've now discovered that SteveC deleted the boundary relation for Tower Hamlets in Feb 09 I think it's time for another tidyup session... -- Regards, Thomas Wood (Edgemaster) ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] Estimating coverage
Oh just a thought, does the calculations include toll roads? Do the DfT monitor these in their figures? (M6 toll, Severn Bridge etc.) Yes they do. It's in the notes on the DfT website that private toll roads which form part of major routes are included, but private minor roads are not. Incidentally, I had a go at recreating the figures for Shropshire Telford motorways I referred to earlier, excluding sliproads. I've done these by just making a new way down the centreline of the motorway on a local copy - and just finding the total length. I get 12.4km for Shropshire (spot on with the DfT figures) I get only 9.5km for Telford (DfT 12.0km) - which suggests that perhaps the DfT are including some sliproads for that one, but not for Shropshire - maybe to do with the more complex junction 5 being included in the total - or perhaps each authority submits the figures to the DfT - and Telford has chosen to include sliproads?? ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb