Done All.--Nilfanion (talk) 23:34, 4 March 2011 (UTC)
Hi. Could you make a pin map locator for Isle of Skye and if possible for any of the administraive districts of Highland. The thing is is it such a large area that I think we need better defined locator maps. So long as if has a UK locattor at the side window..Blofeld Dr. (talk) 17:53, 30 January 2011 (UTC)
Isle of Skye is straightforward enough - I'll start work on that later. Some mainland subdivisions are possible too - what exactly would you want? The data means I can do the current wards and simple combinations, such as the current ward management areas. See w:Highland Council wards created in 2007 and File:Highland UK ward map (blank).svg for a description and visual depiction of the wards.--Nilfanion (talk) 11:14, 31 January 2011 (UTC)
Hi. Yes I ideally I want locators for the areas of Highland. The only thing is that they've tended to chop and change historic counties which may confusingly overlap with new ones. I'll notify Ben, who is one of the most active of WP:Scotland and a native of Skye the best way to do this. I'm currently reorganizing the villages on wikipedia and noticed that the highland region is so large that I think we could do with some subdistrict locators.Blofeld Dr. (talk) 00:12, 2 February 2011 (UTC)
Tricky. Using this map Caithness and Sutherland correspond well to the much better known and more useful geographically districts. Likewise Nairn and Badenoch and Strathspey. Lochaber is roughly Caol and Mallaig plus Fort William and Ardnamurchan. Inverness is 13, 18, 20 plus the city. Lochalsh is the problem - it has essentially vanished, so it either has to be recreated or incorporated into Wester Ross. If we are using the Old Highland Region districts, we should do the former and put everything else into Ross and Cromarty. There is a case for Wester Ross and Easter Ross being separate, but I think we need to be consistent. W. L. Tarbert (talk) 09:18, 2 February 2011 (UTC) (aka MacDui) PS Rumours of my Skye nativity are greatly exaggerated.
My concern is we probably should use the current, administrative areas for these maps; as that is the general UK-wide advice for writing about localities and matches better what is done elsewhere - Renfrewshire is split for instance. This also means we can point the legend to a correct place.
I'll start with the ones where the current ward borders essentially match the older districts: Inverness, Caithness, Nairn, Badenoch and Strathspey and Lochaber. Sutherland is currently split in two but can be merged easily, but like you said Wester/Easter Ross, Cromarty and Lochalsh (especially) will be awkward.--Nilfanion (talk) 11:17, 2 February 2011 (UTC)
The problem is that the new boundaries make sense from the point of the view of the boundary commission, but do not geographically and few people would recognise them. It is not ideal of course. W. L. Tarbert (talk) 17:24, 3 February 2011 (UTC)
I agree its not ideal. The problem really is that the only extant boundaries in the Highlands are wards, which councils/the BC fiddle with constantly (the example of Aberchalder on WT:SCOTLAND for one) unlike districts which are rather more static. I can only provide maps using the current boundaries, and I think any political lines on "current" maps - such as location maps - should represent the here-and-now reality not how it was 5, 15 or 70 years ago. This works for some cases, but not all.
One way around this is to not use the boundaries as the basis, they aren't necessary anyway. For example, a map of the Small Isles is easy, even though they have no political distinction from the mainland. If I have a set of maps for the right general areas, the only tricky bit is making sure the location map template calls the right one.
Like I said, I'll get the easy ones up and then we can start worrying about the more complex ones (easier to discuss with concrete examples!).--Nilfanion (talk) 22:00, 3 February 2011 (UTC)
Excellent. These are now in use. Can you make one for Sutherland, Lochaber, Inverness, Badenoch and Strathspey? Mm Ross and Cromarty will be tough as it stretches across. I think it would be better to also make one for Cromarty and one for Rosshire or something to make the areas more specific. I suspect that Sutherland could also be done that way as it is also very big, although of course for consistency sake we also need ones covering the whole areas of both Sutherland and Ross and Cromarty as they are today.Blofeld Dr. (talk) 21:51, 19 February 2011 (UTC)
Nairn and Lochaber are both up now too. Just to point out the base maps are the current wards of the Highland Council, which roughly correspond to the historic district/counties. Inverness and Badenoch/Strathspey are next on the to do list.--Nilfanion (talk) 22:18, 19 February 2011 (UTC)
Super, thanks. WOSlinker is sorting out the infobox UK place to be able to display them.
Note its just the awkward ones left now - Ross, Cromarty and Lochalsh. The reason these are more awkward is Lochalsh is not extant in the present political situation, and is merged with Wester Ross.--Nilfanion (talk) 13:25, 24 February 2011 (UTC)
Can you make the Ross and Cromarty map sometime?Blofeld Dr. (talk) 11:36, 3 March 2011 (UTC)
Added File:Ross and Cromarty UK location map.svg now. That completes the set for the Highlands. However, I strongly suggest that the set is reviewed to ensure they are suitable - as the maps are based on the modern wards not the historic districts after all .Morven, Caithness is an indication they aren't ideal.--Nilfanion (talk) 23:34, 4 March 2011 (UTC)
I have GIS data in a dbf file, as well as the coordinate data in xls format for the core area, buffer zone and boundary. I would like the map to resemble the excellent work you did on the Exmoor map. Maybe a different shade of green for the core area, and buffer zone. I'm working on getting a GA, my progress is here. Cheers, Jack (talk) 18:17, 4 March 2011 (UTC)
Good to know, though I'm not entirely sure what you want from me? Do you want me to create a map (in which case I'll need that data) or to do so yourself? If the latter, what assist do you want: I could do anything from give you advice to creating the base map (so you can then overlay the reserve data). Using two colours for the core and buffer seems sensible to me. One query - what is the source of the data?--Nilfanion (talk) 22:16, 6 March 2011 (UTC)
I've had a go at creating the map, but can't seem understand how all the GIS data would be incorporated into Inkscape and overlayed. I didn't get very far at all to be honest. I'll email you to send you the data and see what you think. The data is sourced from a GIS programme run by the Devon County Council. Cheers, Jack (talk) 11:23, 9 March 2011 (UTC)
Hi. Was just wondering if you could upload a ward map of Brighton & Hove. I'm aware there is already one as part of the East Sussex ward map, but am unsure as to whether a standalone map might be available? Thanks - TheLeftGlove (talk) 21:15, 9 March 2011 (UTC)
Done - File:Brighton and Hove UK ward map (blank).svg. As it happens, the ward map of Brighton was one of my early experiments (and has been on Commons for months), I've updated it to the finalised style used elsewhere.--Nilfanion (talk) 23:53, 10 March 2011 (UTC)
Brilliant, thanks - TheLeftGlove (talk) 21:11, 13 March 2011 (UTC)
Would like a decent UK/ England locators of York, Manchester, Bristol and Birmingham (Unitary Authority locators) replacing the current poor quality pink one in e.g in the infobox.Blofeld Dr. (talk) 19:46, 3 April 2011 (UTC)
The exception is Bristol (as it is a ceremonial county), and I have created File:Bristol UK locator map 2010.svg. Reason I didn't put it in article is I think the pink map may works better as an illustration (despite its bad quality). Creating a district map for Avon and showing Bristol within that would work better than both. That would be consistent with everywhere else.--Nilfanion (talk) 11:19, 4 April 2011 (UTC)
Hi would it be possible to create an additional ward map for West Somerset. There were boundary changes in West Somserset in 2011 so the current File:West Somerset UK ward map 2010 (blank).svg only applies to the elections in 1999, 2003 and 2007. A new ward map for the current 2011 boundaries would be great. Thanks. Davewild (talk) 13:39, 26 November 2011 (UTC)
I would just get on with this, this really ought to be a quick 5 minute job. However, for some bizarre reason the current BoundaryLine shapefiles are in a weird order (to give example the counties are in alphabetic order, apart from Devon which is tacked on the end). The output SVG retains this order, which would be pointless and frustrating for users like yourself. I'm trying to spoof it into behaving.
If you notice any other cases where the current borders are drastically different to the 2010 ones, please let me know ASAP.--Nilfanion (talk) 00:07, 28 November 2011 (UTC)
I've had a quick look using here and, unless I'm missing any, the only other boundary changes all seem to be very small changes (not even enough for the BBC here to say there has been boundary changes), so probably not worth changing the maps for. Davewild (talk) 10:01, 28 November 2011 (UTC)
Ignore my struck comment, I misread this page. The councils with major boundary changes in 2011 seem to be as follows:- Bedford, Central Bedfordshire, Cheshire East, Cheshire West and Cheshire, Stoke on Trent, Mansfield, Northampton, Sedgemoor, South Derbyshire and West Somerset. Wellingborough also had changes but they may have been more minor. Davewild (talk) 20:52, 28 November 2011 (UTC)
Thanks. Davewild (talk) 18:35, 29 November 2011 (UTC)
Done Bedford, Central Bedfordshire, Cheshire East, Cheshire West and Chester, Stoke on Trent, Mansfield, Northampton, Sedgemoor, South Derbyshire.--Nilfanion (talk) 01:27, 7 December 2011 (UTC)
Hi there. I am a big user of your maps and was wondering if it would be possible for you to create a ward map for the former county of Cleveland (made up of Hartlepool, Stockton-on-Tees, Middlesbrough etc). Also similiar maps of Humberside and Avon would be great if it's not too much trouble. Thanks very much, 07bargem (talk)
I can certainly do these, and should have them up in next few weeks. Be aware that the maps will be the current wards, so may be of only limited utility for articles about Cleveland (or Avon or Humberside) when it existed. I doubt they will be much use for election maps for instance.--Nilfanion (talk) 22:51, 17 November 2011 (UTC)
Actually, it's the current wards that I wanted! I'm using the maps to redesign parliamentary constituencies so that's ideal. Also I couldn't help noticing that your Northamptonshire maps have the old wards for Mansfield. Just to let you know. 07bargem (talk)
Fair enough (see above for point re Northants).--Nilfanion (talk) 00:07, 28 November 2011 (UTC)
Thanks. I've also found that Stoke-on-Trent and Bedford have a similar situation. 07bargem (talk) 19:06, 28 November 2011 (UTC)
Not urgent, but given the high density of articles on buildings, parks etc for London we could really use some high quality maps of areas of Greater London as the one generally used is a little too plain. Do you have the ability to make them showing any streets, or at least the major ones, you use ordenance survey or OSM right? I have made some OSM maps but they're not very good quality and not svgs.Blofeld Dr. (talk) 22:34, 19 February 2011 (UTC)
I've been meaning to work on these myself at some point, the blank ward maps in particular. I assume that the insert showing the location of the borough within Greater London would be best? I'll put this one on the todo pile after the Scottish one above.
The main dataset I have been using for the larger areas is the OS Meridian 2, which is potentially less accurate than OSM at the larger scales (especially well mapped areas). OSM will have better data in London I'm sure. However, OS StreetView is much more detailed and better for urban stuff. Both the OS products I've mentioned include all roads: Motorways to B roads and unclassified. I'll pick one and upload that first to allow better discussion - any preference as to which?--Nilfanion (talk) 22:45, 19 February 2011 (UTC)
Depending on the scale and size of the borough, streets and any urban area marking of parks or whatever would be ideal. Ideally I want them like the OSM maps like this but higher quality and which are really good locators which you can see exactly where a landmark is. Yes, I think a London borough locator marker in the corner would be fine, a UK locator would be unnecessary I think, unless you could find a way of showing both (a south england london locator) without it bloating the template. I'd imagine though that just a Greater London area shade would be fine in the corner. Yeah we could also use better quality blank London ward locator maps too which could be done and then used as a the window in the more specific maps. Blofeld Dr. (talk) 22:49, 19 February 2011 (UTC)
I would still like these. Any chance of making them?Blofeld Dr. (talk) 18:08, 2 April 2011 (UTC)
Actually I was just starting to work at this this shows the raw data for the City. One question is what, if any, labels you think should be put on the map?
However, note that my current priority is to get the set of district-level ward maps complete in time for the local elections. As there's no elections in London this time around the capital will be last.--Nilfanion (talk) 19:00, 2 April 2011 (UTC)
I'd say just the neighborhood titles e.g Soho, Knightbridge etc.Blofeld Dr. (talk) 19:44, 3 April 2011 (UTC)
OK, that should be doable - I will prep a label-free version too (when I get round to them), as for example having "Knightsbridge" on the image isn't really appropriate for .--Nilfanion (talk) 10:55, 4 April 2011 (UTC)
I created an article on earlier on Gilgamesh (restaurant). I wondered if you could make a map of Camden as a starter. I'd say a Greater London locator window in the corner ought to do the trick. Obviously we also need a seperate image loaded up that to replace this. Blofeld Dr. (talk) 17:23, 18 April 2011 (UTC)
Yeah I see what you mean, I will get to work on if/when I can. I should have a map for Camden tomorrow.--Nilfanion (talk) 19:39, 1 May 2011 (UTC)
I've uploaded a blank ward map (File:Camden London UK blank ward map.svg) and a test location map (File:Camden London UK location map.svg) for Camden. I say "test" because it shows the available data, but needs many tweaks. There are some minor ones like putting the ward boundaries on top (or skipping entirely), and adding labels, which are easy enough to sort. Two much bigger concerns are getting the colour scheme right and getting a good appearance for the roads.--Nilfanion (talk) 11:56, 3 May 2011 (UTC)
I could try it with OSM-style colours and format, the big problem there is the background colour (OSM uses grey, small-scale WP location maps use beige)--Nilfanion (talk) 12:02, 3 May 2011 (UTC)
Maybe more greyer or earthier tones for the roads on the test map, One thing though can you also add a London england locator in the bottom right of the current window, like the Greater London one? ?Blofeld Dr. (talk) 12:19, 3 May 2011 (UTC)
I'll experiment with alternate colours later. As for insets, I really don't like the idea of having two on a map - a single image can only do so much. I'll start a discussion on the insets at w:Template talk:Infobox UK place.--Nilfanion (talk) 08:32, 7 May 2011 (UTC)
If and when you get chance, it would be great if you could create ward maps for the other London boroughs as you have for File:Camden London UK blank ward map.svg. I would like to create election result maps for London Borough elections derived from these, as I have been doing for some English districts. Davewild (talk) 15:16, 11 June 2011 (UTC)
I'd like to get street-level location maps sorted from this, but I'm not really sure on way forward. I'm open to suggestions.--Nilfanion (talk) 23:08, 28 July 2011 (UTC)
So would I. With the Camden map do you think you could blank out the surrounding area like in File:Barking and Dagenham London UK blank ward map.svg. What I want are maps like Camden but with a defined boundary and the surrounding blanked out.Blofeld Dr. (talk) 18:51, 17 September 2011 (UTC)
I've uploaded File:Camden London UK location map 2.svg as a further demo. This only includes information within Camden - it certainly makes the borough boundary a lot more obvious.
My current thinking on this:
The colours should be somewhat more muted, perhaps even to match the scheme on the county maps. This is to ensure the pushpin is obvious.
The admin borders need to be more prominent - putting them on top and making them thicker should do this.
I don't think the info should stop at the borough border, but it might be good if the outside is fainter - making the area of interest more prominent.--Nilfanion (talk) 22:28, 19 September 2011 (UTC)
Agreed, fainter would be better outside the area rather than blank.Blofeld Dr. (talk) 12:35, 26 September 2011 (UTC)
I've simplified the content slightly, but it works now I think. There are some slight differences at the different scale levels - what works for the City of London won't work for Harrow. The only really noticeable one is that for the very large outer boroughs (eg Bromley, Hillingdon) the unclassified roads are dropped entirely, as is display of individual buildings. At the resolution these are used this makes little practical difference.
Biggest thing missing really is Heathrow, Hounslow doesn't look right without it. Its not in the OS data, so I may just grab it off OSM.--Nilfanion (talk) 23:33, 15 November 2011 (UTC)
Map of Reading unitary authority area or borough to help locate things within Reading. Currently articles located within Reading use a map of Berkshire which is not much help. BaldBoris 10:09, 17 April 2011 (UTC)
I'll get this uploaded tomorrow hopefully.--Nilfanion (talk) 19:39, 1 May 2011 (UTC)
Done Ok - so tomorrow is 6 months later(!) Perhaps I meant mañana? Anyway got there in end, sorry bout the ridiculous delay.--Nilfanion (talk) 01:08, 29 November 2011 (UTC)
Given the shape of Cornwall the scale of the current map is not the best. Could you make maps of the individual districts of Cornwall, especially the Peninsular part, Penzance district etc? Just like you did with the Highland districts, as long as it has an England locator in the corner.. I'll get them then to update the UK infobox.Blofeld Dr. (talk) 12:37, 26 September 2011 (UTC)
I'm not that keen on this idea - both in general for England and specifically for Cornwall. The situation is a bit different to Scotland, where the Highland council area is "too big" and has logical subdivisions.
Established consensus on en.wp has been to use the English ceremonial counties as the background for the locator maps (outside the cities), because that is deemed the "most useful" level of detail for WP. A switch to district-level mapping should probably be discussed first. If consensus is for district-level I can provide that easily enough.
As for Cornwall specifically, once the Isles of Scilly are excluded, it is a unitary authority with no current sub-districts. The former districts like Kerrier have no real meaning today (with exception of Penwith, which makes them even less useful as divisions. This treatment might also irritate the Cornish nationalist POV types unecessarily - showing the location in a defunct area, and in England, but not Cornwall itself. :)--Nilfanion (talk) 21:28, 26 September 2011 (UTC)
OK can you just to one then for the Peninsular then? It would just be very useful to have a better scale map for articles on things like Quoits and dolmens and other buildings. I know you have an interest in Devon and can see why you might not want to give Cornwall any special treatment. But one better scale map of the peninsular would be very useful for landmarks..Blofeld Dr. (talk) 16:26, 29 September 2011 (UTC)
Its not about special treatment for Cornwall. My recent uploads should indicate my interests are on both sides of the Tamar :) My concern is with the basemap being the most informative scale for the subject. IMO location-in-county is the most appropriate scale for general usage on en. Replacement of county-level with district-level may allow greater precision, but loses the context. For instance, w:Lanyon Quoit would be served better by having a map showing it in Cornwall rather than it in the Penwith.
That said, I certainly agree maps of smaller areas will be useful for certain cases. A map of the immediate area is better for w:Stonehenge Landscape than Wiltshire. I have no problem with creating more maps, but have concerns about systematic replacement as it needs to be the right level of detail for the encyclopedia. The National Parks also have maps, as they are an appropriate background for features within them. I'll start a discussion later at the UK place template to see what level of detail should be aimed for in the general.
Question here: What do you mean by "the Peninsular"? There's already maps for Cornwall + Isles of Scilly and mainland Cornwall only - is the second what you want? Ie removing the Isles of Scilly to give more room for the mainland? To me the Cornish peninsula is everything west of the Tamar, so if you mean a subregion of Cornwall you'll have to say more precisely.--Nilfanion (talk)
See this. That's what I meant.Blofeld Dr. (talk) 12:13, 1 October 2011 (UTC)
OK, I can do a map covering similar area to that one, it won't be this week though as I'm going to be away for a bit. I'll get that sorted and then try to finally finish off the London ones ;)--Nilfanion (talk) 21:34, 1 October 2011 (UTC)
(Partially) Done - File:Southwest Cornwall UK location map.svg. How to use the file is a WP editorial concern, its clearly not useless so should be uploaded. Whether it is more appropriate to use that version in place of the county one depends on context (my opinion is default to not, but there are instances where it is likely better).--Nilfanion (talk) 00:33, 13 October 2011 (UTC)
I notice (above) that you intend to map former postal counties. What is the source for those boundaries? Would you be able to consider mapping postcode boundaries? It would be great to have an outline map of all UK postcode areas, and a larger scale map for each Great Britain postcode area showing the boundaries its constituent postcode districts. Great Britain postcode midpoints are provided by OS OpenData, so these could be approximated to boundary lines using smoothed medians. (This approximation might be misleading for properties near boundaries that have postcode-points and different distances from each border but should be a fair representation at area and district level when appropriately confined within coastal boundaries.) Outside Great Britain, the BT postcode area simply corresponds to all of Northern Ireland and it may be that there are no open source postcode district boundaries. Similarly, GY/IM/JE corresponding to each of the Crown Dependencies. Apologies if I've previously made a similar request, as I appreciate that there are many competing priorities! — Richardguk (talk) 23:30, 16 October 2010 (UTC)
Well GY/IM/JE (and BT) fall outside scope of OS source, I need to further info sources. The former postal counties will be tricky of course, getting (N/S) Humberside and Avon are highest on list for obvious reasons. I agree postal areas and districts (sectors would be excessive(!)) would be good. Looking at the Opendata info to see if I can do stuff with it; I'm not sure I can do much without having to deal with RM's copyright... OS has revised the Boundary Line product - I'm just checking diffs there before I do anything else.--Nilfanion (talk) 23:40, 16 October 2010 (UTC)
If you can derive approximate postcode district boundaries from the postcode points, nearly all the 1990s postal counties are amalgamations of these – specifically, they were amalgamations of post town boundaries, but most districts split across post towns were at least in the same postal county, except I think for Warwickshire / West Midlands (CV8), Norfolk / Suffolk (IP21), Leicestershire / Rutland [pre-1974 county reinstated to reflect unitary area] (LE15, LE16), Surrey / West Sussex (RH6), and Berwickshire / Northumberland (TD12). The correlation between district and county is listed at w:List of postcode districts in the United Kingdom. — Richardguk (talk) 00:13, 17 October 2010 (UTC)
Just had a thought. The 2001 census output areas were, I think, amalgamations of unit postcodes, subject to census ward boundaries. These boundaries are available "free at the point of use". If an output area contains postcode points of only one postcode district, it would be reasonable to assume that the entire OA is within the district. This might help to refine the median-line approximations to precise street boundaries, at least where OA boundaries coincide with postcode area/district boundaries. — Richardguk (talk) 00:28, 17 October 2010 (UTC)
The Code-Point data will be a pain to convert, but I'll probably be able to do so. My biggest concern is converting the points to polygons will inevitably involve original research. The boundary is what Royal Mail says it is, which may or may not be median lines. Those lines definitely are not the district borders in all cases, for instance in the region of SX4164 - the postal districts will respect the county boundary (and the river), but the halfway line definitely cuts the corner and so would put some of Devon in a Cornish postcode district. That problem is small and will only be a few hundred metres out at most (which wont matter on a large scale map), but there are more serious ones which would show up on postal area-level maps. Where is the EX/PL boundary between Okehampton and Princetown? There's a >16km gap with no postal addresses there, where is the line drawn?
As for the census material I'm not sure if it is free enough, as that page states using the vector info commercially requires a OS license. I am going to contact the ONS to clarify, as they have a bunch of other data such as the TTWAs that I don't have already, which look like they may be free under the Opendata agreement. That includes OAs so I may be able to construct postal areas..--Nilfanion (talk) 10:47, 17 October 2010 (UTC)
I was hoping you could somehow make a map of the postcodes. I'm really interested in surnames/given names, and I'd love to be able to fiddle with adding distribution maps to name articles. This website uses postcodes, from I think about 2008. Here's the postcode map they use. [1]. If you just do a quick search for any surname you can see how the represent the data. It'd be cool if we had an svg template to make maps like that, it's something that just can't be described in words.--Brianann MacAmhlaidh (talk) 10:56, 6 March 2011 (UTC)
Comment The free Code-Point Open only provides point data for each postcode unit (the non-free Code-Point with polygons includes polygonal information as well as more metadata). I'm exploring that data at present to work out how to use it. Converting the point data to polygons looks like a pain (which is why the more useful product isn't free...), just creating polygons for the postcode areas will likely be extremely time consuming; and I'd like to be able to get plot down to district level.--Nilfanion (talk) 22:16, 6 March 2011 (UTC)
I've uploaded a blank map of Shetland showing the postal districts (ZE1, ZE2, ZE3) as a test/demonstrator (follow the instructions here to highlight the individual districts. I'm giving myself a crash course in GIS editing at present, so should have a area-level (then district, then sector...) map for GB in the next few months.--Nilfanion (talk) 12:32, 20 March 2011 (UTC)
I've uploaded the one the right to show the IV area. Primary purpose is to present the underlying data - I'm concerned that the map includes original research on my part: I had to draw boundary lines that "look right", around the supplied data points...--Nilfanion (talk) 14:21, 29 March 2011 (UTC)
I think that as long as the description notes that certain boundaries are reasonable approximations, it should be OK. Most maps are approximations. If the Royal Mail refuses to release the data you need, I don't suppose there will ever be totally accurate current maps that are free. What advantage does it give them to hoard the data of boundaries? I understand you are keen on accurate data in your other maps, I guess postcode maps will never be as accurate. It sure looks good. It's your call though. This webpage came up in a Google search, but it just covers the double-digit codes [2]. Oh look at this page, look at the "Scotland Postcode Districts" over-lay they have there [3].--Brianann MacAmhlaidh (talk) 09:48, 1 April 2011 (UTC)
That website is using the same data as me, and predictably has different lines. Every map I've inspected has slightly different borders. The easiest bit to compare is the KW/IV boundary, as you can compare the area maps too. ::I think I am being overly paranoid here, as RM assign delivery points to postcodes, not geographic areas to postcodes. For instance, if you ask Royal Mail what the postcode is for the summit of Ben Nevis, you won't be told "PH33 6SY" but "there isn't one, we don't deliver there!".
As long as description says the borders are represenative only and not definitive, and every postcode (apart from clear exclaves) are in the correct region, they should be ok. In more built up areas, this isn't a problem as the postcode density is much higher meaning the lines will be more precise.--Nilfanion (talk) 10:37, 1 April 2011 (UTC)
Comment I'm slowly getting this sorted - gradually prepping a set of boundaries (I've got the Scottish postcodes "ready" to district level). I am tempted to look at Royal Mail data directly on this...--Nilfanion (talk) 22:58, 10 September 2011 (UTC)
British postcode areas map.svg Only just noticed your comment above, so have inadvertently been duplicating some of your work, using Voronoi polygons to convert OS Code-Point Open unit postcode median points to postcode area, district and post town polygons. (Learning how to use QGIS and Inkscape with Scour, plus mapshaper.org to simplify polygons without losing contiguity.)
My first upload is a labelled postcode area map for the entirety of the UK and Crown dependencies: File:British postcode areas map.svg(right).
Given the vastly differing sizes of the areas, there is a legibility trade-off between small labels versus disruptive inset maps (I've tried to keep the labels large where space permits), as well as a tension between label legibility and boundary-line distinctiveness (hence the very pale red lines to avoid blurring into the black labels). As you can see, in the national map, I've used untitled insets for the areas around Manchester and London, and variable-sized labels, to maximise readability, though I'm not sure whether that makes the map less coherent than a uniform but small font, so I'd be interested in your point of view.
I propose next to produce a map of each area, with postcode district and post town boundaries and labels in a broadly similar style to the national map, with urban areas shaded as in your existing mid-scale maps. In areas with extremes of population density (such as AB), inset maps will probably be necessary to show city-centre districts without making the rural districts too large.
I aim to stick as closely as possible to the style of your OpenData locator maps, eg showing urban areas as shaded polygons in larger scale maps, though I think road and rail networks would be too confusing to overlay, as would administrative boundaries (particularly since the postcode boundaries are only derived from mid-points so could misleadingly appear to deviate from actual common boundaries – though there is an existing dual-themed example at File:Map of the administrative geography of the British Isles with postcode areas.jpg). At present, I think postal counties are not worth indicating, as they are obsolescent and would conflict with post town labels.
Your existing maps are very impressive, so let me know if you've any thoughts about the above direction, or indeed whether you are already in a position to upload your own maps of each area – in which case, I apologise for treading on your toes!
Also, any feedback on the commons description text would be welcome, as the licensing and attribution requirements are a bit confusing. I've used an equirectangular projection, expressed on the description page in terms of a "latitude of true scale" (choosing 55N for the national map and its insets), as this can directly be specified as a projection in QGIS whereas a stretching factor can only be specified in terms of its arcsecant (I think lat_ts=55 implies stretching of about 174%). But the boundary bearings are exact, which I assume is the key requirement for compatibility with locator templates.
Don't worry about treading on toes here either. I'd prefer to use Royal Mail defined polygons to "best guess" voronois, I'm pretty sure they exist but aren't free, so our own will have to do. (They won't be identical for two reasons - first is we only have unit-level data, not delivery points, using the full info will provide different polygons. The second is PCs are defined with reference to local geography, not ignoring it - major rivers meander and the boundary will meander with them :) ) More practically, when making the polys one potential issue is RM sites such as mail centres - there's typically a bunch of postcodes for that spot which may distort things, especially on large scale maps of city centres.
Specific feedback on the UK map:
The single most useful modification you could make is to have one polygon for each postal area, and have these clearly labelled (via id tags). This makes derivative works trivial: File:English ceremonial counties 2010.svg to File:Lancashire UK locator map 2010.svg takes a few seconds in a text editor, highlighting the AB area from the Postal map isn't possible as there's no one polygon representing it.
There's a line around the edge of the map. I'm not sure what the source of it is, but you should get rid of it.
The coastline should have a thin line representing it (#0978ab).
Not sure about the border colour, the standard #646464 is probably better. I realise the desire to separate from the text labels, but if you can't see the boundaries the areas aren't that clear
Some of those labels a bit overwhelming: I'd pick a size and go no larger than that. If intented to display at scale of image to right, probably about the size of the "TQ" label looks about right. Smaller labels to suit small areas are OK.
Insets might need more careful treatment, its not that clear what bit is expanded. I know what each shows, but at a glance its not obvious that the top inset is from the centre of the map and the bottom one is from the SE.
The "stretch" is irelevant for the purposes of a location map template: What that needs is the lat/long coords of the 4 corners, and equirectangular projection (so lat is horizontal and long vertical at all points). For consistency, I'd suggest that you choose corner locations to match File:United Kingdom location map.svg. Ideally the "stretch" too - to further increase consistency and it makes derivatives slightly easier again.--Nilfanion (talk) 22:59, 25 November 2011 (UTC)
Great feedback, all good points.
QGIS doesn't seem to allow id attributes in its SVGs. For the national map, I could add these manually (as was done with the clipping elements in the initial upload). But instead of exporting maps as SVG using the standard QGIS composer and then customising the output, a more radical approach would be to compose the SVG file in a very precise way using SQL to customise the creation of each element. I'll try this method and, if it works, it would also automate most of the production of the area maps, making further amendments very easy to apply consistently.
With the border colour, I was hoping red would look a bit more "postal" than grey (like a post box!), to distinguish the maps at a glance from similar scale maps of administrative boundaries. But since red has poor contrast with black label text, I tinted the red to pink, as shown, so perhaps standard grey is preferable to such a compromise. The situation is more complicated for area maps, which I propose should show boundaries of areas, districts and post towns (but not postal counties); as well as coastline, urban area shading and probably rivers and lakes for context. This colour scheme may affect the best choice for area borders on the national map. Will experiment and upload an area for comparison.
If necessary - the id elements could be added in post-processing in Inkscape, as long as you have the right polys to attach them to. Your map does have polygons for the individual districts in London area for instance, so should be feasible to add them for the others. Ideal is if you can get it sorted on the GIS-side of course, that way you don't have to figure if path #235 is AB32 or PH24 :)
Your plan for area maps sounds sensible - I've been using Meridian to provide the mid-scale data. Once you get results will be able to see what works I guess.
And incidentally, I'm happy to see someone else taking interest in UK maps - trying to do thousands of them is just not feasible(!)...--Nilfanion (talk) 11:34, 26 November 2011 (UTC)
As an aside from postcode mapping, congratulations on your brilliant work with the Land-Form Panorama relief location maps. Out of interest, did you choose JPEG over PNG format solely because of compression efficiency?
Meanwhile, I was wondering whether you've seen use made of coordinate metadata in SVG files? I've seen microformat enthusiasts on Wikipedia for article text. But image metadata (other than standard EXIF resources) seems confined to loosely-structured text on the description page. I was considering including metadata in the imminent postcode SVGs, as it seems simple in principle to specify datum and bounds in the file itself as well. But the W3C explanation is a bit complicated for me to apply <crs:CoordinateReferenceSystem svg:transform> with confidence, so I'm currently unwilling to risk getting it wrong.
Incidentally, my interpretation of this is that Royal Mail, at some level, are themselves relying on maps which, in some ways, are not much more refined than the Voronoi polygons we're generating from OpenData! It's unclear whether they're referencing postcode points or individual delivery points, but the illustration appears to show only the former – though of course they're working with sectors rather than districts. My working assumption is that sectors change too frequently and are of insufficient interest to be worth documenting and maintaining on enwiki/Commons.
Whilst accepting your caveats on Voronoi derivations, I would add that, for places (such as parts of Dartmoor) that are permanently devoid of deliverable addresses, postal boundaries are not very meaningful even (or especially) in the hands of Royal Mail! For inland areas, interpolation is a least-worst necessity to avoid gaps. But I propose to remove from the postcode SVGs extrapolated postcode boundaries for unpostcoded remote islands (more than 4km from postcoded islands and the mainland), i.e. those that have no postcode median-points mapped to them. The anomalous island boundary lines are just visible north of Scotland and west of Wales in the current upload, and would have been even more confusing in the area-level maps. An arbitrary 2km coastal buffer simplifies the seaward boundary for the vast majority of islands and islets, with results that look reasonable. Remoted inhabited places such as Fair Isle, Lundy and the principal Isles of Scilly have postcodes of their own, so can be shown in the relevant areas.
I hope to upload a revised version of File:British postcode areas map.svg in the next day or so. Have been busy cogitating and data cleansing to ease production of the subsequent area-level district maps.
Re the relief stuff, thanks. Smaller filesize is only real reason I used jpg, and only noticeable problem is the red bit on the insets. There's a number of things I'd like to improve on them still, such as I'd like an svg set ultimately, but ones like File:Dartmoor topographic map.svg are prohbitively expensive (need to patch the vectors). I might write up a personal wishlist somewhere for things like that.
The pink boundary lines (the new ones) on the map you linked look like they utilise delivery points - as it seems to follow property boundaries. Of course, at the scale map that you will produce, those differences will be irrelevant. I don't think there are any units with widely seperated delivery points, as that defeats the object for RM, who have no issue with single-point units for farms.
Agreed on concept of postcode boundaries in empty areas - if you asked RM what's the postcode for the Cranmere Pool letter box was the response would be "who cares?" :) They aren't going to be waste their time pondering hypotheticals like "if someone built a house at the top of Ben Nevis, which postal sector will it get assigned to?" Their attitude there will be similar to ours, an arbitrary choice to aid clarity of the bits that matter (the populated bits) on small scale maps. The 2km threshold for islands seems reasonable too, the few outliers look odd.
I can think of one example which will show up on small scale maps and I'll try to check against RM-approved mapping. How is the Glen Affric YHA handled? Nearest postcode is IV637YW which is Cluanie to the south, however it "should" be in the same district as Cannich (IV4). If it has a real postbox at the foot of the glen, that box ought to have a IV4 postcode.--Nilfanion (talk) 12:06, 30 November 2011 (UTC)
Looking more closely at the Geoplan map, they've wrongly divided the L-shaped Oliver Road between sectors SM1_3 and SM1_4; in fact, their boundary line should be further to the north east as all the addresses are in SM1_4QF (the postcode point next to the "V" of "Oliver" is correctly mapped within SM1_4). Ironically, the Voronoi boundary (shown in a screengrab from my sector-level polygons at File:SM1 3-SM1 4.png) is postally correct but visually ugly around this point, because of the two postcodes which cause a tied island at the south of Benhill Road; Benhill Road itself gets wrongly divided by the Voronoi near its junction with Oliver Road. The unoccupied recreation ground causes a predictable but postally harmless divergence. Extrapolating wildly, this seems to demonstrate how the Voronois are ugly but fairly accurate for densely populated areas.
NH079202 / 57.232,-5.1834 which is indeed within the IV63 polygon instead of IV4, though near the boundary: File:GlenAffricPostcodeDistricts.png (this is a screengrab, not an exact prototype of the area-level maps I'm working on). At least the Strategi urban area shading (which include rural settlements) will implicitly indicate which places are sparsely populated, and hence more vulnerable to misallocation at the margins using the "as the crow flies" boundaries.
the grey lines for the postcode area boundaries are thinner than on existing UK location maps;
I've provisionally used a shade of red (bold with a 1/30 white outline stroke) for the label text; you might want to compare with black text.
Geographic extents for the main map now match the standard UK location map.
I've customised the SVG IDs, but for cross-file consistency the main map now uses the coastlines, lakes and Ireland border from File:United Kingdom location map.svg, instead of from VMAP0 and Strategi, so the first parts of these two files are now byte-for-byte identical, except for deletion of internal political borders and changing Crown dependency land from grey to yellow.
Let me know your thoughts so I can take account of the feedback when prototyping an area-level map. Cheers!
Looks pretty good now :) Only two things I can think of that I'm not totally sure about:
The treatment of coastal borders. I don't think its necessary (or pretty) to have a parallel line slightly off the coast. It would probably be better to include the marine borders between areas only, so only the perpendiculars outwards. Or even just cropping at the coast. The first may make more sense - as that would make it clearer which islands are IV, PH or PA. As for GY, JE and IM putting partial lines at about the mid-points File:United Kingdom location map.svg would be clearer.
The insets, or more accurately the grey rectangles they come from, dominate smaller thumbnails. I'd suggest using a more transparent fill (or perhaps completely so) and adding a thicker border. Giving the two insets labels might be logical too. eg putting "See Inset 1" in the grey rectangle of the London area.--Nilfanion (talk) 22:24, 3 December 2011 (UTC)
Great feedback as always, fair points.
Since offshore locations are even less likely to be assigned postcodes than Dartmoor, I can't deny the logic of your point about coastal borders (though the appearance varies greatly with scale). Given your previous suggestion to have identifiable path elements for each area in the SVG text (for ease of adaptation by others), I propose to retain the current area polygons, including the offshore parts, but to overlay the sea (and inland water with the area-level maps) on top, as an opaque blue-filled polygon, thus obscuring the offshore boundaries. Though this makes the file size larger than if the obscured lines were deleted, it retains the underlying information and retains flexibility for polygon shading etc. I think water-over-land layering is a technique you've used in your own boundary-SVGs, but let me know if this change of approach sounds reasonable. The downside would be that each island would no longer be a separate element, but a hole within a water element (either the sea element or, at area level, an inland water element).
- As a side-effect to the above layering change, I would no longer be able to use the File:United Kingdom location map.svg file ("UKLM") as a base layer. Instead of trying to import UKLM's paths for inverting land to sea, I propose to revert to generating the coastline from VMAP0 and OS Strategi data (as with the first uploaded version, but retaining the compatible geographic extents of the current version). The coverage and proportions would be the same and the coastine would look the same to casual viewers, but there would be greater consistency between the national and area-level maps, and I would gain control over the markup formatting.
- As a secondary side-effect, I further propose to use an exact aspect ratio for the nominal SVG image dimensions. UKLM seems to have been created in Adobe Illustrator and has non-integer dimensions which mean that image-size calculations and aspect ratios aren't exact. Currently, 885.546 × 1,368.581 in the SVG attributes is interpreted as 886 × 1,369 by mediawiki. It would be better to choose scalings for the national and area-level maps that result in exact integer dimensions. The national file has 170% stretching and covers 13.2 × 12 degrees, so dimensions should be in a ratio of 132:204, eg 1,056 × 1,632. I believe that, for thumbnailed wiki images, the width of images is the sole factor when rendering in articles, so for consistent path thickness the nominal image width needs to be roughly constant. Is a width of around 1,000 units a good target? Or should account be taken of the height too when deciding the nominal scale, so that narrow areas don't have much thinner lines if rendered with less than the default width? I'd be interested to know the approach you took when scaling UK location maps of differing aspect ratios.
For the inset-coverage shading, before uploading I'd experimented with 10% opacity which was barely noticeable but the current 25% does look too dark so will experiment further with an intermediate value. I'm reluctant to add descriptive text to the insets, in case it is misinterpreted as part of the postcoding nomenclature; "LONDON" would be apposite for the southern inset but there's no way to label the northern inset without introducing arbitrary or misleading descriptions, and numbering the insets (as suggested) introduces a non-postal concept and also English-language dependency. Perhaps distinctive coloured shading for each, with the inset maps shaded slightly to match non-grey shading of the corresponding rectangles on the main map?
Meanwhile, I've noticed that mediawiki renders the Arial label text as Liberation Sans, so I'll set font-family to "Liberation Sans,Arial,sans-serif" as recommended.
Finally, I'm increasingly tempted to add some coordinate reference system metadata elements (see 2nd bullet of 02:05, 30 November 2011, above). Puzzled that the SVG1.1 standard for this seems largely unimplemented (on Wikipedia/Commons and more generally online). I'll see if I can throw in the appropriate markup, in case it's one day of use.
I've made most of the changes discussed, but retaining the File:United Kingdom location map.svg elements as the base layers, for visual consistency with the other series of UK location maps. To hide the offshore area boundaries, I defined a clip-path using the Great Britain mainland element (it's not possible to clip against group elements but eventually I realised that none of the area boundaries cross non-mainland islands so a single clipping shape is sufficient). Accordingly, even though the offshore boundaries are now invisible, the polygon data is still present and adaptable for reuse (with the 2km fringe as before). Ad hoc dashed lines now clarify coastal boundaries in the Hebrides and Channel Islands.
Instead of language-dependent labels or coloured tints to contextualise the two insets, I've experimented with a 5% shadow connecting each inset with its coverage area.
Your feedback welcome as always.
— Richardguk (talk) 06:28, 28 January 2012 (UTC)
Can't think of much more to say now on this; everything works pretty well. Boundary lines work for me, as do the insets. The only other stuff I can think of is some niggles with the labels. This is based on the nominal size of the SVG and default thumbnail size, where the text is just about legible. At higher res there's no issue, and article thumbnail the labels will never be readable but work.
The BT label looks offset to the top, a more natural location IMO is west of Lough Neagh.
Some tweaking of the Liverpool/Manchester inset - to get the vertical of the L off the coast line, M and HD look off centre (should move up and move up and left respectively), as well as adjusting sizes in a similar manner to the London inset.
Coastal postcodes in general: There are several labels which cross over the coast, affecting readability. Sometimes this is minor (KY), sometimes more serious (BH, SS). Adjusting label placement to minimize clashes would be good. In some cases, might be an idea to put the label a distance out to sea and connecting to the land with a line; would help for ones like SR and SS. Overdoing that would be counterproductive.
On similar vein, the label for PA would be clearer if the Scottish lochs were dropped. I dislike the loch selection on the UK location maps in general, it seems too arbitrary
Offshore islands: Lundy doesn't need a label (EX is implicit IMO), ditto the Isles of Scilly with TR. Fair Isle's ZE could be dropped if a marine boundary line was added, like for the Channel Islands. The KA label Isle of Arran is a maybe.
Sorry for delay in reply (been distracted by categorisation...)--Nilfanion (talk) 00:12, 5 February 2012 (UTC)
Worth waiting for, thanks.
Pleased that the unconventional way of contextualising the two inset maps is working out OK.
Will seek to refine the label positioning as suggested. Also, most SVG renderers seem not to render well the white strokes of the glyph outlines (applied to aid legibility of the red text against dark background graphics), so might try rendering the letters twice (with a use element): placing the white outline-stroked bold text under a layer of unstroked non-bold red text.
Agree that, with the introduction of dashed lines for ambiguous island borders, the secondary area labels can be dropped. To compensate, and for consistency, I'm minded to introduce additional dashed lines: for EX–SA (to clarify Lundy), KW–ZE (to clarify Fair Isle), and possibly PO–SO (to clarify the Isle of Wight, which would be consistent with the other islands, but the Solent might well be too narrow to justify this particular addition and the existing PO label already covers the island and mainland).
As to the loch rendering and basic land shapes: Despite my original musings, now that you've changed my mind I'm minded to continue using the exact paths and element structure of United Kingdom location map.svg for 100% visual consistency, even though this limits flexibility. But if you wanted to introduce an improved version of that SVG, I'd be happy to adopt it (in which case, if it were a major revision, it would be handy to have the sea overlaying the land, to avoid the issue discussed below, as well as having meaningful IDs for each element).
The current methodology will produce a problem when adapting the labelled national area map to produce unlabelled locator maps (showing each area highlighted in its national context, similar to Northern Ireland in United Kingdom.svg). I think these will be necessary for {{Infobox UK postcode area}} and should be consistent with the standard national map. (These would be in addition to the detailed maps of districts within each area, which I intend to derive solely from OS OpenData as previously discussed – already the source for the coastal paths on the two inset maps). The workaround I'm now using for clipping borders to land only works for clipping to one basic shape element (as required by the various SVG specs and most renderer implementations), i.e. mainland Britain. So, as currently structured, a locator map of ME with the ME_areapath element highlighted would fail to highlight the part within the Isle of Sheppey (or, if the offshore island were separately highlighted with a fill property, the mainland part of ME would overlay the interior half of the mainland coastline stroke, as clipping works along the centreline, introducing a slight inconsistency for the coastline width of areas that are not wholly within mainland Britain).
Apologies for cluttering your userspace with my ongoing thoughts. If you prefer, feel free to move the above dialogue to a new page or File talk:British postcode areas map.svg (which currently cross-references this subsection).
Hi, could I request regional ward maps for 2012 in much the same way you've done for 2011 - I imagine much is the same, beyond the likes of Hartlepool, Swindon, Broxbourne, Daventry, Rugby and Rushmoor which have undergone boundary changes to prompt all-out elections, and Epping Forest which has had minor changes? I've not yet finished colouring your wonderful 2011 maps, but thought I'd ask early. Thanks once again. HeadlightMorning (talk) 02:16, 6 May 2012 (UTC)
OK, this one is easy enough to answer - no, unfortunately I can't at this time. Problem is OS hasn't released a new boundary set since October which contains the old boundaries. When new data becomes available I'll make the necessary updates.--Nilfanion (talk) 10:01, 6 May 2012 (UTC)
The most recent version of Boundary-Line now includes the 2012 datasets, so I should be able to tackle this in the near future. I'll redo the 6 districts you mention above before doing rest - though I won't do any with no changes or only minor tweaks barely perceptible at the map's native scale.--Nilfanion (talk) 21:46, 29 June 2012 (UTC)
Done The 6 above now.--Nilfanion (talk) 19:25, 7 July 2012 (UTC)
Done File:Crawley UK location map.svg--Nilfanion (talk) 20:48, 26 September 2010 (UTC)
Comment I've uploaded blank maps for the City of London and all the boroughs ((City of London, Barking and Dagenham, Barnet, Bexley, Brent, Bromley, Camden, Croydon, Ealing, Enfield, Greenwich, Hackney, Hammersmith and Fulham, Haringey, Harrow, Havering, Hillingdon, Hounslow, Islington, Kensington and Chelsea, Kingston upon Thames,
Lambeth, Lewisham, Merton, Newham, Redbridge, Richmond upon Thames, Southwark, Sutton, Tower Hamlets, Waltham Forest, Wandsworth and Westminster).
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
.svg){kind=link}
