Mapping dirty London

In the past couple of months I’ve been able to combine work and my mapping hobby, working on a web site about air pollution in London. I’m going to be speaking about this at the October geomob meeting.

I’m lucky enough to live in one of Europe’s most polluted cities. Air pollution causes more early deaths than obesity and road collisions, and is only bested by smoking. The Mayor published some really good open data on pollution levels, which of course is incomprehensible to ordinary folk. So despite having a sense that it’s not the cleanest city, Londoners don’t know all that much about the problem or how it could be solved. We want to help change that.

Our first splash was a map showing the quantities of some major pollutants dropped on sections of roads across the capital, so Londoners could find out – how polluted is my road?

Mapping dirty London

Lots of people loved that. The GLA’s GIS team did the mapping part, using our Ordnance Survey license and data to match pollution data up to ITN road sections. They also produced league tables for each borough, which we sent round to all the local papers. The Guardian featured it on their homepage and TimeOut blogged about it, driving many thousands to have a look.

Next, I checked a list of schools known to be within 150m of heavily polluted roads – there being strong scientific research to suggest a link between that proximity to pollution and higher rates of asthma in children. Currently there are estimated to be 1,148 schools suffering from this problem., revealed through fantastic work by the Campaign for Clean Air in London. We’ve mapped these, so you can see if your school is affected. This was really easy – turn the schools into GeoJSON and stick them into a Leaflet map, using the markercluster plugin to make it usable when zoomed out.


That wasn’t very difficult, but I think the map tells the story well – that this problem affects schools all over London, not just in the centre.

I’ve now been able to do some of the GIS work myself, and what fun it was! I’ve never had much call to really use Quantum GIS, but it’s a wonderful tool.

I was able to take raster files showing nitrogen dioxide concentrations across London from the London Atmospheric Emissions Inventory 2010, vectorise them, and filter them to find areas where levels were above legal limits. With this, I can then play around with other open data to see what lurks in areas suffering from illegally high levels of air pollution.

Mapping dirty London
Areas of London expected to exceed legal limits for the annual average concentration of nitrogen dioxide in 2020. For context, London was supposed to be under these limits in 2011 in order to comply with a European Directive introduced a decade ago.

My first experiment was to clip this to London’s road network. I used the Overpass API to extract all the roads from OpenStreetMap (for some reason I can’t connect to OSM-GB at work of late). From this I was able to determine that in 2020, around 45% of London’s main road network is still expected to exceed legal limits. Nasty!

I was also able to determine that in seven years time, there will still be 928 schools near to heavily polluted roads. So thousands of young Londoners will spend their whole time in primary school breathing in illegally high levels of air pollution.

I then started to think: where do I go that means I’m next to main roads for long periods of time? Pubs, cafes, bus stops, parks. Well, these are all in OpenStreetMap as well!

I started with bus stops, because we have a pretty comprehensive dataset there after the NAPTAN import. I did all the GIS analysis, producing tables of data for boroughs and the like. But it was only when I used Maperitive to produce tiles for a slippy map that it struck me – there are still LOTS of duplicate nodes where someone has manually added the bus stop years ago, then we imported the NAPTAN stop. So actually OpenStreetMap is a completely useless source for bus stops.

I got around this by just downloading the original NAPTAN data and using that instead. But it’s a shame because NAPTAN is really inaccurate. Where OpenStreetMappers have added bus stops, or manually checked NAPTAN stops, the locations are much more precise. It would be great if we could try to clean this data up to remove duplicates. Perhaps over the winter meetups, Harry?

With this, I produced a snazzy web page showing info on London in 2020.


I haven’t tried pubs and cafes because our coverage is so patchy. One day there may be enough contributors for OpenStreetMap to have a really excellent geodatabase of these features. What an amazing resource that will be! Though I wouldn’t want to be put off some of my favourite pubs.

One final step I didn’t take was routing. I’d really like to see somebody integrate the pollution data with a routing engine, to try and find reasonably direct walking and cycling routes that keep you off the most polluted roads. I blogged about this last year, and I still think it would be both cool and genuinely useful.

My friend Robert also suggested a routing engine where the polluted roads are off-limits, and tiles without those roads drawn. Getting around today without using those roads at all would make for an interesting challenge!

All of this work has had quite an impact. Take this cutting from my local paper:

It was also the top story on BBC London News for the whole of Wednesday when Jenny Jones AM questioned the Mayor about our findings:

Now we just need to fix the problem.


  1. dancarins said:

    Brilliant work, Tom.

    12th September 2013
    • Tom Chance said:

      Thanks, it would be good if similar data were released for other cities like Manchester and Birmingham so people there could do the same analysis.

      13th September 2013
  2. This is a very nice piece of work. For the past couple of years I’ve tried to demonstrate that OSM could be used for analytical purposes. In this case you have actually tried to use it for a genuine analysis project, and the absence of complete and comprehensive data hinders this. We probably only need to be 80-90% complete but coverage does need to be comprehensive.

    It’s interesting that you cite problems with non-deduping of NaPTAN stops in OSM: for most purposes (routing, finding the nearest bus stop, setting proximity alarms so as not to miss an unfamiliar stop) the data is ‘good enough’. Trying to use data for analysis places a different kind of bar with respect to quality and highlights areas we need to work on. It is therefore not just useful in its own right, but should be another driver for improving the map.

    Lastly, you could use FHRS Open Data to source cafes/restaurants or pubs/bars. In the latter case coverage is not complete (not all pubs serve food), but is pretty comprehensive (most London Boroughs participate in the scheme).

    13th September 2013
  3. Jan Eek said:

    We actually do it here in my city, Bergen, Norway, and the measuring results are NOT good.
    Especially under a special weather condition, people with asthma, kols, heart condition or reduced immune system are adviced to stay inside in several places in the city.

    Beside being a member of the Green party, I am also member of a local party that focuses only on the airpollution. We managed to get one of us into the City Counsil and that has given some results.

    The cars are of course the worst polluter, BUT, we are a port city sourrounded by seven mountains, so we have the ships. There is a lot of traffic in that narrow port; supportship for the oil business, fastgoing vessels for transporting people along the coast, and then we have the cruise ships. We are one of the busiest port for cruise ships in Europe (“The gate way to the fjords”) and this year we had/have 260 visits from cruise ships.

    So, we are fighting a war on two fronts: The cars, but not least, the pollution from the ships. Until we brought it up, nobody was even thinking of the pollution from the ships. Without going into too much detail on the chemistry, the toxic fumes from the huge diesel engines from the ships are polluting much more than the cars. And the particles are so small that you can’t see them and they are more dangerous.

    The University hospital in Bergen made an assessment in 2009, and conluded that at least 143 people died “too early” because of the pollution. It really is a serious problem and the UK of course have their full share of ports.

    The solution is so simple: Electricity from land. Many cities in Europe have established that and we have had a hard fight for two years with the Counsil, and now, finally, we are going to set up a “plant” with the proper equipment to supply ships with electricity to ships.

    We are far away from covering the whole port, but it is a start and when the cost/benefit analysis we have presented finally gets through, things may speed up.

    So, the cars……The mountains are of course a problem, BUT, it is also a blessing. TUNNELS! So, now, finally, there are concrete plans to make long tunnels to keep the cars, well, as many as possible, out of the city centre.

    So, things are going the right way, but it will take time……

    13th September 2013
  4. There are 7 million of us living in here. And the roads are clogged up because people are driving cars on their own. At the end of the day that means that some people are going for a spin with an empty sofa (the cars back seats) and an arm chare (the passenger seat).

    The most popular car in the UK at the moment is the Ford fiesta, it weighs1.6 Tons. Let’s assume that the Fiesta is about average size for cars in the UK. That means every time you “pop out” in the car, that’s 1.6 tons of metal you decide to cart about.

    That is the equivalent of 538 red bricks. Think about what an effort it would be to drag that much weight to the shops on foot, every time you want to get some milk, or go to the cinema. You wouldn’t do it. It would be a profligate waste of energy. It is a profligate waist of energy.

    At some point we are going to have to just accept that, the parties over. We can’t keep needlessly wasting energy like this.

    Nisan has come up with the Leaf. But is that really the answer? It is still massive. It is still a machine for moving Goods and Passengers. What we need is personal transport to get us to work. Electric Vehicles could be perfect for Car Clubs. Short journeys were you need to move something heavy, but getting to work like this is simply mad.

    I would like to see regulation change to promote the use of small single person electric vehicles. Yep, I want to see lots of adults whizzing around on scooters and possibly some on skate boards. What a joyful thing that would be!

    Electric motors are small and powerful enough that these things can now do hundreds of miles on a charge and move at 20mph going uphill. You can fold them up and take them on the train. And they fit comfortably in any small house. They are there when you need them and they are much faster the current average speed of cars in cities.

    They are also so small that the batteries themselves don’t require too much land to be destroyed getting the minerals from and open cast mine.

    So why are they illegal in the UK?

    13th September 2013
  5. malenki said:

    Your writeup about combining the pollution data with maps is fascinating.

    The disturbing part for me was about the NAPTAN import.
    The importer should have taken care to avoid the problem of duplicates instead of obviously just dumping his data into OSM.

    @ Alexander Baines-Buffery:
    Vehicles powerd by electricity small enough to fold and light enough to carry into a train don’t go “hundreds of miles”.
    I assume you talk about pedelecs
    The average e-bike weighs minimum 25 kilogram and can go about 50 km before its battery is exhausted.¹ Then you have a nice training effect by having to pedal the extraweight. 🙂


    29th September 2013
    • Tom Chance said:

      The NAPTAN import was one of the first large-scale imports, one of a few which led the community to realise issues like duplicates needed a bit more attention. Somebody did create a tool to try and help people weed them out after the import, but as I said this has never been done very comprehensively.

      30th September 2013

Comments are closed.