Football Ground Nearest Station Mark II

When I wrote the blog post “Settling a Coffee Break Debate – Football Travel Distance” It was done to do just that, settle a coffee break debate. Getting the Football Supporters’ Federation to share it came as an afterthought. In doing so the true power of the crowd was unleashed and we received several comments from fans pointing out minor errors in the results. Mainly missing railway stations or errors in the location of the football ground. As such I wanted to prove that the Geospatial Engineering group are nothing short of perfectionists and rectify these results.

Firstly I had to retrieve all the locations of railway stations in the UK. I did this by using the OpenStreetMap extended API and download all features in the UK bounding box that contained the tag “railway=station”. This gave me a selection of nodes and polygons. I generated centroids of the polygons and merged them with the nodes. As a few commenters pointed out the last piece of analysis didn’t include tram, light railway and subways. So for completeness I also carried out the same process on features marked “railway=tram_stop” (light rail and subway stations are included in “railway=station”).

I then had to retrieve the locations of football stadiums. Again I used the OpenStreetMap extended API and downloaded all features that contained both “sport=soccer” & “leisure=stadium” tags. And likewise generated centroid of the polygons and merged with the points. However due to either the stadium not being tagged as such or simply not mapped out yet a few grounds were missing so in these places I have used the previous locations. I also had the problem that the openstreetmap features didn’t contain information on the club and the stadium name were in some cases slightly different to the names I had stored ( e.g. Emirates, and Emirates Stadium) so I had to do some fuzzy text matching to complete this mapping.

Once I had the 2 new datasets it was just a simple case of calculating the nearest station/stop to each stadium and the distance between them. Again I achieved this by loading both datasets into PostgreSQL/PostGIS. Below are the full set of results. With the representing the distance from the centre of station/stop to the centre of the stadium, so not necessarily the distance a fan would walk. Also this does not take into consideration that the station or stop might not be open on match day.

Author: Neil

Software Engineer/ Research Assistant Urban Observatory View all posts by Neil