Question for you all - What is a Train station? What is an Underground?

Open Data
1

Hey Everyone

I am starting to resolve some of the NaPTAN data … insanity*… and we are starting with an audit of stations in the 910 (rail) and 940 (metro) data sets.

I wanted to ask youse as a community who deal with data - what in your minds and in passengers minds are the differences between a Train station and an Underground station (in the TfL area only …)

I also want to find places where there are both a Rail and a Metro station at the same stop - e.g. Chalfont and Latimer - where the same platforms might be used

I am stuck in analysis rabbit holes and thought this community might be a good one to ask about how these differ in the “give data to Passenger”

I am also happy to take any improvements to the data feed.

No actual fix times will be given - I am at the figuring out the scope stage

Cheers

Dr J

*Insanity - there are five Clapham Junctions in NaPTAN, there are only three Rail Platforms, most of the Metro in London after some time 2015 has only one Metro Platform - these we know - tell me more insane moments you have to code around!

3 Likes
2

Hi Dr J,
For me, I’d say the difference is a ‘Train station’ is for any mainline train station which comes under the jurisdiction of Network Rail perhaps. That means London Overground comes under ‘Train Station’. ‘Underground’ to me would be any normal London Tube line or DLR service.

As for your other question some instances of the ‘underground’ and mainline rail services sharing the same platforms are:

  • Bakerloo Line/London Overground between Queens Park and Harrow & Wealdstone (Queens Park mostly the Bakerloo line trains use their own platforms but I believe they can use the Overground platforms occasionally to make use of a turnback at Kilburn High Road).
  • Metropolitan Line/Chiltern between Harrow On The Hill, Rickmansworth, Chorleywood, Chalfont & Latimer and Amersham
  • District Line/London Overground between Richmond, Kew Gardens, and Gunnersbury.

Purely for information, if you are looking at this as part of your accessibility data, it’s worth noting instances like Rayners Lane to Uxbridge which has the Metropolitan Line and Piccadilly line trains, these offer different levels of accessibility. Same platforms for both sets of trains.
Acton Town & Ealing Common similarly, served by both District Line and Piccadilly line.

3

Hi @drjharrison,

I remembering look at the NaPTAN data quite a while back. I thought at the time it could be very useful but also that it was quite complex. Maybe insane’s a better description :slight_smile:

For a (not necessarily 100%) list of ‘train stations’ these can be found at tubestation.uk. Those with the old BR sign.

Mike

4

It’s pretty complicated and without knowing what you’re trying to achieve it’s difficult to know whether there is an answer that would satisfy you.

For me, a Train Station would be one managed by Network Rail, but I’m looking from an operational point of view rather than a customer’s one.

If it helps, as far as I’m aware, train services in the TfL are are provided by:

Network Rail
London Underground (the Tube network)
Rail for London (the Overground and Elizabeth line)
DLR (the Docklands Light Railway)
Tramlink (the Croydon Tram)

The last four operating on behalf of Transport for London.

I would expect all stations to ‘belong’ to either ‘Network Rail’ or ‘Transport for London’ (the only exception I’m aware of being Heathrow Terminal 5 which belongs to British Airports)

…but I can’t think of a ‘rule of thumb’ for working out which organisation a particular station would belong to.

Note.
To Pete’s list of shared platforms you could perhaps add Putney Bridge to Wimbledon, where the District line runs over Network Rail Tracks. The two services don’t share platforms at either Putney Bridge or Wimbledon and, although they do both run through the same platforms at East Putney, Southfields and Wimbledon Park I don’t believe Network Rail services are ever timetabled to stop there?

Good Luck.

5

Thanks all for your responses

I am trying currently to audit what we have in NaPTAN and make sense of decisions taken in 2006 (and prior) without any documentation to help out.

As I have the chance to give a clear “this is a station” giudance for “future me” I wanted to get a wide set of viewpoints on this.

For example - Beryl is travelling in London - she wants to go from the M&S in Oxford Circus to Grainery Square in Canary Wharf…

When this journey is being presented to her - does it matter to her that the Elizabeth Line platforms at Tottenham Court Road Station are National Rail Platforms?

Is it that she can use her card to tap in and out?
Is it that she can see how many minutes to the next train?

What would make it “not a tube”?
What do we need to communicate?

I am aware my questions may seem very esoteric and I do question things we “all just know” - I did spend 2020 and 2021 asking “What is a Bus Stop?” to arrive at the definition ‘A Bus stop is where the bus stops’ - I do this as it teases out all of the assumptions and ‘it just is…’ out of us all.

Thanks!

6

Interesting questions but I suspect that the flexible ticketing provided by TfL is too complex to incorporate into NaPTAN simply by defining StopPoint categories for “train” stations.

Looking wider than Beryl’s short journey but still within “London” the answer would depend on the ticket type (e.g. Contactless, Oyster, Freedom Pass), start/finish locations, and in some cases (e.g. Freedom Pass) the time of day.

Leaving that aside, London has 6 “train” systems for which traditional names make a clear distinction with one exception: Tube/Underground, Overground, DLR, Tram, National Rail. This would be suffice if Tube was reserved for the small train diameter lines keeping Underground for only District, Metropolitan, etc. Whatever you do with NaPTAN, it is inevitable that many London “train” stations will fall into multiple categories.

7

It can also get very station-specific. Using the Elizabeth line as an example:

Reading-Iver inclusive: same platforms as National Rail, same gatelines, pay-as-you-go available, not Oyster, but Freedom passes OK after 9:30.

Heathrow Terminals 2 & 3, 4, 5: separate gatelines for Piccadilly, shared platforms with National Rail, all tickets OK, but only Heathrow Express ones valid on HX trains.

Hayes & Harlington-West Ealing, Acton Main Line: shared platforms & gatelines with NR.

Ealing Broadway: shared gateline for all services, separate platforms for Central & District.

Paddington: almost too complicated to explain. What we have in effect is four separate stations
A: Mainline station - four different gatelines (one for HX only), plus some open-access platforms, used by NR,HX and early morning Elizabeth
B: Elizabeth line station: own gatelines and platforms, but has escalator link to Bakerloo (see C)
C: Bakerloo/Circle/District (towards Bayswater): three sets of gatelines, either side of Praed Street and also south of & below the main NR concourse. Escalator link from Bakerloo platforms to Elizabeth line (see B)
D: Circle/District (towards Royal Oak): separate station at the far end of the complex (approx 1/2 mile walk within the station)

As far as C & D go, it doesn’t matter which one you pick if you’re heading to Edgware Road, as both lines go there, but is extremely important if you’re heading the other way.

Bond Street-Farringdon & Whitechapel: common gatelines, separate platforms

Liverpool Street:
platform-specific gatelines for NR/Overground & early morning Elizabeth, common area for main Elizabeth/Central/Circle/Hammersmith & City/Metropolitan
This then links, via the Elizabeth line platforms and a long walk to Moorgate, connecting to Circle/Hammersmith & City/Northern & (different) NR service

Canary Wharf: three entirely separate stations, one each for Elizabeth, Jubilee, DLR.

Custom House: one station, two separate gatelines for Elizabeth & DLR

Woolwich: Elizabeth line only, Woolwich Arsenal is a 5-minute walk away

Abbey Wood: common gatelines with NR, separate platforms

Stratford: Once common concourse area for all platforms & services, mostly dedicated platforms, but some sharing between NR & Elizabeth.

Maryland-Harold Wood: common gatelines for NR & Elizabeth (& Overground at Romford)

Brentwood-Shenfield: same, but Oyster not available, Freedom passes OK

1 Like
8

Weeping in Data…

I will be back on the week of the 23rd - if anyone has a brain wave on “what is a train station?” and “What does that mean to Beryl?” - I’ll be back then to pick them up…

9

I personally feel that there’s no “one size fits all” solution to this problem, as there are just so many things to consider, e.g:

  • a station can be owned and managed by different organisations, e.g. Abbey Wood is owned by Network Rail, leased out to Transport for London, and operated by MTR Elizabeth line.
  • it can be jointly managed by multiple organisations, e.g. Stratford is operated jointly by MTR Elizabeth line and London Underground.
  • it can be served by a mix of heavy, metro, light, and very light rail modes of transport, e.g. Wimbledon is served by National Rail, London Underground, and Trams.
  • one mode in the station can be more visible, prominent, frequent, or heavily used than that of the station owner/manager, e.g.
    • Queen’s Park is managed by Arriva Rail London, but the Bakerloo line serves it more frequently than the Overground;
    • Canning Town is managed by London Underground but has more DLR platforms than Jubilee line platforms.
    • Bond Street is a London Underground managed station, but 2 out of the 4 entrances are branded almost exclusively for the Elizabeth line, despite all 4 entrances being able to get you to any of the 3 services serving the station.

I find the difference that you’ve pointed out in these 2 quite fascinating, and it brings up even more questions.

Operationally, the DLR part of Custom House and the Elizabeth line part of Custom House are 2 separate station. One can operate without the other, they can be closed independently of each other, and both the stations themselves and the railway/services running through them are operated by 2 separate companies.

When I travel through Custom House as a passenger I would personally say that, like at Canary Wharf, they feel like separate stations because to interchange between the 2 services you need to tap out and exit of one station, walk over, and enter and tap in to the other.

So if some passengers feel they are 1 station while Canary Wharf is 3 stations, then that brings up the question of what would qualify 2 directly adjacent stations to be considered 1 larger station? How far apart do they have to be? What sort of physical or ticketing distinction does there need to be to keep the 2 considered separate? Do they need to be distinct buildings - and what about if they are part of a larger non-railway building, but at opposite ends?

If one were to consider Custom House to be 1 station rather than 2 adjacent stations, despite them being operationally separate and having barriers between them, would you then have to say that Moorgate+Liverpool Street, or parts B and C of Paddington, are also 1 station?

I think as with a lot of things, my answer here is “it depends”. I will slightly change your question to be a journey from Bond Street to Canary Wharf, because of some specific signage present at Bond Street and a handful of other Elizabeth line stations.

While at Tottenham Court Road all the interchange signage names specific lines, at Bond Street there is some signage that simply points to “Underground”, which in the case of that specific station is referring to the Central and Jubilee lines, but doesn’t name them explicitly. I think for this reason it is important to highlight the distinction of the Elizabeth line not being a London Underground service, so that some occasional vague signage does not get the passenger lost within a station.

That being said, I wouldn’t necessarily say that it needs to be explicitly labelled as a “National Rail” platform, as that then also contradicts the signage, with TfL keeping the Elizabeth line branding separate from the wider National Rail brand. So while sticking a National Rail double arrow into the journey planner and explicitly mentioning National Rail at Bond Street would be a little overkill in my opinion, I would say that some attention would need to be paid to how the platform is described and what iconography/colours go along with it, to be clearer on which part of a larger hub station the passenger would need to head towards.

I feel like this is an absolutely monumental task that is bound to always have more than 1 answer and lots of differing opinions, so I wish you the best of luck, and look forward to seeing the result of your work!

10

So if some passengers feel they are 1 station while Canary Wharf is 3 stations, then that brings up the question of what would qualify 2 directly adjacent stations to be considered 1 larger station? How far apart do they have to be? What sort of physical or ticketing distinction does there need to be to keep the 2 considered separate? Do they need to be distinct buildings - and what about if they are part of a larger non-railway building, but at opposite ends?

TfL publish a spreadsheet of out-of-station interchanges where you can swipe out and back in without breaking your journey as far as ticketing is concerned provided it’s done within a specified time window.

Sometimes this is nominally within the same station (Blackfriars LU and Blackfriars NR) and sometimes they’re completely different (Monument LU and Fenchurch Street NR)

11

As far as I can see/work out, the difference between ‘Rail’ and ‘Metro’ stations in NaPTAN is that ‘Rail’ stations are licenced by, or have services licenced by, the ORR (and are regarded as part of the National (ex-BR) Rail network), while ‘Metro’ services/stations are not ORR licenced - London Undergound’s licence is from the Secretary of State, DLR has an exemption (I think as it’s a physically seperate network), and Tramlink is not a ‘railway’ (within the meaning of the Act), so doesn’t need one! Rail for London (which I believe covers London Overground and Elizabeth Line) does have an ORR licence. So it is really just a legal or administrative distinction, not a fundamental one, and - operator apart - something that doesn’t really matter to me as a passenger.

But, that being said, there are many stations that are both - they have both ORR-licensed services and not ORR-licensed service, with various degrees of integration. So it really comes down as to what can be regarded as ‘one station’, and when does it become two, which can be a difficult distinction to draw. (I raised this question on a different, enthusiast, forum - see https://www.railforums.co.uk/threads/trivia-how-many-stations-and-when-does-one-station-become-two.263440/ if you’re interested - and no conclusion was reached).

My own thinking is that a single station is a place (entity), with one identity and one person (organisation) in charge, that you can enter to board a train, etc. of any operator: it becomes a second station when you have to leave it and enter another (whether this means crossing a road or just passing through a gateway) to get a different train. A greater distinction than just having seperate gatelines. So on Tramlink (my local system) at Wimbledon and Elmers End it uses platformsthat are within those (National Rail) stations, and so uses them, rather than have its own, but elsewhere (e.g. Mitcham Junction, and both East and West Croydon) it has its own stops, that are outside and distinct from the NR stations.

12

Reading that thread and comments on here suggests that my previous post was on the right lines (sorry!) by referring to categories for the type of train system rather than type of station.

Just define a train station as (e.g.) a StopPoint for vehicles on metal wheels running on metal rails. Each type of train system can be allocated a specific code and icon which tables or maps can show as appropriate for each station. This makes sense even for trams, as shown by the Croydon Tramlink where many of the stops are either normal railway station platforms or unmanned halts within a fenced off railway track.

13

I am inclined to agree, though I would use the term ‘railway station’ (‘train station’ I regard as an ugly Americanism) for all railway Stop Points (both London Underground and DLR are railways, with their stops called ‘stations’, after all), but my thought is that perhaps tram stops should be kept seperate, as ‘tram stops’ - that is what they are normally called, and, certainly in London (Croydon Tramlink), the tram system is outwith the normal rail mode ticketing/fare structure.

Trams are in the unfortunate position of indecision between being ‘bus on rails’ and ‘train in the streets’, generally ending up as a hybrid, not quite either…

14

Hello Dr J,

I am a software engineer of Aubin, a multi-modal journey planner covering the whole of Great Britain using open data, which is currently in the beta phase.

Our journey planner has the ability to route passengers direct to the platform inside stations, with step by step navigation outside and inside stations, including which lifts to use, as long as the data exist in OpenStreetMap. It can tell you in real time which platform the train departs and guide you on the map to the platform as well.

We use Naptan AtcoCode as the identifier for all stops, but after checking the data, we have to make some adjustment on top of the data.

The first thing is that, a station, or even a single platform, can serve both mainline rail (National Rail) and metro services. However, the current data set have both the 9100 and 9400 codes for these stations / platforms where on the ground they are the same. As a result we have to process the feed to replace the 9400 codes by 9100 codes if the stations / platforms are shared.

The second thing is that, the Naptan schema guide specifies the codes for a stop area, access place and quay (using Transmodel terminology), however in the real dataset, I can’t find any real data for the platforms (quays) for mainline rail stations, but only the access places in the format of 9100TIPLOC, and a rail station will have multiple of these if there are multiple TIPLOCs (where the scheme guide specifies that the platforms should be in the format of 9100TIPLOC1). Even if there are multiple TIPLOCs for a rail station, they are part of the same station so they need to belong to the same stop area (910GTIPLOC), where the primary TIPLOC is used (the primary TIPLOC is the one in the National Rail timetable data station file with an interchange status not equal to 9).

Therefore we are not directly using the Naptan data for mainline rail stations / platforms but generate the codes directly from National Rail timetable data.

Using Clapham Junction as an example, we have data for various platforms in the following form, using the GTFS model:

{
      {
        "gtfsId": "GB:9100CLPHMJW3",
        "name": "Clapham Junction (Platform 3)",
        "platformCode": "3",
        "parentStation": {
          "gtfsId": "GB:910GCLPHMJC"
        }
      },
      {
        "gtfsId": "GB:9100CLPHMJM9",
        "name": "Clapham Junction (Platform 9)",
        "platformCode": "9",
        "parentStation": {
          "gtfsId": "GB:910GCLPHMJC"
        }
      },
      {
        "gtfsId": "GB:9100CLPHMJC14",
        "name": "Clapham Junction (Platform 14)",
        "platformCode": "14",
        "parentStation": {
          "gtfsId": "GB:910GCLPHMJC"
        }
      }
}

Another defect in the data is that, in the dataset, some tube stations only have one platform code but in the real world, there are multiple platform islands. For example, there is only one code listed for Finchley Road where services for both directions are assigned the same code, but in reality, there are 4 platforms, where both northbound platforms are on an island, and both southbound platforms are on another island. It is impossible for us to correctly tell a wheelchair user that changing trains between the same direction is possible, but changing between different directions is impossible. We therefore need accurate platform data in the dataset to support these use cases.

In order to provide more accurate navigation instructions, we also replace the location provided in the dataset with OpenStreetMap data if a match can be found.

Whether the platform is a rail platform or a metro platform is not important to us, and it is meaningless if both rail and metro services use it. The more important thing is to ensure that in the timetable data, both rail and metro services use the same code if they share the same platform in the real world, and use different codes if they use different platforms.

In regard to showing the symbols and branding, the stations are totally irrelevant. Only the service matters. If the service in an itinerary is a Metropolitan service, the tube logo is shown, if the service is a Chiltern service, the National Rail logo is shown, if the service is a heritage railway service, a generic train logo is shown. If the user plans a journey from Amersham to Rickmansworth, both the metro and the rail services are shown side by side to each other in a similar manner that the user can choose between multiple bus routes, as they use the exact same stops.

You can try our API, especially the stops endpoint, on OTP GraphQL Explorer and our web app on https://beta.aubin.app/ which uses that API.