Proposal:Public transport schema

Workshop

The proposal is the result of a community workshop which took place from 16th to 17th of May, 2009, at Geofabrik, Karlsruhe (Germany). The list of all workshop participants can be found below:

• Markus Bader
• Christian Berreth
• Christoph Eckert
• Heiko Jacobs
• Matthias Merz
• Melchior Moos, author of a public transport map based on OSM data
• Frederik Ramm
• Thomas Reincke, employee of Aachener Verkehrsverbund GmbH (public transport authority of Aachen, Germany)
• Frank Sautter
• Sebastian Schwarz
• Jochen Topf
• Malte Ahrens

All individual aspects of the schema presented in the text below arose from the following topics discussed during the workshop:

• basic concept for line-like traffic infrastructure
• basic concept for point-like traffic infrastructure
• basic concept for network information (lines and routes)
• basic concept for network information (public transport networks)
• general separation of infrastructure data and network information
• public transport routing and timetable information
• rendering and visualization
• (micro-)tagging
• realization of automated data matching

Related proposals

Currently, there are some other activties within the OSM community working on enhanced modeling schemas for PT related data, too, which can be found in the transit category of the OSM wiki. Some aspects of the schema described in this proposal base upon two proposals which can also be found in the mentioned category:

• Proposed features/unified stoparea
• QROTI

In the NaPTAN category of the OSM wiki, all activities concerning the NaPTAN data import can be traced. These activities were also payed attention to during the developing process of this proposal.

Modeling multi-track railways

Multi-track parallel railways in close proximity can either be modeled as a single way with tracks=* or as a number of parallel ways. If individual tracks have different tagging requirements (electrification, gauge etc.), the tracks should be modeled appropriately. The tracks=* tag should be used to record the number of tracks with a default value of 1 being assumed where this is not supplied.

Railways for rails

To distinguish between different types of railways for rails, in addition to railway=rail the tag usage=* shall be used – it describes the usage of a railway for rails:

• usage=main for main lines
• usage=branch for branch lines
• usage=industrial for industrial railways
• usage=military for military railways
• usage=tourism for touristically used railways

Further tags for railways for rails are:

Railways for light rails

In addition to railway=light_rail, additional tags for railways for light rails can be:

Railways for metros

Metros can be divided into two different categories:

• railway=subway for subways
• railway=monorail for monorails

Modeling subways, a mapper shall always use the tag tunnel=* with the value yes for railways underground and with the value no for railways overground.

Modeling monorails, a mapper shall always use the tag monorail=* with the value hanging for suspended monorails and with the value magnetic for magnetic monorails.

Additional tags for railways for metros can be:

Railways for trams

Railways for trams shall always be modeled as separate geometries, most notably when the railways are grooved. Otherwise, if only one geometry is used for a road and the grooved railway on it, additional tags cannot be clearly assigned to either the road or the railway.

Additional tags for railways for trams can be:

Funiculars

Funiculars did not appear as an independent category in OSM so far: From now on, they shall be modelled as ways with the tag railway=funicular.

Additional tags for funiculars can be:

Public conveyors

Public conveyors also did not appear as an independent category in OSM so far: From now on, they shall be modelled as ways with the tag highway=conveyor.

Additional tags for public conveyors can be:

Waterways for passenger ferries

From now on, the waterways for passenger ferries shall be modeled as ways with the tag waterway=ferry_way to separate infrastructure data from network information.

Basic model for stops

Therefore, a basic model for stops shall be established: The basic structure of this model is independent from the types of public service vehicles serving a stop or the traffic way a stop is related to. The model is kept as simple as possible and allows mappers to model very simple stops as well as very complex stops with many details. Therefore, the model consists of four components:

• the stop position of a public service vehicle on a traffic way
• the access for passengers to the public service vehicles or the stop itself
• the stop area
• the stop area group

The graphic above shows that the model consists of three steps at most: A stop area (as a relation) contains at least one stop position and either no access or an arbitrary amount of accesses. Consequently, a stop area builds the relationship between all the elements it contains. A stop area group (as a superior relation) shall be used if multiple stop areas have to be combined – e.g. because they are just representing multiple parts of one stop.

Stop position

Stop positions shall be modeled as nodes and tagged with public_transport=stop_position. The nodes shall be placed directly on the ways representing the traffic ways the stop positions are related to: This makes routing more easy and leads to more uniformity.

Although the described model for stops makes the stops independent from the public service vehicles, additional tags shall be used for stop positions to allow rendering systems to visualize them with adequate symbols and icons:

If those additional tags shown above are missing, the absent information could still be read out from the line relations containing the stop position. But this would be considerably more difficult to implement.

Another important aspect concerning stop positions is the following questions: When to record one stop position and when to record more than one? The answer: One stop position shall be modeled if there is just one in reality, if two bus stops are directly facing each other or if the stop relates to a single-track railway. In all other cases, more than one stop position shall be modeled, namely as much as there are in reality.

Access

The geometry and the characteristical tag an access shall be modeled with depends from the type of the access:

Additional tags for accesses can be:

Buildings, ferry terminals, piers or other facilities can be integrated as accesses in a stop area relation, too. The existing modeling schema for those elements shall be maintained.

Stop area

A stop area relation shall be tagged with public_transport=stop_area. The additional tags a stop area relation can be equipped with mostly serve as unique identifiers of the stop area:

A stop area relation is not always necessary: It would be superfluous if it contained just one stop position as a member. In such cases, the isolated stop position shall be equipped with the tags listed above which uniquely identify it – including the tag operator=*.

Stop area group

A stop area group relation shall be tagged with public_transport=stop_area_group and should not be mistaken for a "transfer relation" but shall be understood as a type of relation which can be used to combine multiple stop areas – e.g. because those stop areas are just representing multiple parts of one stop: One stop area tagged with name=Church Square (East) and another stop area tagged with name=Church Square (West) can be combined by a stop area group tagged with the name name=Church Square.

A stop area group can also contain an arbitrary number of additional nodes which are representing other point-like traffic infrastructure:

• taxi stands and call boxes
• rickshaw stands
• rent-a-car facilities
• rent-a-bike facilities
• car sharing facilities
• water taxi stands and call boxes

Examples for the application of the model for stops

In the most simple case there is only one stop position with a name:

If two accesses are "joining" the stop position a stop area shall be created:

If two (not necessarily adjacent) stop areas belong to one stop a stop area group shall be created:

Backward compatibility of the model for stops

The backward compatibility of the model described above is guaranteed because existing point-like traffic infrastructure can be preserved – merely the meaning of those map features will be interpreted in a different way from now on: The existing map features for stops (which are tagged with highway=bus_stop, railway=halt etc.) will be interpreted as stop positions if they are placed directly on traffic ways; if not, they will be interpreted as accesses – like all the other map features for accesses already existing (which are tagged with highway=subway_entrance, railway=platform etc.). By additional tagging, more information (public_transport=stop_position, public_transport=platform etc.) can be attached to the existing map features without any problems. The stop area and stop area group relations can be created on top of the existing data.

Suitability of the model for stops

The model for stops described above is suitable for:

• bus stops and bus stations
• trolley bus stops and trolley bus stations
• rail halts and rail stations
• light rail halts and light rail stations
• commuter rail halts and commuter rail stations
• metro halts and metro stations
• tram halts and tram stations
• funicular stations
• aerialway stations
• passenger ferry terminals

Public elevators

Public elevators did not appear as an independent category in OSM so far. As the model for stops described above is not suitable for the modeling of public elevators, those elements shall be tagged as nodes with highway=elevator from now on.

Basic model for lines

Therefore, a basic model for lines shall be established: The basic structure of this model allows mappers to easily design different types of lines as well as different ways there and ways back. The model is kept as simple as possible and allows mappers to design alternate line courses in a comprehensible way, too. Therefore, the model consists of five components:

• line
• line variant
• traffic way (as a relation member)
• stopping place (as a relation member)
• access (as a relation member)

The graphic above shows that the model consists of three steps at most: A line variant (as a relation) contains an arbitrary amount of traffic ways, stopping places and/or accesses. A line (as a superior relation) shall be used to combine all the different variants of a line.

Line variant

Separate relations shall be used for each direction of a line. Those relations contain all stopping places, accesses and traffic ways as members. The sequence of these members in the ordered list exactly expresses the connection between the source location and the target location of the line in reality.

Accesses shall be included in line variant relations because of two reasons: Firstly, this approach makes it possible for pedestrian routing applications to refer pedestrians to the correct accesses (e.g. to platform F which grants accesses to subway line M). Secondly, this approach guarantees the backward compatibility of the whole model for lines, because all the existing map features tagged as stops (e.g. with highway=bus_stop) but being placed beside traffic ways can be preserved as valid relation members.

To model alternate line courses separate relations shall be used, too, because using only one relation for all alternate line courses at once (e.g. applying the role alternate for alternate relation members) would make the correlation of different alternate relation members impossible: Consequently, when processing the relation the alternate line courses could no longer be reassembled.

For a line variant relation, only the following tags are required:

By modeling a line variant relation, the mappers shall – apart form the topics mentioned above – consider further aspects, too: Firstly, the ways representing the traffic ways being included in the line variant relations don't have to be segmented at each stopping places. Secondly, stopping places served on demand can be assigned with the role on_demand. Thirdly, the line variants of public service vehicles running on railways shall either include the correct rail track it is running on (if individual rail tracks are modeled) or the approximately correct rail track (e.g. if three of five rail tracks are modeled) or the only rail track or the trace (if only one rail track or a trace is modeled).

To model so-called "telescope lines", line variant relations are not necessarily required: Telescope lines feature occasional extensions by one or more stops, for example at certain times of day/night. To model telescope lines, it is sufficient to model one relation and to assign the role additional to the appropriate members representing occasional extensions.

Differentiation between railway lines and railway routes

There are railway lines and railway routes which shall be distinguished from now on: Railway routes are several ways that form some kind of entity, for example the railroad stretch from one city to another. Railway lines are the scheduled lines running on rail tracks, for example the subway "U96". So just as with cycle routes etc., routes are a way to bring several ways together under one name. Lines, on the other hand, are formed by vehicles moving regularly on this infrastructure.

To realize the differentiation between lines and routes, the tag route=* attached to relations will be ignored from now on if its value does not equal rail. The tag route=railway shall be used from now on to indicate relations for routes, all the other tags route=* will thus become obsolete. To model a line, the tag line=* shall be used from now.

Bus lines and trolley bus lines

Tags for bus lines and trolley bus lines can be:

Rail lines

Tags for rail lines can be:

Light rail lines

Tags for light rail lines can be:

Metro lines

Tags for metro lines can be:

Tram lines

Tags for tram lines can be:

Passenger ferry lines

Tags for passenger ferry lines can be: