Transport modes

The mobility behavior of a population depends in part on the perceived cost of the different modes of transport that are available. The perceived cost of a given trip with a given mode aggregates factors in its cost of distance (€/km), its cost of time (€/h), and a cost constant (€/trip). We can then estimate how likely it is that a person chooses some mode instead of another, by comparing their perceived costs.

To be able to compute these perceived costs, Mobility estimates trips distances and travel times between all possible origin * destinations pairs for all transport zones under study (with a travel time threshold so we avoid computing unlikely trips such as 3 hour walks).

The available modes of transports that can be used in Mobility are :

• Unimodal modes:

  • Walk.

  • Bicycle.

  • Car.

• Multimodal modes:

  • Carpool:

    • Access to the carpool pick up point : car, single occupant.

    • Carpool ride : car, multiple occupants.

  • Public transport:

    • Access : any of the available unimodal modes.

    • Public transport ride : any combination of public transport services (bus, tramway, subway…).

    • Egress : any of the available unimodal modes (usually walk).

Unimodal modes

OpenStreetMap Data Preparation

The geographic data describing road, cycling, and pedestrian networks is sourced from OpenStreetMap. Maximum speeds are inferred from the road category when not explicitly provided in the data.

Transport Graph Creation

OpenStreetMap data is converted into a transport graph and simplified to accelerate computations:

• Successive road segments without intersections are merged by summing their distances and travel times, and averaging their capacities and congestion parameters.

• Dead-end segments are removed. It is assumed that buildings representing transport zones are located near roads that are not dead ends, and that dead ends will never be used for routing between transport zones.

The following table indicates which OSM road categories are retained for each of the three modes:

OSM Category	Car	Walk	Bike
bridleway	No	Yes	Yes
cycleway	No	Yes	Yes
ferry	Yes	Yes	Yes
footway	No	Yes	Yes
living_street	Yes	Yes	Yes
motorway	Yes	No	No
motorway_link	Yes	No	No
path	No	Yes	Yes
pedestrian	No	Yes	Yes
primary	Yes	Yes	Yes
primary_link	Yes	Yes	Yes
residential	Yes	Yes	Yes
secondary	Yes	Yes	Yes
secondary_link	Yes	Yes	Yes
service	Yes	Yes	Yes
steps	No	Yes	Yes
tertiary	Yes	Yes	Yes
tertiary_link	Yes	Yes	Yes
track	No	Yes	Yes
trunk	Yes	Yes	Yes
trunk_link	Yes	Yes	Yes
unclassified	Yes	Yes	Yes

Public transport

GTFS Data Preparation

The data detailing public transport service is sourced from feeds published in the GTFS (General Transit Feed Specification) standard: stop locations, days and times of service, and transport modes (bus, tram, train, metro, etc.).

Operator feeds are filtered to retain only lines with at least one stop in the studied transport zones. They are then aligned to a common date before being merged into a single feed.

Transfers between lines that were not originally specified are added within a 200-meter radius around each stop, with transfer times calculated based on straight-line distance between stops.

The public transport service from the Tuesday with the most operational services is used as the reference for modeling.

The GTFS data versions used for the Greater Geneva study are from late 2024.

Public Transport Graph Creation

The public transport offer is converted into an equivalent transport graph, allowing the construction of multimodal graphs by combining it with access and egress mode graphs (walking, biking, driving):

• Creation of entry and exit nodes in the public transport network by grouping nearby stops (within 40 meters) to reduce their number.

• Computation of average travel times between two stops on the same line, and of waiting times between arrivals and departures of the same line at a stop.

• Computation of average minimum transfer times between each arrival and the next departures of the other accessible services. Transfers exceeding 20 minutes are discarded.

• Calculation of initial waiting time, based on average headway: 50% of the headway if under 10 minutes, otherwise 5 minutes.

• Calculation of perceived waiting time, modeling the risk that the expected service is missed and the user must wait for the next one. This is derived from the line’s average headway and a perceived probability of missed service.

• Calculation of arrival time differences, between the target arrival time and all possible actual arrival times at the destination stop.

• Calculation of straight-line distances between stops, used for distance-based metrics.

The resulting travel times are therefore approximations of actual travel times. In reality, the departure time impacts travel duration, as several factors vary throughout the day: transfer times, vehicle speeds, etc.

Travel Time

The total travel time is composed of:

• A waiting time at the access node of departure (precautionary time).

• (Optional) A travel time to an intermediate stop for a transfer.

• (Optional) A waiting time for the transfer.

• A travel time to the egress node.

The number of transfers is not limited by the model, but a maximum travel time is enforced to exclude unlikely trips.

Functions

You must describe the transport modes that you want to model. Most usual modes are available, and you can use a combination of any mode with public transport. Available modes : walk, bicycle, car, carpool, public transport (+any mode before or after using public transport).

class mobility.transport.modes.IntermodalTransfer(*, max_travel_time: Annotated[float, Gt(gt=0)], average_speed: Annotated[float, Gt(gt=0)], transfer_time: Annotated[float, Ge(ge=0)], shortcuts_transfer_time: Annotated[float | None, Ge(ge=0.0)] = None, shortcuts_locations: list[Coordinates] | None = None)

Parameters governing transfer between two transport modes.

classmethod coerce_shortcuts_locations(value: Any) → list[dict[str, float]] | list[Coordinates] | None

Accept both {‘lon’,’lat’} items and [lon, lat] / (lon, lat) pairs.

model_config = {'extra': 'forbid'}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class mobility.transport.modes.ModeRegistry(modes: List[TransportMode])

Register transport modes by id.

get(mode_id: str) → TransportMode

Return a registered mode by id.

class mobility.transport.modes.TransportMode(name: str, travel_costs, generalized_cost, ghg_intensity: float, congestion: bool = False, vehicle: str = None, multimodal: bool = False, return_mode: str = None, survey_ids: List[str] = None, parameters: TransportModeParameters | None = None, parameters_cls: type[TransportModeParameters] | None = None)

A base class for all transport modes (car, bicycle, walk…).

Attributes:

name (str): the name of the mode. travel_costs (PathTravelCosts | PublicTransportTravelCosts | DetailedCarpoolTravelCosts):

a travel costs object instance that computes and stores travel costs (time, distance) between transport zones.

generalized_cost : congestion (bool):

boolean flag to enable congestion, if the travel costs class can handle it (only PathTravelCosts).

build_congestion_flows(od_flows_by_mode)

Build congestion-relevant flows for this mode.

Args:

od_flows_by_mode (pl.DataFrame): Aggregated person flows by

["from", "to", "mode"].

Returns:

pl.DataFrame | None: Congestion-relevant flows for this mode, or None when this mode does not directly refresh congestion.

class mobility.transport.modes.TransportModeParameters(*, name: str, ghg_intensity: Annotated[float, Ge(ge=0)], congestion: bool = False, vehicle: str | None = None, multimodal: bool = False, return_mode: str | None = None, survey_ids: list[str] | None = None)

Common parameters for transport mode definitions.

model_config = {'extra': 'forbid'}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].