Filed timetables on 545 city to city corridors

One service date per corridor

Each corridor was asked about once, for one service date, and every row carries that date and the instant of the fetch. It holds no frequency, no weekend pattern and nothing about next month.

1,254 corridors were put to the source. 545 came back with services, 694 had a stop at each end and nothing running between them, and 15 had no stop in the feeds within 60km of one end. Coverage follows who files an open feed.

502 of the corridors were probed for September 28, 2026 and 43 for September 30, 2026, which is the second date the probe tried where the first came back empty. Some itineraries run for several days, so departures in the file reach as late as October 6, 2026.

What is in it

Counted from the same snapshot the files are built from, so this table and the CSVs cannot drift apart.

• 545Corridors with services
• 1,035Itineraries
• 3,025Legs
• 11,222Calling points
• 3,368Named stops
• 1,267Of those, with a position
• 184Operators
• 33Countries
• 371Legs with a platform or track
• 620Legs filed as wheelchair accessible
• 2,132Legs carried in the GTFS feed
• 1,074Stops carried in the GTFS feed

Legs by mode. Bars are relative to the largest.

Ferry is a quarter of the legs, because these are corridors between cities rather than a network: an island is reached by ship or not at all.

The GTFS feed

The same snapshot as a GTFS Schedule feed, because that is what a transport researcher already has a tool for. One trip per leg, so trip_id is the leg_id in legs.csv and the two join directly.

111 agencies, 1,074 stops, 1,121 routes, 2,132 trips and 7,347 stop times across 9 service days. It validates against MobilityData's canonical GTFS validator with zero errors.

Every time in the feed is UTC

The reference requires one timezone per feed, and this snapshot spans Europe, Asia and the Americas, so every agency and every time in stop_times.txt is UTC, the instant the operator filed. Each stop carries its own stop_timezone to put the local clock back.

Deliberate deviations

1. One service date per corridor, not a repeating schedule. calendar.txt has one row per service day with start_date equal to end_date and a single weekday flag set.
2. Air is not in the feed: the reference defines ten route types and none of them is an aircraft. The 103 air legs are in legs.csv with mode = air and gtfs_route_type empty.
3. 893 of the 3,025 legs are not in the feed, almost all because a stop at one end has no position and GTFS requires a coordinate. gtfs_trip_id in legs.csv is empty on exactly those rows.
4. Some trips call at fewer stops than the operator filed, for the same reason. calling_points.csv always carries the full sequence, including the stops the feed had to drop.
5. Extended route types are folded into the ten the reference defines: 101 becomes rail, 200 becomes bus, 1004 becomes ferry. The original value is kept in legs.source_route_type, so the fold can be undone.

https://gtfs.org/documentation/schedule/reference/ · MobilityData canonical GTFS validator 8.0.1

How each row is evidenced

Every row in every file carries source, source_url, fetched_at and evidence.

filed

The operator published it: stop names, service numbers, platforms, departure and arrival instants, calling points, accessibility flags. Every row in four of the five tables.

derived

This package computed it: in stops.csv, the rows whose position was worked out from a gazetteer rather than given by the operator.

derived columns

A filed row can still carry a computed column. Durations, distances, counts and the folded GTFS route type are arithmetic over filed values, and the Table Schema names each one as derived.

2,405 of 3,025 legs are filed as not wheelchair accessible, which says how feeds are filled in rather than what the rolling stock is.

How to cite it

Attribution is the whole of the license. If you use this, this is the line.

@dataset{orlero_corridor_timetables_1_0_0,
  author    = {{Orlero}},
  title     = {{Filed timetables on 545 city to city corridors}},
  year      = {2026},
  version   = {1.0.0},
  publisher = {Orlero},
  url       = {https://orlero.com/data/corridor-timetables},
  note      = {CC BY 4.0}
}

Identifier