KTH Open and FAIR Data
Published September 2, 2026 | Version v1
Dataset Open

Stockholm Public Transport (SL) static GTFS merged with the observed arrival/departure times reported via GTFS-RT TripUpdates and ServiceAlerts, 2024 Dataset

Authors/Creators

  • 1. ROR icon KTH Royal Institute of Technology

Description

This dataset is not a snapshot or sequence of snapshots of vehicle-trip progress, but rather a daily summary of all arrival and departure observations. It is the result of us independently polling that live feed every 15 seconds throughout 2024, matching each time snapshot against the planned GTFS Regional static timetable active for that service day (including planned vehicle-trips not present in TripUpdates for unknown reasons). Thus, this dataset provides complete daily planned supply with its observed performance. In processing, we consolidate the many snapshots covering each stop visit into a single record of the observed arrival and departure time (if observed). In other words, we keep only the final observations instead of the full trip progress, or we wait for the vehicle trip to complete and keep its last streamed snapshots. This results in a significantly smaller dataset of observed arrival and departure times.

One day of data from Trafiklab's stream is about 4.5 GB, so this dataset is the result of collecting and processing 1.7 TB of data during 2024. This dataset already contains ready-to-use, per-stop-visit observed delay, arrival, and departure times across the Stockholm Region, saving users from collecting and processing raw GTFS-RT snapshots themselves. Note that the same underlying live feed (https://www.trafiklab.se/sv/api/gtfs-datasets/gtfs-regional) is also archived, unprocessed, by Trafiklab in the KODA repository (https://www.trafiklab.se/sv/api/our-apis/koda); we collected this dataset independently of KODA, so its polling interval and coverage may differ from ours, KODA is a good alternative source if you need raw feed snapshots for other operators/regions, or wish to reproduce this dataset's processing yourself.

Abstract

Coverage: 2024-01-01 to 2024-12-31 (calendar year, Europe/Stockholm operating dates).

The GTFS-RT TripUpdates and ServiceAlerts feed published by Trafiklab is a live, snapshot-based service: each request returns only the feed's current state (updated on the Trafiklab server approximately every 15 seconds). Each vehicle-trip present in the dataset is streamed with a scheduled plan, and observations are either predicted or observed depending on progress during the trip.

This dataset is not a snapshot or sequence of snapshots of vehicle-trip progress, but rather a daily summary of all arrival and departure observations. It is the result of us independently polling that live feed every 15 seconds throughout 2024, matching each time snapshot against the planned GTFS Regional static timetable active for that service day (including planned vehicle-trips not present in TripUpdates for unknown reasons). Thus, this dataset provides complete daily planned supply with its observed performance. In processing, we consolidate the many snapshots covering each stop visit into a single record of the observed arrival and departure time (if observed). In other words, we keep only the final observations instead of the full trip progress, or we wait for the vehicle trip to complete and keep its last streamed snapshots. This results in a significantly smaller dataset of observed arrival and departure times.

One day of data from Trafiklab's stream is about 4.5 GB, so this dataset is the result of collecting and processing 1.7 TB of data during 2024. This dataset already contains ready-to-use, per-stop-visit observed delay, arrival, and departure times across the Stockholm Region, saving users from collecting and processing raw GTFS-RT snapshots themselves. Note that the same underlying live feed (https://www.trafiklab.se/sv/api/gtfs-datasets/gtfs-regional) is also archived, unprocessed, by Trafiklab in the KODA repository (https://www.trafiklab.se/sv/api/our-apis/koda); we collected this dataset independently of KODA, so its polling interval and coverage may differ from ours, KODA is a good alternative source if you need raw feed snapshots for other operators/regions, or wish to reproduce this dataset's processing yourself.

Terminology

SL (Storstockholms Lokaltrafik) is Stockholm's public transit operator. GTFS (General Transit Feed Specification) is the static schedule format used for its planned timetables; GTFS-RT (GTFS Realtime) is the companion real-time feed format, used here for TripUpdates (vehicle-reported arrival/departure estimates) and Service Alerts (disruption notices).

  • SL (Storstockholms Lokaltrafik): The public transport authority/operator for the Stockholm region, responsible for bus, metro, train, tram and ferry services.
  • GTFS (General Transit Feed Specification): A standard file format for publishing static public transit schedules and associated geographic information (stops, routes, trips, timetables).
  • GTFS-RT (GTFS Realtime): A companion format to GTFS that conveys real-time updates: vehicle positions, trip updates (predicted/observed arrival and departure times) and service alerts (disruption notices), as protocol-buffer feeds.
  • TripUpdates: A GTFS-RT feed type reporting real-time predictions or observations of arrival/departure times for scheduled trips.
  • Service Alerts: A GTFS-RT feed type reporting service disruptions, delays, or other notices affecting routes, trips, or stops.
  • AVL (Automatic Vehicle Location): Onboard vehicle tracking used as a source for the real-time positioning/timing data merged into this dataset.

Acknowledgment

Original data source: Trafiklab (https://www.trafiklab.se)

Data collection and processing: Intelligent Mobility Lab (iMobility Lab), KTH (Matej Cebecauer and Anastasios Skoufas).

Dataset prepared by: Daniel Chaves and Matej Cebecauer, Division of Transport Planning, Department of Civil and Architectural Engineering, KTH.

Recommended acknowledgment for dataset use:

We thank Trafiklab (Trafiklab.se) for the data and the KTH iMobility Lab members who helped collect and process Trafiklab's API live-stream data into the dataset used: Matej Cebecauer, Daniel Chaves, and Anastasios Skoufas.

Table of contents

Dataset

Title

Files 

Format 

1

Observed vs. scheduled vehicle arrival/departure events (AVL + GTFS-RT TripUpdates, merged with GTFS Regional Static planned time tables) 

12 files, one per month 

CSV, gzip-compressed (.csv.gz) 

2

Raw SL GTFS-RT Service Alerts 

1 file 

Parquet (zstd) 

3

Static GTFS Regional Static reference data (Stops, Transfers, Trips, Routes) 

4 files 

CSV 

Technical info

1. Observed vs. scheduled arrival/departure events (gtfs_rt_YYYY_MM.csv.gz)

One row per (trip, stop) observation: the scheduled stop time from the static GTFS regional timetable, merged with the observed arrival/departure reported via GTFS-RT TripUpdates for that stop visit. 

 

Field 

Type 

Description 

date 

date 

Service date the trip operated on 

LineNumber 

string 

Public-facing line number (e.g. "18", "323") 

LineID 

integer 

Numeric part of route_id — SL's internal line number 

route_id 

string 

Route identifier. Join key to routes.txt for full route metadata 

TransportMode 

string 

Mode: BUS, METRO, TRAIN, TRAM, SHIP (derived from route_type in routes.txt) 

direction_id 

integer (0/1) 

Direction of travel for the trip (0 = outbound, 1 = inbound) 

trip_id 

integer 

Trip/departure identifier on route_id. Join key to trips.txt for full trip metadata 

VehicleID 

integer 

Vehicle identifier, as reported in GTFS-RT TripUpdates/VehiclePositions 

shape_id 

integer 

Foreign key to shapes.txt (route path geometry — not included in this dataset) 

planned_departure_sfm 

integer (seconds) 

Scheduled trip departure time, seconds-from-midnight of the service date, from the static GTFS regional timetable 

planned_departure_tripupdates_sfm 

integer (seconds) 

Scheduled trip departure time, seconds-from-midnight, as carried in the merged GTFS-RT TripUpdates message 

stop_sequence 

integer 

Order of this stop within the trip 

dist_traveled 

number (meters) 

Distance traveled from the previous stop to this observed stop 

previous_stop_id 

integer 

stop_id of the preceding stop. Join key to stops.txt 

previous_StopAreaID 

integer 

Parent-station identifier of the preceding stop (stops.txt → parent_station) 

observed_stop_id 

integer 

stop_id of the observed/current stop. Join key to stops.txt 

observed_StopAreaID 

integer 

Parent-station identifier of the observed stop (stops.txt → parent_station) 

scheduled_arrival_time_sfm 

integer (seconds) 

Scheduled arrival time at this stop, seconds-from-midnight 

observed_arrival_time_sfm 

integer (seconds) 

Observed (real-time) arrival time at this stop, seconds-from-midnight 

observed_arrival_delay 

integer (seconds) 

Observed arrival delay vs. schedule (+ late / − early) 

observed_arrival_uncertainty 

integer 

GTFS-RT reported uncertainty of the observed arrival estimate 

scheduled_departure_time_sfm 

integer (seconds) 

Scheduled departure time from this stop, seconds-from-midnight 

observed_departure_time_sfm 

integer (seconds) 

Observed (real-time) departure time from this stop, seconds-from-midnight 

observed_departure_delay 

integer (seconds) 

Observed departure delay vs. schedule 

observed_departure_uncertainty 

integer 

GTFS-RT reported uncertainty of the observed departure estimate 

tripupdates_and_gtfs_not_merged_within_service_id 

boolean (0/1) 

Data-quality flag: 0 = static GTFS and GTFS-RT TripUpdates matched on the same service_id/service date; 1 = not matched 

tripupdates_not_merged_with_stoptimes_or_trip 

boolean (0/1) 

Data-quality flag: 0 = merge succeeded using stop_times/trips timing information; 1 = not matched 

stop_headsign 

string 

Headsign displayed at the stop for this trip 

  

Merge quality: a row where both tripupdates_and_gtfs_not_merged_within_service_id = 0 and tripupdates_not_merged_with_stoptimes_or_trip = 0 indicates a fully successful match between the static schedule and the real-time TripUpdates report. 

2. Raw GTFS-RT Service Alerts feed (service_alerts_2024.parquet)

Unprocessed GTFS-RT Service Alerts feed, one row per (alert × informed entity) pair. 

 

Field 

Type 

Description 

id 

bigint 

GTFS-RT alert entity ID 

start 

timestamp 

Alert active-period start 

end 

timestamp 

Alert active-period end 

cause 

int 

GTFS-RT Alert.Cause enum value 

description_text2 

string 

Alert description text (Swedish) 

header_text2 

string 

Alert header/summary text (Swedish) 

route_id 

string 

Affected GTFS route, if specified 

trip_id 

bigint 

Affected GTFS trip, if specified 

schedule_relationship 

smallint 

GTFS-RT TripDescriptor.ScheduleRelationship for the informed entity 

stop_id 

bigint 

Affected stop, if specified 

date_inserted 

timestamp 

When this row was captured into the database 

date 

int 

Capture date, YYYYMMDD 

alert_message_timestamp 

timestamp 

Feed message timestamp 

Row count (2024): 1,972,779. Source: Trafiklab. 

This is the raw alert feed, one row per informed entity, so a single real-world incident typically spans many rows/alert IDs over its lifetime. No deduplication or incident-clustering has been applied. 

3. GTFS Static data

GTFS Regional Static Data 

Regional GTFS reference tables (Trafiklab GTFS Sverige 2 / GTFS Regional). Source: trafiklab.se GTFS static data overview (https://www.trafiklab.se/api/gtfs-datasets/gtfs-regional/). 

The files below extract unique values from 366 daily GTFS Regional Static file feeds available during 2024 through the Trafiklab API. (https://www.trafiklab.se/sv/api/gtfs-datasets/gtfs-sweden/static-specification/). All attribute values reflect their latest occurrences; for example, if the coordinates associated with a specific "stop_id" changed during 2024, this dataset provides only the last observed coordinates. The same rule applies to all attributes when referencing GTFS Regional static files in this dataset. For exact coordinates, trip head-sign, route names, etc. active on a specific day, we refer you to the historical database KODA (https://www.trafiklab.se/sv/api/our-apis/koda/).

 

stops.txt 

 

Field 

Type 

Description 

stop_id 

ID 

Identifies a stop, station or station entrance 

stop_name 

Text 

Name of the location, in local/tourist vernacular 

stop_lat 

Latitude 

Latitude of the location 

stop_lon 

Longitude 

Longitude of the location 

location_type 

Enum 

Type of location — see table below 

parent_station 

ID 

Defines hierarchy between locations in stops.txt — contains the stop_id of the parent location 

platform_code 

Text 

Platform identifier for a platform stop 

  

location_type values: 

 

Code 

Meaning 

0 (or blank) 

Stop/Platform — where passengers board or disembark; called a platform when it has a parent_station 

1 

Station — physical structure containing one or more platforms 

2 

Entrance/Exit — where passengers enter/exit a station from the street 

  

routes.txt 

 

Field 

Type 

Description 

route_id 

ID 

Identifies a route 

route_short_name 

Text 

Short, abstract identifier for the route (e.g. "32", "100X") 

route_long_name 

Text 

Full, more descriptive name of the route 

route_type 

Enum 

Mode of transportation used on the route — see table below 

route_desc 

Text 

Description providing useful, quality information about the route 

 

route_type values (extended/regional codes)

Code 

Name 

Example 

100 

Railway Service 

Train traffic not assignable to a more specific category 

101 

High Speed Rail Service 

SJ Snabbtåg, X2000 

102 

Long Distance Rail Service 

International (Utrikes) 

103 

Inter Regional Rail Service 

SJ InterCity 

105 

Sleeper Rail Service 

Night train (Nattåg), EuroNight 

106 

Regional Rail Service 

County traffic (Länstrafik) 

401 

Metro Service 

Metro (Tunnelbana) 

700 

Bus Service 

— 

714 

Rail Replacement Bus Service 

Replacement bus for train traffic 

900 

Tram Service 

Trams in Göteborg 

1000 

Water Transport Service 

SL Pendelbåt, Stockholm–Helsinki 

1501 

Communal Taxi Service 

On-demand public transport with advance order (Förbeställningstrafik) 

  

trips.txt 

 

Field 

Type 

Description 

route_id 

ID 

Identifies a route 

trip_id 

ID 

Identifies a trip 

trip_headsign 

Text 

Text on signage identifying the trip's destination to riders 

direction_id 

Enum 

Direction of travel for the trip — see table below 

shape_id 

ID 

Identifies the geospatial shape describing the vehicle's travel path 

 

direction_id values 

 

Code 

Meaning 

0 

Outbound — travel in one direction 

1 

Inbound — travel in the opposite direction 

 

transfers.txt

 

Field 

Type 

Description 

from_stop_id 

ID 

Stop/station where a connection between routes begins 

to_stop_id 

ID 

Stop/station where a connection between routes ends 

transfer_type 

Enum 

Type of connection for the (from_stop_id, to_stop_id) pair — see table below 

min_transfer_time 

Non-negative integer 

Time, in seconds, required to permit a transfer 

from_trip_id 

ID 

Trip where a connection between trips starts 

to_trip_id 

ID 

Trip where a connection between trips ends 
 

transfer_type values

Code 

Meaning 

0 (or blank) 

Recommended transfer point between routes 

1 

Timed transfer point — the departing vehicle waits for the arriving one 

2 

Transfer requires the minimum time specified by min_transfer_time 

3 

Transfers not possible between routes at this location 

 

Files

transfers.csv

Files (4.0 GB)

Name Size Download all
md5:89c540a03343d011f9f936f75ddb740a
343.6 MB Download
md5:91b4ac4b37c3f6fcdaedef9beb3b52ce
321.2 MB Download
md5:870c1a842cfafdbb3c66a448bfa3d108
340.9 MB Download
md5:3123dad4bb8de10432515972487d728e
334.5 MB Download
md5:3c4d82dcefeb0c08ad0f1b2b26bfc71e
343.5 MB Download
md5:e5080438a0dba0127e66028c029ebaa9
316.4 MB Download
md5:956c6e0a18553aec9831428b70f45ae4
314.2 MB Download
md5:8c50d606c49e63a0c2e36b19a52cb7a0
331.8 MB Download
md5:f9655fd51f9bc97ef6c1e11172b20505
338.1 MB Download
md5:b5c0307fa44ec09a601a71dc80918c5d
349.9 MB Download
md5:f3a132892faf93257b5960bd64892870
324.5 MB Download
md5:077ab2c4d9a38e63921cb1a704cd70cd
316.8 MB Download
md5:71aa1d3d111cdd5d350b114bad8a0c4f
21.3 kB Preview Download
md5:1dc9997c4ad0ad39b8f71297f3297887
29.0 MB Download
md5:01d962818bc291a16a0797f5675d6deb
1.4 MB Preview Download
md5:4ea76948cad0be43ed6b220be2bb4d20
3.0 MB Preview Download
md5:18854bce1551d8c58d3a1f35f0f95b7e
17.9 MB Preview Download