Hugging Face's logo

Datasets:
alreadyopen
/
world-flights

Modalities:
Geospatial
Tags:
geospatial
aviation
ads-b
geoparquet
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
built: timestamp[s]
counts: struct<days: int64, aircraft: int64, points: int64>
  child 0, days: int64
  child 1, aircraft: int64
  child 2, points: int64
files: struct<index.json: struct<sha256: string, bytes: int64>, latest.json: struct<sha256: string, bytes:  (... 3076 chars omitted)
  child 0, index.json: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 1, latest.json: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 2, v/2026-10-09/replay/00.bin: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 3, v/2026-10-09/replay/01.bin: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 4, v/2026-10-09/replay/02.bin: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 5, v/2026-10-09/replay/03.bin: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 6, v/2026-10-09/replay/04.bin: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 7, v/2026-10-09/replay/05.bin: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 8, v/2026-10-09/replay/06.bin: struct<sha256: string, bytes: int64>
      child 0, sha256: string
      child 1, bytes: int64
  child 9, v/2026-10-09/replay/07.bin
...
: string, used_by:  (... 8 chars omitted)
  child 0, item: struct<name: string, owner: string, url: string, licence: string, docs: string, used_by: string>
      child 0, name: string
      child 1, owner: string
      child 2, url: string
      child 3, licence: string
      child 4, docs: string
      child 5, used_by: string
days: list<item: struct<day: timestamp[s], release: string, source: list<item: string>, aircraft: int64, p (... 175 chars omitted)
  child 0, item: struct<day: timestamp[s], release: string, source: list<item: string>, aircraft: int64, points: int6 (... 163 chars omitted)
      child 0, day: timestamp[s]
      child 1, release: string
      child 2, source: list<item: string>
          child 0, item: string
      child 3, aircraft: int64
      child 4, points: int64
      child 5, replay_points: int64
      child 6, acas_events: int64
      child 7, files: list<item: string>
          child 0, item: string
      child 8, bytes: int64
      child 9, kaggle: string
      child 10, bucket_files: list<item: string>
          child 0, item: string
      child 11, bucket: list<item: string>
          child 0, item: string
replay_layout: string
licence: string
expected_daily_volume: struct<parquet_gb: int64, heatmap_raw_gb: int64, replay_gb: double, note: string>
  child 0, parquet_gb: int64
  child 1, heatmap_raw_gb: int64
  child 2, replay_gb: double
  child 3, note: string
hf_days: int64
new: list<item: string>
  child 0, item: string
new_day: timestamp[s]
to
{'built': Value('timestamp[s]'), 'counts': {'days': Value('int64'), 'aircraft': Value('int64'), 'points': Value('int64')}, 'licence': Value('string'), 'hf_days': Value('int64'), 'expected_daily_volume': {'parquet_gb': Value('int64'), 'heatmap_raw_gb': Value('int64'), 'replay_gb': Value('float64'), 'note': Value('string')}, 'replay_layout': Value('string'), 'days': List({'day': Value('timestamp[s]'), 'release': Value('string'), 'source': List(Value('string')), 'aircraft': Value('int64'), 'points': Value('int64'), 'replay_points': Value('int64'), 'acas_events': Value('int64'), 'files': List(Value('string')), 'bytes': Value('int64'), 'kaggle': Value('string'), 'bucket_files': List(Value('string')), 'bucket': List(Value('string'))}), 'new_day': Value('timestamp[s]'), 'new': List(Value('string'))}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2951, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 547, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              built: timestamp[s]
              counts: struct<days: int64, aircraft: int64, points: int64>
                child 0, days: int64
                child 1, aircraft: int64
                child 2, points: int64
              files: struct<index.json: struct<sha256: string, bytes: int64>, latest.json: struct<sha256: string, bytes:  (... 3076 chars omitted)
                child 0, index.json: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 1, latest.json: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 2, v/2026-10-09/replay/00.bin: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 3, v/2026-10-09/replay/01.bin: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 4, v/2026-10-09/replay/02.bin: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 5, v/2026-10-09/replay/03.bin: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 6, v/2026-10-09/replay/04.bin: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 7, v/2026-10-09/replay/05.bin: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 8, v/2026-10-09/replay/06.bin: struct<sha256: string, bytes: int64>
                    child 0, sha256: string
                    child 1, bytes: int64
                child 9, v/2026-10-09/replay/07.bin
              ...
              : string, used_by:  (... 8 chars omitted)
                child 0, item: struct<name: string, owner: string, url: string, licence: string, docs: string, used_by: string>
                    child 0, name: string
                    child 1, owner: string
                    child 2, url: string
                    child 3, licence: string
                    child 4, docs: string
                    child 5, used_by: string
              days: list<item: struct<day: timestamp[s], release: string, source: list<item: string>, aircraft: int64, p (... 175 chars omitted)
                child 0, item: struct<day: timestamp[s], release: string, source: list<item: string>, aircraft: int64, points: int6 (... 163 chars omitted)
                    child 0, day: timestamp[s]
                    child 1, release: string
                    child 2, source: list<item: string>
                        child 0, item: string
                    child 3, aircraft: int64
                    child 4, points: int64
                    child 5, replay_points: int64
                    child 6, acas_events: int64
                    child 7, files: list<item: string>
                        child 0, item: string
                    child 8, bytes: int64
                    child 9, kaggle: string
                    child 10, bucket_files: list<item: string>
                        child 0, item: string
                    child 11, bucket: list<item: string>
                        child 0, item: string
              replay_layout: string
              licence: string
              expected_daily_volume: struct<parquet_gb: int64, heatmap_raw_gb: int64, replay_gb: double, note: string>
                child 0, parquet_gb: int64
                child 1, heatmap_raw_gb: int64
                child 2, replay_gb: double
                child 3, note: string
              hf_days: int64
              new: list<item: string>
                child 0, item: string
              new_day: timestamp[s]
              to
              {'built': Value('timestamp[s]'), 'counts': {'days': Value('int64'), 'aircraft': Value('int64'), 'points': Value('int64')}, 'licence': Value('string'), 'hf_days': Value('int64'), 'expected_daily_volume': {'parquet_gb': Value('int64'), 'heatmap_raw_gb': Value('int64'), 'replay_gb': Value('float64'), 'note': Value('string')}, 'replay_layout': Value('string'), 'days': List({'day': Value('timestamp[s]'), 'release': Value('string'), 'source': List(Value('string')), 'aircraft': Value('int64'), 'points': Value('int64'), 'replay_points': Value('int64'), 'acas_events': Value('int64'), 'files': List(Value('string')), 'bytes': Value('int64'), 'kaggle': Value('string'), 'bucket_files': List(Value('string')), 'bucket': List(Value('string'))}), 'new_day': Value('timestamp[s]'), 'new': List(Value('string'))}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

World flights

Every aircraft position adsb.lol's feeders received, one day at a time, plus near-live snapshots around Hong Kong. Rebuilt by AlreadyOpen/world-flights: the daily bake runs at 04:30 UTC, after adsb.lol publishes the previous UTC day (about 03:26 UTC); the Hong Kong snapshot runs every 15 minutes.

Where the data is

  • Full history: Kaggle, one public dataset per day, created once and never changed or deleted: https://www.kaggle.com/datasets/helenkwok/world-flights-YYYY-MM-DD, e.g. helenkwok/world-flights-2026-10-09. Each holds the complete day: every file listed below, at the dataset's root.
  • Index of all days: helenkwok/world-flights on Kaggle (updated daily): index.json with every day's dataset slug, aircraft, points and size, latest.json, and the newest complete Hong Kong day (live/hk-YYYY-MM-DD.parquet).
  • Hugging Face, for maps: the last 7 days only. The bucket huggingface.co/buckets/alreadyopen/world-flights (CORS and HTTP Range) holds v/<YYYY-MM-DD>/traces-NNN.parquet and v/<YYYY-MM-DD>/replay/ for the newest 7 days, index.json, latest.json and live/hk*. A day leaves the bucket only after its Kaggle dataset is listed by Kaggle. This dataset repo holds only this card, index.json and manifest.json: no copy of the data.

Files

  • v/<YYYY-MM-DD>/traces-NNN.parquet: GeoParquet (geometry = POINT(lon lat), WGS 84), one row per trace point, every point and every field of adsb.lol's daily globe_history release (planes-readsb-prod-0). Split into parts of about a gigabyte of source JSON each; an aircraft is never split across parts. Columns:
    • per aircraft (repeated on each of its rows): icao (24-bit address, hex; ~ prefix = non-ICAO), registration, type (ICAO type code), description, owner_operator, year, db_flags (readsb bitfield: 1 military, 2 interesting, 4 PIA, 8 LADD), no_reg_data, readsb_version, day_start (Unix seconds, 00:00 UTC of the day)
    • per point (readsb's trace array, in its order): t (seconds after day_start; time = the same as a timestamp), lat, lon, altitude (ft, barometric unless flag 8; NULL when on the ground or unknown), on_ground, ground_speed (kt), track (degrees; can be true heading when on the ground), flags (1 position stale, 2 start of a new leg, 4 vertical rate is geometric, 8 altitude is geometric), vert_rate (ft/min), details (readsb's full aircraft object as JSON, present when it changed: flight, squawk, category, nav_*, nic/nac/sil, wind, temperatures, ...), source (adsb_icao, mlat, tisb_*, adsr_*, ...), alt_geom, geom_rate, ias, roll
  • v/<YYYY-MM-DD>/replay/HH.bin (one file per UTC hour, so a browser loads one hour at a time) and replay/index.json (hours with their aircraft and point counts and file sizes): the same day for a map that replays it. Downsampled: per aircraft, only the first position in each 30-second bucket is kept, and only t, lon, lat, altitude, ground_speed and track (rounded: 1e-5 degree, whole feet, whole knots, 0.1 degree). Everything else is in the Parquet, which is complete. An aircraft flying across hours appears in each hour's file; hour 24 holds the few points readsb logged just after midnight. Layout (no library needed: DataView and typed arrays): AOR1, a uint32 header length, a JSON header (hour, aircraft: [icao, registration, type, first point, count] in icao order, columns, layout), then one little-endian column after another: t int32, lon int32, lat int32, altitude int32 (-999999 ground, -2147483648 unknown), ground_speed int16 and track int16 (-1 unknown). An aircraft's points are contiguous and time-ordered.
  • v/<YYYY-MM-DD>/raw/: every other file of the release, as delivered: acas/acas.json.gz, acas/acas.csv.gz, heatmap/NN.bin.ttf (48 half-hour files), README.txt, LICENSE-ODbL.txt, LICENSE-cc0.txt.
  • v/<YYYY-MM-DD>/acas.parquet: GeoParquet of raw/acas/acas.json.gz, one row per ACAS/TCAS resolution advisory message (8,391 on 2026-10-09), every field: record is the whole line (readsb's aircraft object plus acas_ra with time, advisory, ARA/RAT/MTE/RAC, threat id and the raw message bytes), with time, hex, flight, registration, type, lat, lon, alt_baro, advisory, advisory_complement, threat_id_hex and full_bytes as columns. acas.csv.gz holds the same events in a shorter, headerless text form; it is kept as delivered and not parsed separately.
  • The heatmap files are not parsed. They are readsb's binary heatmap (gzip, then little-endian int32 records of 16 bytes: 10-second slice markers, positions and callsign/squawk records). The format is defined only by readsb's and tar1090's source code, not documented, so this dataset keeps the originals rather than guess at fields. The trace Parquet holds the same positions at full resolution.
  • index.json (the day's entry; the top-level index.json lists every day): counts, file names and sizes, the Kaggle URL, and bucket URLs while the day is on Hugging Face. latest.json: the newest day.
  • live/hk-YYYY-MM-DD.parquet (UTC day): GeoParquet, one row per aircraft per snapshot of the adsb.lol API within 250 nm of (22.3 N, 114.2 E), every 15 minutes; aircraft holds the full API object as JSON, with common fields as columns. Built up on the bucket during the day, then archived in that day's Kaggle dataset. live/hk.json (bucket only): the latest response.
  • manifest.json (Hugging Face): sha256 and size of the files of the latest daily bake, the counts and the sources.

Kaggle stores .gz files decompressed, so the Kaggle copies of raw/acas/*.gz arrive without .gz, at full size.

Expected volume (estimate from 2026-10-09, a 4.1 GB release; also in index.json expected_daily_volume): about 6 GB of trace Parquet per day, plus about 1 GB of raw heatmap files and 0.7 GB of replay files: roughly 8 GB per day on Kaggle (about 2.9 TB a year; Kaggle allows 200 GB per dataset), and about 7 x 6.7 GB = 47 GB on the Hugging Face bucket.

Source and licence

Data from adsb.lol contributors (historical data, GitHub releases, API), made available under the Open Database License (ODbL) 1.0. Credit: "adsb.lol contributors, ODbL 1.0". Feeders waive their rights to the raw data under CC0; adsb.lol releases the database under ODbL.

Every file, here and in the Kaggle datasets, is a derived database under the ODbL: if you publicly use or share a database adapted from it, you must release that adapted database under the ODbL (share-alike), credit adsb.lol contributors, and keep it open. That is why this dataset holds adsb.lol data only and is kept apart from AlreadyOpen datasets under other licences; combining them is your choice and your share-alike obligation. Aircraft registration and type fields come with readsb's database as bundled in the release. Provided as is, with no warranty, and not endorsed by adsb.lol.
Downloads last month
-
Total file size:
29.3 kB