WTF ADS-B field notes
About WTF ADS-B
Where the observations come from, what receives them, and how stories are written and checked.
WTF ADS-B is an independent receiver project run by Nearl Crews near Oakland County International Airport (PTK), Michigan. It looks for observations worth explaining, from repeated loops and uncommon aircraft types to numbers that disagree, and each story shows the measurements behind it. A flight pattern alone cannot say why an aircraft was there.
How the data arrives
Aircraft broadcast their identity, position, altitude, and speed on 1090 MHz and, in the United States, on 978 MHz (UAT). Two software-defined radios listen on those bands through a filtered amplifier and two 26-inch antennas. The receiver software also accepts multilateration (MLAT) estimates from cooperating receivers, and stories keep those apart from positions an aircraft broadcast. Reception gaps limit what can be reconstructed: the journal is a selection of observations, not a complete record of traffic around PTK.
The software
Ultrafeeder decodes 1090 MHz with readsb and keeps tracks in tar1090; dump978 decodes 978 MHz. Two open-source projects do the rest: wtf-adsb-engine turns the saved aircraft records into checked findings, and wtf-adsb-publisher selects, writes, and publishes the stories to a Ghost site with a custom theme. Both are MIT licensed and run as containers.
Writing and review
Stories publish automatically, at most nine a day with room for rare exceptional events. Each one starts as template text built from the recorded data, pinned reference tables, earlier stories here, and researched background about the aircraft type. That background comes from an AI research step that may cite only an allowlist of publishers, and every researched fact appears with its source. An AI writer may rewrite the headline, summary, and story text from the same facts; a validator then checks every number, name, date, and place against those facts and rejects text that mentions the receiver's location, guesses who was aboard, or claims an emergency or a mission without hedging. A story that fails keeps its template text.
Measurements, record limits, and aircraft links come straight from the data and are never written by AI. A current registration or type match may differ from an aircraft's historical assignment, and a database association does not say who was aboard.
Data sources and credits
- Aircraft type and operator data: the Mictronics aircraft database, under the Open Data Commons Attribution License 1.0 (ODC-By).
- ICAO address allocation ranges: the table in tar1090, which transcribes ICAO's allocation plan.
- Transponder code meanings: the FAA's National Beacon Code Allocation Plan and Aeronautical Information Manual.
- Watchlist categories: Plane Alert DB, under the Open Database License 1.0 and Database Contents License 1.0. A category describes the list entry, not the people aboard.
- Organization names and descriptions behind watchlist links: Wikidata, under CC0 1.0. People are never described this way.
- Upper winds for airspeed comparisons: Open-Meteo, weather data by Open-Meteo.com under the Creative Commons Attribution 4.0 license (CC BY 4.0).
- Photos: representative aircraft photos carry their own credits and licenses; photos of an observed aircraft come from Planespotters.net, with the photographer named beside the image. That photo was taken on another occasion, and your browser loads it from Planespotters.net directly.
A type's usual ground speed, quoted in speed-record and type-mismatch stories, is the 10th to 90th percentile of the median ground speeds this receiver logged for that type in one flight phase and altitude band over the 30 days before the story was first written, counting each aircraft once, leaving out the story's own aircraft, and requiring at least 20 other aircraft seen over 14 days. It describes these records, not what the type can do.
Aircraft on a map
Maps show the saved positions, centered on the recorded path, on topographic backgrounds from USGS The National Map (map services and data available from the U.S. Geological Survey, National Geospatial Program). Each background is saved with the track image, so the map works without a live service; it is a current reference map, not the ground on the flight date. Nearby-place headings and fallback maps use Natural Earth, the 2025 U.S. Census Gazetteer, and OurAirports; a place name or airport gives context and does not establish where a flight began or ended.
The scale bar is approximate at the map's middle latitude, and the recorded path length sums the accepted positions in each connected segment, so gaps and position errors affect it. In a replay the marker follows the saved positions, movement between reports is interpolated, the readings shown belong to the saved report, and the marker hides across gaps. Altitude with an unspecified reference is labeled as such. The receiver's exact position, marker, and coverage rings stay private; gaps stay visible, and a selected part of a flight is labeled as such.
Support the project
If you find these observations useful, you can support WTF ADS-B: Buy me a coffee.
How publication works
Qualifying observations publish automatically after evidence and privacy checks. Each week, one selected observation receives additional sourced background. The archive describes this selected record; it does not establish complete receiver history.