Skip to main content
← Getting The Data

Open Flight Data

The whole following-planes claim was built from screenshots. Somebody photographed an aircraft-history table on a tracking site, somebody typed the rows into a spreadsheet, somebody tallied the spreadsheet, and somebody read the tally on television. Every step is a place a row could be misread, and published audits say a great many of them were.

This page is what happened when we skipped the chain and went to the primary data ourselves, using only sources that are free, need no account, and can be repeated by any reader with a terminal.

Everything below was retrieved on 24 August 2026. Anything said about an API without a date on it is worthless six months later.

What APIs they have​

The single most important thing on this page​

Almost every free ADS-B network serves live positions only. Ask adsb.lol, adsb.fi, airplanes.live or anonymous OpenSky about an aircraft that is not airborne at this second and you get HTTP 200 with an empty array — which reads exactly like "no such aircraft" and actually means "not flying right now."

A researcher who does not know that will look up SU-BTT, get nothing, and conclude the aircraft was hidden. It was parked.

SourceEndpointHistorical?AuthStatus on 24 Aug 2026
adsb.lol globe historyglobe_history/YYYY/MM/DD/traces/hh/trace_full_HEX.jsonYESnone200 — the workhorse
adsb.lol v2/v2/hex/HEX, /v2/reg/REGno, live onlynone200, empty when not airborne
adsb.fi/api/v2/hex/HEXno, live onlynone200, empty when not airborne
airplanes.live/v2/reg/REGno, live onlynow gated403 — email them first
OpenSky/api/flights/aircraftyesOAuth2403 anonymous
OpenSky/api/states/allno, live onlyoptional200
adsbdb.com/v0/aircraft/REGno — metadata onlynone200
ADS-B Exchangeglobe_history/...yespaid403
OurAirportsairports.csv, runways.csvreference datanone200, CC0
Internet ArchiveCDX APIsnapshotsnone200

adsb.lol globe history — the one that works​

The only free, no-account source of historical ADS-B tracks we have found:

https://adsb.lol/globe_history/YYYY/MM/DD/traces/<last 2 of hex>/trace_full_<hex>.json

Gzip-compressed JSON. It self-identifies the aircraft in r (registration) and t (type) from the feeder database, and trace is an array of [seconds after timestamp, lat, lon, alt, ...] where alt is the literal string ground when the aircraft is on the surface — so takeoff and landing times fall straight out of it.

How you use them​

Everything is in apis/public_open_source/code/. No dependencies, Node 18+.

node globe_history.js SU-BTT 2025-09-10          one aircraft, one day
node globe_history.js --following 2025-09-04 2025-09-11   the fleet, a range
node pull_all.js --apply                          every claimed date in the spine CSVs
node adsbdb.js                                    tail to hex, type, owner
node live_networks.js SU-BTT                      the live-only networks side by side
node wayback.js                                   Internet Archive snapshots
node ourairports.js                               public-domain airport reference data

Every script saves the raw response with a .meta.json beside it recording the URL, the HTTP status, the byte count and the UTC time of the request. Nothing overwrites a previous pull — a re-pull lands beside the first with a timestamp suffix, because the diff between two pulls is how we show something stopped being retrievable.

OpenSky changed its authentication on 18 March 2026: username and password are gone, it is OAuth2 client credentials now, tokens last 30 minutes, and there is a daily credit budget — 400 anonymous, 4,000 registered, 8,000 for accounts that feed data back. Without credentials, historical calls return 403 You cannot access historical flights.

Command-line tools installed for this work: duckdb for querying the spine CSVs and the FAA registry in place, and monolith for freezing a whole tracking page into one self-contained file. The ADS-B decoders readsb and dump1090-fa are installable but decode radio off the air — they do not query anybody's API.

What we got so far​

478 (aircraft, date) pulls on 24 August 2026. 178 traces retrieved. Landed in three places: the raw archive, each aircraft's own directory, and each overlap's own directory.

The September 10 departure, settled​

The disputed question of when SU-BTT left Utah has a primary answer now.

First seen, on the ground13:07:53 UTC — 07:07 MDT — at 40.2303 N, 111.7283 W, Provo Municipal (KPVU)
Wheels up13:14:04 UTC — 07:14 MDT
Last seen, on the ground16:50:37 UTC — 12:50 EDT — at 39.6631 N, 75.6046 W, Wilmington New Castle (KILG)
Trace points1,229

That is Provo to Wilmington, Delaware, departing about five hours before the assassination and landing about ninety minutes before it. It matches the "7:08 AM MT" figure that has circulated almost exactly, and it is inconsistent with the competing accounts of a Dugway departure.

Source: adsb.lol/globe_history/2025/09/10/traces/d3/trace_full_0101d3.json, retrieved 24 August 2026, saved at Planes/SU-BTT/data/adsb/.

What it does not show. It says where an airframe was. It says nothing about who was aboard, who tasked it, or why it went to Delaware. No public record establishes tasking in either direction.

The result that matters most​

We ran every exactly-claimed date in this repo's own spine files through the primary archive, without looking at anyone's audit first. Then we compared.

The repo's audit verdict on the claimPrimary trace found
accurate18 of 24 — 75%
partial7 of 8 — 88%
inaccurate5 of 44 — 11%

An independent primary-data check, run blind, lands in the same place the audit did. The pairings the audit called sound are the ones with tracks behind them; the pairings it called wrong are overwhelmingly the ones with no track at all.

And underneath that sits a second, sharper split:

  • Flight legs — the individual "this aircraft was at this field on this date" rows — verify at 141 of 144, 98%.
  • Overlap pairings — the "and therefore it was following them" rows — verify at 30 of 79, 38%.

The aircraft movements are largely real. The claim that those movements constitute shadowing is where the sheet falls apart. That is a more precise finding than either side of this argument has published, and it does not flatter either of them.

What we did NOT get​

The most useful section on the page.

  • 300 of 478 pulls returned nothing. 52 of those were exactly-claimed dates.
  • The archive does not reach 2022. The earliest trace we obtained anywhere is 24 February 2023. The trackers date the pattern from 2022, and those rows cannot be tested for free at all. Any statement about 2022 on this site rests on somebody's screenshot, not on data we hold.
  • A whole band of the archive returns 403. Every date from 12 October 2025 to at least 15 December 2025 returns HTTP 403 for every aircraft we tried, including one with no connection to this case. 11 October works; 31 December works. This is an archive-wide condition, not a suppression of any aircraft, and it must never be reported as one. It happens to cover the dates a tracker claims SU-BTT was photographed on the Sharm el-Sheikh tarmac — which means the free archive cannot check that claim either way.
  • airplanes.live is no longer open. Its API now returns 403 and asks you to email a description of your project first. It was open before. That is a real reduction in what a member of the public can retrieve, and it is why this page carries dates.
  • ADS-B Exchange historical is paid. globe_history returns 403 to the public — and several of the original screenshots were taken from that site while it was open.
  • OpenSky historical is unavailable to us. No credentials. Its flight-list endpoint is the best free cross-check on adsb.lol and we are not using it. This is the cheapest gap on the list to close.
  • The tracking pages behind this claim were never archived. An Internet Archive sweep of 30 target URLs on 24 August 2026 found 22 with zero snapshots — including both the Flightradar24 and the ADS-B Exchange pages for SU-BTT, the lead aircraft of the whole claim, while every other Egyptian tail has at least one. See capturing the tracking websites.
  • No Erika Kirk aircraft was checked, because there is nothing to check. Her flight logs are reported erased, and this repo holds one sourced Erika location before 10 September 2025 — for a window the trackers counted 73 overlaps across. Every Erika pairing rests on the foreign aircraft's track plus a claimed location for her. See Erika Kirk's flight logs reported erased.
  • The hex code for SU-BTT is disputed between two community databases. adsbdb returns 0101F0; the trace at 0101D3 self-identifies as SU-BTT and carries the September 10 track, while 0101F0 has no trace that day. We use 0101D3 and record the disagreement rather than resolving it silently.

An empty result is not a finding​

Every gap above is recorded in a MISSING_DATA.md file inside the aircraft's directory and the overlap's directory — 90 of them — naming the tail, the UTC date, the endpoint, the HTTP status, and who claimed the flight.

Each one says the same thing, and this page says it too: a 404 means the network holds no trace. It does not establish that the aircraft did not fly, and it does not establish that a transponder was switched off. The ordinary explanations come first — parked and silent, outside volunteer receiver coverage, or a claimed date that was simply wrong. This network sees only what its volunteers heard, and most of the Atlantic, most of North Africa and much of the rural United States at low altitude is not covered by anyone.

The holes cut both ways, and they cut against us as often as for us.

Update — three of the holes above have since been closed​

Retrieved 24 August 2026, after the pass described above. Assuming the data had been erased, we went looking for backups. Three of the gaps listed in "What we did NOT get" turned out to be recoverable, and this section corrects them rather than leaving the older text to stand. The full account is on Flight Data Recovery.

SourceHistory?AuthStatus on 24 Aug 2026
globe.airplanes.live globe_historyYESnone200 — 5–10× the byte count of adsb.lol for the same aircraft-day
adsb.lol GitHub releases (ODbL)YESnone200 — one ~3 GB release per day, including days the live API refuses
samples.adsbexchange.com/tracesYES, monthlynone200 — one day per month, back to July 2016
  • The 12 October – 15 December 2025 403 band is recoverable. adsb.lol publishes its entire archive to github.com/adsblol/globe_history_2025 as one Open-Database-Licence release per day, and every date in the band is present. globe.airplanes.live serves the same days over HTTP. The 403 is a serving condition, not a deletion, and this page should never have implied otherwise. It also means the free archive can now check the 13 October 2025 Sharm el-Sheikh tarmac claim.
  • 2022 is now partly reachable. ADS-B Exchange's free monthly sample archive reaches July 2016. It produced this investigation's first primary 2022 traces — SU-BTT departing Wilmington, Delaware on 1 June 2022 and arriving Wichita, Kansas on 1 September 2022. It is one day in thirty, so it cannot test a claim about a specific mid-month date, and the great majority of 2022 rows remain untestable.
  • The N102DZ FlightRadar24 history was recovered from the Internet Archive — twelve flight rows from a snapshot taken 11 September 2025, eight months before the reported removal. The reported removal itself could not be confirmed — the FlightRadar24 page loads normally in a browser, and the HTTP 403 a script gets is a robot block that hits control aircraft identically. The archived copy is still worth holding, because FR24 shows the public only seven days and that window has long since rolled past.

And one hole that got wider, correctly. adsb.lol holds no 2026 data at all. That was tested against control aircraft with no connection to this case — an unrelated Ryanair and an unrelated Lufthansa airframe return 404 identically — so it is archive-wide retention, not suppression of anything. Naming the control test matters more than the result: it is what separates a retention window from a cover-up, and without it the two look the same.

airplanes.live remains 403 on its live /v2/ API — the gating recorded above is real. Its historical globe_history path is open. Those are two different services on one domain and this page previously conflated them.