Learjet 60 N722BD loss of control and crash, Greenacres, Florida, 1 November 2026

HEXBIRD formal assessment · version 19 · data as of 2 November 2026, 03:50Z · revised in numbered versions as the analysis changes; the page does not pull live data · times in UTC (local time was EST, UTC−5)

Controlling analytic message

Key judgments

The weather encounter likely contributed to the upset

HEXBIRD assesses that it is likely the weather encounter contributed to starting the upset.

The rationale is timing. The aircraft flew for more than 2 hours without any recorded anomaly. The upset began a little over a minute after the crew reported icing and moderate turbulence, and 17 seconds after they reported precipitation and asked to climb. The close timing favours a weather contribution but does not exclude an unrelated fault. Some faults that could start an upset, such as icing of airspeed sensors, would themselves be weather-linked.

Confidence is low. Radar observations made independently of the crew show convection and variable flow aloft near the upset point, but they do not establish the conditions at the upset or identify the starting mechanism. A fault unrelated to the weather would produce a similar track, and there is no recorder or wreckage evidence.

The mechanism cannot yet be assessed

HEXBIRD cannot presently assess what started the upset or why the crew did not recover. The live candidates are a turbulence or updraft upset (H1), an icing effect on airspeed sensors (H2), an aerodynamic stall or a response to a real or false stall warning (H3), engine power loss (H4) and a flight control, trim or autopilot fault (H5). These can combine in sequence. The analysis of competing explanations sets out the evidence, including the possible reasons the crew did not recover.

At least one person aboard was communicating during the dive

HEXBIRD assesses that it is almost certain at least one person aboard was conscious and communicating from 17:48:27Z to 17:49:02Z, the first 47 seconds of the dive.

Confidence is high. The radio calls were verified by ear and give descent rates that match ADS-B. They show communication, not what either pilot was doing with the controls, and they do not exclude partial or transient impairment. A later transmission, from 17:49:17Z to 17:49:24Z, may also have come from the aircraft, but its words and speaker are uncertain, so the judgment does not rely on it.

The 17:45Z icing report came from N722BD

HEXBIRD assesses that it is almost certain the 17:45Z PIREP (pilot report) came from N722BD.

Confidence is high. No other aircraft was recorded near the reported position at a similar level (finding 5), and 2 minutes later the crew reported the same conditions to Miami Center. That radio report comes from the same crew, so it supports the attribution rather than independently confirming the weather.

No evidence of deliberate targeting

HEXBIRD identified no flight-specific evidence that N722BD was deliberately directed at Mar-a-Lago, Palm Beach International Airport (PBI) or the President. The recorded route and communications point to an ordinary flight that met weather and then had an emergency:

• Miami Center had cleared the aircraft direct to Freeport
• it had earlier tracked directly toward Freeport
• its turn north was an approved weather deviation
• the crew asked to climb, not descend
• the crew declared a Mayday and described an uncontrolled descent
• the dive turned south, away from Mar-a-Lago

The dive did pass within 0.8 nm of PBI. Track data alone cannot exclude deliberate action, so it stays in the analysis of competing explanations. A crew could not easily have known whether the President had left: the schedule gave no departure time, and the restriction around Palm Beach was published as active until 18:30Z.

The President had highly likely left Palm Beach

HEXBIRD assesses that it is highly likely the President had left Palm Beach about 43 minutes before the crash. The presidential C-32A departed PBI at about 17:06Z, and he is inferred to have been aboard.

Confidence is moderate. The White House schedule, seen only as a search snippet, and the C-32A's departure agree. No pool report or wheels-up time yet confirms he was aboard. The distances involved are in findings 15 and 16.

Key findings

These are source facts, measurements or derived results. They carry measurement caveats, not probability terms. Exact radio wording is in appendix A.

Aircraft and route

1. The aircraft was Learjet 60 N722BD, serial 60-271, Mode S hex A9AB17, according to the FAA registry. Its ADS-B track ends at the crash site. The aircraft is on the FAA's limited aircraft data display list, and this report does not name its owner.
2. It took off from Austin (KAUS) at about 15:37Z for North Eleuthera (MYEH), according to Flightradar24's flight history, and cruised at FL410 over the Gulf of Mexico. Miami Center cleared it direct to Freeport at 17:13Z and again at 17:31Z, warning of moderate to extreme precipitation about 100 miles across along that route, and at 17:36Z told it to descend to FL350. The filed flight plan has not been seen; "Freeport" in these clearances is consistent with the Freeport VOR (ZFP), a routing point toward the Bahamas.
3. From about 17:38:40Z the aircraft tracked 079° to 080°, the direct course to Freeport. At about 17:40Z Miami Center approved deviations left and right of course, and the aircraft turned left onto about 051° for about 7 minutes. From about 17:47Z it turned right onto 097° to 099°, "turning direct Freeport". The direct line from FEMID, the fix near Naples on the cleared route, to Freeport crosses the Florida coast near Boca Raton, about 20 nm south of PBI. The northward deviation put the aircraft over western Palm Beach County, and from there the course back to Freeport crossed Palm Beach. From the upset point Freeport lay at 096°, Mar-a-Lago at 102° and PBI at 103°.

Crew reports and the upset

4. At 17:47:02Z the crew told Miami Center they had "light clear ice that is manageable and moderate turbulence", and at 17:47:34Z gave the outside air temperature as −47 °C. A further phrase at 17:47:22Z, "plus or minus … 700 feet, 600 feet", is incomplete and its meaning is uncertain. At 17:47:58Z they reported "picking up precipitation" and asked to climb to FL370. They were cleared to climb at 17:48:07Z, with an instruction to turn no further right than their present heading because of traffic, and read it back at 17:48:13Z.
5. An official PIREP for 17:45Z describes a Learjet 60 at FL350, 25 nm west of the PBI beacon, with light clear icing at −46 °C. Between 17:38Z and 17:52Z no other aircraft of any type was recorded within 24 nm of that point between FL280 and FL430; N722BD passed 5 nm from it. The 12Z Miami radiosonde gives about −48 °C at FL350, and other aircraft reported −48 to −53 °C nearby by datalink.
6. From 17:45:50Z to 17:47:05Z, on a steady track at FL350, groundspeed fell from 474 to 450 kt, and to 432 kt during the right turn; the data cannot tell whether this was a wind change or a loss of airspeed. A sustained climb began at about 17:47:54Z, before the clearance and at or shortly before the request, and ended at about 17:48:15Z. Pressure altitude reached 35,450 ft (GPS 37,300 ft), about 450 ft above FL350, and the reported climb rate peaked at 1,600 ft/min. In the previous 2 minutes pressure altitude had stayed between 34,950 and 35,075 ft. Vertical air motion, a control input or both could produce this climb; the data cannot tell which, and the archived trace has no selected altitude.
7. This report dates the upset from 17:48:15Z, when the GPS vertical rate crossed zero, and calls that position the upset point. For about 21 seconds the track stayed at 099° to 100°, which suggests the wings were roughly level. The flight path angle reached −23° by 17:48:37Z and held at −21° to −23° until about 17:48:51Z. A tightening right spiral followed, reaching −40° near 21,000 ft and −65° to −68° from 17:49:25Z. The last position was at 17:49:32.75Z, at 26.6327N 80.1127W beside Military Trail, with a GPS height of 325 ft; impact was at about 17:49:33Z. From 17:49:25Z, the last 8 seconds, the aircraft flagged its position integrity as unknown.
8. From 17:48:27Z to 17:49:02Z the crew declared a Mayday and reported descending at 10,000 and then 20,000 ft/min, an overspeed, and "descending [through] 22,000 feet … banking 60 degrees". The descent rates and altitude match ADS-B; ADS-B does not measure bank angle. From 17:49:28Z Miami Center called repeatedly without reply, and Southwest 918's relay on 121.5 MHz drew no answer.
9. HEXBIRD's speed estimates (method in the chart caption) put Mach at 0.73 to 0.77 at the upset. Mach passed the Learjet 60's maximum operating Mach number, 0.81 at FL350, between about 17:48:17Z and 17:48:35Z and peaked at 0.95 to 0.98, near 12,000 ft, in every horizontal-wind and temperature case tested. No five-second-smoothed estimate reached Mach 1, but brief supersonic flight is not conclusively excluded. On the central estimate the aircraft stayed above the type's altitude-dependent speed limit from about 17:48:33Z to the end of the data. Peak calibrated airspeed was about 577 kt near 6,500 ft, where the maximum operating speed is 300 kt indicated: about 1.9 times the limit speed and about 4 times its dynamic pressure. The limits come from the Learjet 60 type certificate data sheets (EASA IM.A.212 and FAA A10CE); they apply to pressure altitude and indicated airspeed, so comparisons with these estimates are approximate.
10. The transponder code stayed 4021 throughout, with no emergency, radio-failure or hijack code. The SPI (IDENT) flag was set twice during the dive, from about 17:48:51Z to 17:49:11Z and from 17:49:14Z to the end of data; its cause is not known. Pressure altitude stayed within −612 to +145 ft of the value predicted from the radiosonde down to 9,000 ft, which counts against a sustained blockage of the static source; it cannot test for a blocked pitot tube. From 17:49:23Z, the last 10 seconds, pressure altitude froze at 8,750 ft while GPS altitude kept falling.

Weather

11. Convective SIGMET 45E, valid 16:55Z to 18:55Z, covered thunderstorms over South Florida including the upset point, "tops above FL450". Lantana (KLNA) reported a thunderstorm in the vicinity at 17:35Z and distant lightning in all quadrants at 17:55Z. Satellite imagery puts the upset point under a broad anvil whose top was about 48,000 ft, roughly 11,000 to 12,000 ft above the aircraft.
12. At FL350 near the upset point, radar showed light to moderate echo before the upset and stronger echo after it. Before: 12.5 dBZ median and 19 dBZ maximum within 2 km on the Miami 5.1° sweep at 17:44:15Z, and 25 and 29.5 dBZ on the Melbourne 3.1° sweep at 17:44:56Z. After: 31 and 37 dBZ on the Miami sweep at 17:49:29Z. The echo top at the point lay roughly between 39,000 and 49,000 ft, and low-level reflectivity under the track reached 51 to 57 dBZ between 17:45Z and 17:50Z.
13. On the Miami 5.1° sweep at 17:44:15Z, 4 minutes before the upset, radial velocity along the beam through the later upset point increased by about 13 m/s over 5 km, at about 36,000 to 43,000 ft. This is consistent with outflow aloft above convection, but a single radar cannot identify an updraft. Spectrum width, the spread of velocities within each radar sample, reached 5 to 8 m/s in a patch near and south of the later track; along the track itself it was similar to the surrounding echo. These measurements are independent of the crew's reports but do not establish the turbulence the aircraft met.
14. No other ADS-B traffic was recorded within 23 nm of the upset point at or above FL200 from 17:45Z to 17:49Z. At 17:48:16Z another business jet at FL350, 122 nm away over the Florida Keys, asked Miami Center to climb to FL370 "due to ice"; its callsign was partly heard and is consistent with SIY900, a Hawker 800. Pilot reports from FL300 and above in the region from 17:15Z to 18:25Z describe light chop, moderate turbulence or smooth air; none reports icing.

The President, restricted airspace and the airport

15. The presidential C-32A 09-0018 (hex ae4aec) took off from PBI at about 17:06Z. At 17:49:33Z it was at FL330 near Jacksonville, about 268 nm from the crash site; N722BD never came within about 170 nm of it. The White House schedule had the President departing Palm Beach for Dubuque, Iowa that day, time to be announced.
16. After the upset the track passed 0.8 nm from PBI at about 17:48:59Z (pressure altitude about 24,400 ft), 0.08 nm from Trump International Golf Club at about 17:49:04Z (about 22,100 ft) and 3.39 nm from Mar-a-Lago at 17:49:06Z (about 20,700 ft), then went south. The crash site is 4.86 nm south-west of Mar-a-Lago.
17. VIP temporary flight restriction 6/4136 (surface to 17,999 ft above mean sea level) was published as effective until 18:30Z but was not in the FAA feed after 17:20Z to 17:30Z. Using GPS height corrected to sea level, N722BD stayed above its ceiling until about 17:49:13Z.
18. At about 17:49:56Z the PBI tower controller called American 638 for a radio check and explained "we just lost power". It is noted because of its timing, about 23 seconds after impact 3.3 nm away; when the outage began and why are not known.

Imagery and datalink

19. Two dashcam videos posted by GlobalOSINTHQ show the final second. The aircraft appears as one compact image, sometimes cross-shaped like a fuselage with wings, with no obvious separate large component or sustained smoke trail. Blur, compression and occlusion mean the video cannot confirm that the airframe was intact. A widely shared still posted by Faytuks Network is the same frame as one of the videos. Details are in appendix B.
20. A search of the airframes.io ACARS archive found no messages from this aircraft, and no fixed traffic camera or archived webcam frame recorded the crash. The PBI digital ATIS broadcasts for about 17:00Z to 18:28Z were not received.

Requirement and scope

The requirement was to establish what happened to the aircraft seen diving into Greenacres in a video posted at 19:22Z, and whether it posed a threat to the President. The scope is the flight of N722BD on 1 November 2026, the weather along its final track, related air traffic control communications, the President's location from 17:00Z to 17:50Z and the restricted airspace around Palm Beach. This product does not assess crew, operator, maintenance, ownership or legal questions.

Assessment base

This product relies mainly on 2 kinds of source with different origins and failure modes:

• ADS-B position and velocity reports from the aircraft's transponder, aggregated by airplanes.live and ADS-B Exchange
• air traffic control radio, recorded by LiveATC.net volunteers

The two ADS-B aggregators carry the same transponder's messages, so they corroborate reception, not the aircraft's own sensors. The radio calls corroborate the ADS-B dive independently and give the routing, the crew's reports and the Mayday. Weather evidence comes from a pilot report, SIGMETs, NEXRAD Level II data from Miami (KAMX) and Melbourne (KMLB), a radar mosaic, GOES-19 imagery, airport observations and the 12Z Miami radiosonde. Dashcam video corroborates the steep dive and gives a partly independent, model-dependent speed estimate; the original video files have not been obtained. The President's location rests on a schedule seen only as a search snippet and on ADS-B of the presidential aircraft; neither confirms who was aboard. Mach and airspeed are HEXBIRD estimates that depend on thin wind data.

The most important gaps are the unarchived Miami Center frequency from 17:40Z to 17:47Z and the absence of recorder or wreckage evidence. Collection priorities lists the rest.

Map of the flight, final dive and weather

route from Austin (2-minute samples)final segment from 17:46Z, about 1 Hzlast positionVIP restriction rings, 10 and 30 nm (not in the FAA feed after 17:30Z)reference points
Low-level reflectivity, shaded at 20, 35, 45 and 55 dBZ: Iowa Environmental Mesonet mosaic of the lowest radar tilt, about 1 to 3 km above the ground here. Reflectivity at FL350, cyan outline: KAMX Level II data gridded at 11.25 km (GPS height at FL350). Cloud tops, purple outlines: GOES-19 band 13 brightness temperature, parallax-corrected for a 14.5 km top. Aircraft data: airplanes.live ADS-B trace of a9ab17. Select a point on the final segment to see its time, altitude and speed.

Radar cross-sections along the track

A radar chart of this storm has been widely shared on social media (the post by @Alkibiades_ at 21:50Z, which carries an @tafische watermark). HEXBIRD's reprocessing of the same KAMX volume reproduces the chart's main reflectivity pattern and approximate values. However, the chart compares pressure altitude with radar heights above sea level, and its upper-level measurements near the upset point were made after the upset: the 4.0° tilt, below the aircraft's height, 61 seconds after, and the 5.1° tilt 74 seconds after. It also labels the climb as an updraft. It therefore does not show that the aircraft entered 25 to 35 dBZ echo at its actual height, or that an updraft caused the climb.

Altitude, Mach and airspeed in the final 2 minutes

The altitude shown is GPS altitude from ADS-B. Mach and calibrated airspeed are HEXBIRD estimates, shown where the 5-second fit has enough points. ADS-B groundspeed and a 5-second fit of the GPS altitude rate give an inertial velocity, from which HEXBIRD subtracts winds reported by nearby aircraft. Temperature and pressure come from the 12Z Miami (KMFL) radiosonde. Above FL240 the winds come from only 2 or 3 aircraft per altitude band, and they differ from the radiosonde by about 20 kt at FL350. The estimates assume no vertical air motion. Across the horizontal-wind and temperature cases tested, Mach at the upset is 0.73 to 0.77; the spread widens to about ±0.06 early in the dive and narrows to about ±0.02 below 15,000 ft. The dashed green line marks the Learjet 60 maximum operating Mach number, 0.81, which applies from about 26,750 to 37,000 ft pressure altitude. The dashed orange line is the type's maximum operating speed at the aircraft's estimated pressure altitude, expressed in knots calibrated: 300 kt below 8,000 ft, 340 kt from 8,000 to 20,000 ft, tapering to 330 kt at 23,000 ft, and the Mach 0.81 limit above about 26,750 ft. It follows the airspeed-limit chart in the EASA and FAA type certificate data sheets, whose altitude breakpoints can be read to about ±500 ft. Pressure altitude comes from the radiosonde pressure at GPS altitude, about 1,900 ft below GPS altitude at FL350. The limits are in indicated airspeed and indicated Mach, which differ slightly from the calibrated values estimated here. The shaded band marks the last 8 seconds, when the aircraft flagged position integrity as unknown.

Dashcam video gives a partly independent estimate of about 590 kt at 66° below horizontal in the final second, about Mach 0.86, broadly consistent with ADS-B; an alternative calibration gives 640 to 660 kt. See appendix B.

3D view of the upset and dive

ADS-B track of N722BD from 17:47:30Z to the estimated impact at about 17:49:33Z, drawn to true scale with no vertical exaggeration. Blue: level flight and the climb before the upset. Orange: from the upset at 17:48:15Z to impact. The translucent curtain and grey ground shadow show the track's position over the ground; grid squares are 1 km. The moving marker replays the flight at 4 times real speed. Drag to orbit, scroll to zoom. Heights are GPS heights converted to height above ground; the final 8 seconds have degraded position integrity. By default the view shows radar reflectivity surfaces at 30 and 40 dBZ from the scan after the upset; the buttons switch to the earlier scan or turn the radar off. The scans are the same two Miami volumes as the radar cross-sections, with the same timing and resolution limits. The surfaces enclose only those two thresholds: blank space can still contain cloud, weaker precipitation and hazardous air motion, and reflectivity is not a measure of turbulence. Radar heights are above sea level and the track is above ground; the difference here is a few metres. Sources: airplanes.live ADS-B trace; NOAA NEXRAD Level II. If the 3D view does not load, the map and chart show the same data.

Analysis of competing explanations for the upset

The matrix tests what started the upset. It is an audit aid, not a vote. C means consistent, I inconsistent, N not diagnostic and U unknown. The hypotheses are not mutually exclusive: they can combine in sequence, for example turbulence followed by an autopilot disconnect and a stall warning, or structural failure after an overspeed. The matrix asks which starting events the evidence supports. Two explanations are handled outside the matrix. A mid-air collision or wake encounter is unsupported by the surveillance evidence (finding 14) but not excluded: reception is not guaranteed, earlier traffic could have left a wake, and military aircraft often do not broadcast ADS-B. An ADS-B data artifact is rejected, because dashcam video and radio calls corroborate the dive. Why the crew did not recover is treated below the table.

The crew's reports of ice, turbulence and precipitation in the minute before the upset are the main evidence for a weather-linked start. They fit H1 to H4 without separating them. The radio calls weigh against complete incapacitation (H7) but not against partial impairment, and weigh weakly against deliberate action (H8). Nothing directly supports or excludes a control, trim or autopilot fault (H5). The dashcam video shows no obvious separate large component in the final second, but it cannot confirm that the airframe was intact (H6).

A report of clear icing at −46 to −47 °C is hard to reconcile with ordinary icing, because supercooled liquid water is very rare below about −40 °C. One explanation is a high concentration of ice crystals. Ice crystals can affect airspeed sensors and engines without building ice on the airframe, and such regions often show only modest radar reflectivity. Other explanations are a misdescribed ice type or a temperature reporting error. Nothing measured the ice-water content or any effect on equipment.

The flight path angle held at −21° to −23° for about 14 seconds from 17:48:37Z, which could reflect a recovery attempt or a quasi-steady state. Several mechanisms could explain the failure to recover, and the evidence cannot separate them:

• high-Mach handling effects and very high control forces once above Mach 0.81
• spatial disorientation in cloud, which fits a tightening spiral
• a continuing fault such as a trim runaway
• structural damage at calibrated airspeeds far above the type's limits

Assumptions that carry the judgments

• The by-ear check of the transcript is accurate. One listener did it, and FAA audio has not been obtained. This carries the communicating and pilot-report judgments; FAA recordings that differ would show it wrong.
• Radar at FL350 represents conditions near the aircraft, though each beam averages about 2 km of depth and no scan coincides with the upset. This carries the weather judgment; finer-resolution radar showing different conditions would show it wrong.
• The 17:44:15Z velocity pattern is relevant to the upset. That depends on how it persisted or changed during the unsampled 4 minutes, which has not been established.
• ADS-B GPS altitude and velocity stayed accurate until 17:49:25Z, and winds reported by 2 or 3 aircraft per altitude band represent the winds along the dive. These carry the speed findings; recorder data would test them.
• The President's departure is inferred from the schedule and the C-32A's movement. This carries the judgment that he had left Palm Beach; a pool report placing him in Palm Beach after 17:06Z would show it wrong.

What would change these judgments

• The weather encounter likely contributed: evidence of an aircraft fault that began before the weather encounter would lower it; evidence of structural failure before the dive would make weather a less direct factor.
• The mechanism cannot yet be assessed: recorder data or audio describing an airspeed disagreement, a stall warning, an engine problem or a trim problem would point to H2, H3, H4 or H5 respectively.
• At least one person aboard was communicating: FAA recordings or a second listener that differ materially from the transcript.
• The icing report came from N722BD: evidence of another aircraft at FL350 near the reported position at 17:45Z.
• No evidence of deliberate targeting: evidence that the crew deliberately departed from a cleared or intended course toward Palm Beach for reasons other than weather or an emergency.
• The President had left Palm Beach: a pool report or other record placing him in Palm Beach after 17:06Z.

Collection priorities

1. Obtain the FAA's own audio and radar data, including 125.075 MHz from 17:40Z to 17:47Z, through the NTSB docket or a records request. This would fill the gap in the crew's communications and test the transcript.
2. Read the NTSB preliminary report, normally issued within about 30 days, for recorder data, wreckage distribution and whether the aircraft broke up in flight. This would test H1 to H6 and the failure to recover.
3. Obtain the filed flight plan or an FAA statement confirming the destination and the cleared route.
4. Check public Terminal Doppler Weather Radar Level II holdings for the radar serving Palm Beach (listed by the Iowa Environmental Mesonet as DJT, about 7 km from the upset point) for high-elevation scans of FL350 between 17:44Z and 17:49Z, and request the FAA's retained data if they are not public. At that distance FL350 lies about 57° above the horizon and the beam would be about 130 m wide, compared with about 2 km for the Miami radar. This would test the weather judgment.
5. Confirm the stall-protection equipment fitted to serial 60-271, and confirm the speed-limit altitude breakpoints against the approved flight manual (FM-123). This bears on H3 and the failure to recover.
6. Obtain the original dashcam video files, which would resolve the airframe better than the posted copies and test H6.
7. Find a White House pool report or Dubuque arrival coverage confirming that the President was aboard the C-32A.
8. Check Florida Power and Light outage records and the PBI tower facility log for the power loss reported at about 17:50Z, to establish whether it is connected to the crash.

Method note and AI disclosure

An AI agent, Claude Opus 5.5 running in Claude Code, did this analysis and drafted this report on 1 and 2 November 2026. It worked with delegated sub-agents of the same model, including a separate challenge pass and a structural review of version 17. OpenAI GPT-6.1-Sol, run through the Codex command-line tool, independently reviewed version 5, the dashcam speed analysis, the version 17 radar and destination changes and the version 20 speed limits. The work used:

• HEXBIRD ADS-B, FAA registry, NOTAM, NASR and OpenStreetMap queries
• public web sources and LiveATC archives
• NOAA radar and satellite data from public cloud archives
• local Python scripts

Speech-to-text used OpenAI Whisper models run locally without prompts. Mach and airspeed estimates use standard compressible-flow relations. Radar data were processed with Py-ART, including region-based velocity dealiasing; the 3D radar surfaces are marching-cubes envelopes of the gridded reflectivity. Processing limits:

• the map's FL350 layer treats masked radar gates as weak echo before gridding, while the cross-sections and 3D surfaces omit them; gridding smooths peaks
• radar spectrum width at 124 km reflects variability within large sampling volumes, and its quality has not been checked for second-trip echo, sidelobe contamination from heavy precipitation below, partial beam filling or wind shear within the beam
• the satellite values were not screened with quality flags
• the ACARS search ran about 270 queries

Scripts, raw data, query text and extracts, and the review records are kept with the investigation files. No human review other than the transcript check has taken place.

Corrections

• Version 17: versions 1 to 16 gave the destination as Freeport, Grand Bahama, inferred from the clearances "direct Freeport". Flightradar24's flight history lists North Eleuthera, and the clearances are consistent with Freeport as a routing point. This affected the controlling message, findings 2 and 3, the targeting judgment and the flight-plan collection priority.
• Version 10: version 9 said the near camera was driving south. It was a rear-facing camera on a car driving north. This affected finding 19 and appendix B.
• Version 9: versions 6 to 8 described the Faytuks Network still as a suspected altered derivative and excluded it. That came from comparing the still and the video frame at different scales; at matching scale they are the same frame. This affected finding 19 and appendix B.

Sources

1. LiveATC.net, archived air traffic control audio: streams ZMA08-FMY-Combined (1700Z), ZMA25-FMY-Combined (1700Z, 1730Z), KPBI1-ZMA01 (1730Z, 1800Z), ZMA26-KMIA3 (1730Z), KPBI2-Twr (1730Z), KPBI2-Guard (1730Z) and KBCT1-Guard (1730Z), 1 November 2026.
2. airplanes.live, ADS-B trace for a9ab17, accessed about 20:00Z. HEXBIRD default.recent carries the same data.
3. ADS-B Exchange, ADS-B trace for a9ab17, accessed about 20:00Z.
4. Aviation Weather Center, PIREPs within 200 nm of PBI, accessed about 19:58Z. The 17:45Z report was received at 17:55:06Z.
5. Aviation Weather Center, METARs for KLNA and KFLL, accessed about 19:58Z.
6. Iowa Environmental Mesonet, SIGMET archive, 15:00Z to 18:00Z, including convective SIGMETs 42E and 45E.
7. Iowa Environmental Mesonet, NEXRAD lowest-tilt reflectivity mosaic, 17:40Z to 17:50Z.
8. NOAA, NEXRAD Level II data on AWS, KAMX and KMLB volumes from 17:40Z to 17:54Z, including Doppler velocity and spectrum width.
9. NOAA, GOES-19 ABI data on AWS, CONUS bands 2 and 13 and cloud-top height, 17:41Z to 17:51Z.
10. Iowa Environmental Mesonet, KMFL radiosonde, 12Z.
11. airframes.io ACARS archive, searched through HEXBIRD on 1 November 2026, including datalink temperatures from other aircraft.
12. Flightradar24, flight history for N722BD, accessed about 03:00Z on 2 November 2026: 1 November flight from Austin (AUS) to North Eleuthera (ELH), actual departure 15:37Z.
13. FAA aircraft registry for N722BD, queried through HEXBIRD, snapshot current on 1 November 2026.
14. HEXBIRD default.tfrs and default.notams: VIP restriction 6/4136, Mar-a-Lago restriction 6/0100, and feed records at 17:20Z and 17:30Z.
15. White House Press Pool, schedule for 1 November 2026, seen as a search snippet only.
16. GlobalOSINTHQ, two dashcam videos of the final descent posted on X, 19:51Z. One is credited to @onlyinpalmbeachtv; the original files have not been obtained.
17. Faytuks Network, dashcam still of the aircraft before impact posted on X, 20:08Z. The same frame as one of the GlobalOSINTHQ videos, cropped and in better quality.
18. Google Street View, Military Trail at 3064 S Military Trail showing the Palm Coast Plaza sign, imagery January 2025.
19. Scope Report, bystander video of the dive posted on X, 19:22Z. It shows a phone re-filming a screen, and its original source is unknown.
20. EmergencyStream, aftermath video posted on X, 19:02Z.
21. Osint613, video of the crater on Military Trail posted on X, 19:15Z, credited to a Facebook user.
22. WPTV, report of the incident at Military Trail and 10th Avenue, 1 November 2026.
23. CBS12, report naming the aircraft and a Bahamas destination, 1 November 2026. Its source for the destination is not stated.
24. @Alkibiades_, radar chart of the storm posted on X, 21:50Z, carrying an @tafische watermark. Discussed under the radar cross-sections.
25. BocaNewsNow, timeline of the Palm Beach County crash, updated to 15:50 local time.
26. European Union Aviation Safety Agency, type certificate data sheet IM.A.212, Learjet M60, issue 04, 12 November 2024, section III.4, airspeed limits chart. It validates FAA type certificate A10CE.
27. Federal Aviation Administration, type certificate data sheet A10CE, revision 69, 5 July 2026, section VIII (Model 60), from a mirror copy; the FAA's own link returned an error. It carries the same airspeed limits chart and a maximum operating altitude of 51,000 ft pressure altitude.

Appendix A: air traffic control transcript

Audio from LiveATC.net, from the archived streams named in each row. Times are UTC from the archive file start; feed clocks can be off by a few seconds. Two speech-recognition models (Whisper large-v3 and large-v3-turbo) made the transcripts on a local computer, with no prompt, and HEXBIRD checked them against ADS-B. The requester then listened to every row against the audio: 22 rows were confirmed as transcribed and 3 were corrected (two wording corrections, at 17:36:20Z and 17:40:05Z, and one speaker attribution, at 17:39:56Z); 1 was marked unclear because its speaker could not be identified (17:15:40Z). In 3 rows the words are clear but a speaker could not be identified. The "Support" column records that check; it rates the transcription, not analytical confidence. Square brackets mark words that are unclear or added for sense. Two crew transmissions, at 17:48:54Z and from 17:49:17Z to 17:49:24Z, are not reproduced because their words are low confidence; they were not part of the by-ear check. Miami Center's 125.075 MHz frequency, which N722BD used from about 17:40Z to 17:47Z, is not archived on LiveATC.

Appendix B: dashcam evidence

Two dashcam videos posted by GlobalOSINTHQ show the final second. This report calls them the near camera and the far camera.

• The near camera was rear-facing on a car driving north on Military Trail. The car had just passed the impact point and was probably several tens of metres from it, plausibly about 40 m. It shows a smeared image with wing-like elements, and fire starts at ground contact.
• The far camera (credited to @onlyinpalmbeachtv) was on a car driving north about 800 m south of the impact. The aircraft appears for about 0.84 seconds as a single compact image, sometimes cross-shaped like a fuselage with wings. No obvious separate large component or sustained smoke or fire trail is discernible in the examined frames.

Blur, compression, limited sampling and occlusion mean the videos cannot confirm that both wings or other components remained attached, and cannot exclude damage or anything shed earlier. Aftermath video shows a crater on Military Trail.

Frame match and location

The still posted by Faytuks Network is the same frame as frame 16 of the near camera, about 0.1 seconds before impact: scaled to match, its poles, roadside tree, sign and aircraft streak line up. It is not an independent view, and which posted copy came from which is not known. Google Street View from the northbound lanes of Military Trail at about 26.6330N 80.1129W, looking south-south-west, reproduces the scene: the Palm Coast Plaza sign at 3064 S Military Trail is on the left, the median palms and the green 10th Avenue North sign are ahead, and the crossarm poles and strip mall are on the right. The still is consistent with a genuine frame; the original video file has not been obtained.

Impact speed from the far camera

The far camera gives a partly independent estimate of the speed in the final second. Under the assumptions below, it puts the aircraft at about 590 kt, diving at about 66° below horizontal. That is broadly consistent with ADS-B, but the uncertainty is wider than the statistical ranges alone suggest.

How the estimate was made:

• the aircraft was tracked in 22 positions over about 0.84 seconds; it disappears behind a tree before ground contact, so the contact time comes from the model
• the video file repeats a near-duplicate frame in a regular pattern consistent with conversion from 25 to 30 frames per second, so successive positions are about 40 ms apart
• the lens scale was calibrated from the scene, using a later U-turn by the car and the straightness of utility poles
• the camera position comes from its GPS overlay, speed readout and a mapped power pole; the range of about 800 m comes from the estimated impact point
• a constant-velocity straight path ending at the ground was fitted to the positions; motion along the line of sight is weakly observed, so the track direction (about 290°) was taken from the final ADS-B positions

Limits:

• the ranges in the table come from a Monte Carlo propagation of the inputs under the adopted calibration, geometry, timing and constant-velocity assumptions; they exclude unresolved calibration and model uncertainty
• an alternative lens calibration, which the analysis could not exclude, gives about 640 to 660 kt, about Mach 0.94 to 0.97
• the Mach numbers use the Lantana surface temperature of 27 °C at 17:55Z, which puts the speed of sound at about 675 kt, after removing the surface wind of 12 to 19 kt from the east; the wind changes Mach by less than 0.01
• the estimate is not fully independent of ADS-B: it uses ADS-B positions for the range and track, and the analyst had seen an earlier ADS-B-derived speed range, though no ADS-B speed or altitude values were inputs to the fit

Comparison with ADS-B: horizontal speed and track agree closely, at 228 to 245 kt on a track of about 277° to 290° over the last few seconds. Averaged over 2 to 4 seconds, ADS-B gives about 594 to 601 kt total, inside the video's conditional 68% range. Over the last 1.1 to 1.6 seconds it gives about 638 to 661 kt, at or beyond the upper end of the video's conditional 95% range. The two use different averaging periods and share some geometric inputs, so they do not support a narrow impact-speed interval. All estimates are subsonic, from about Mach 0.86 to 0.97, slightly below the dive's peak (finding 9) because the warmer air near the ground raises the speed of sound.