UAP
case files.

Puerto Rico / Aguadilla — no split, no water entry

The recording. On 26 April 2013, an infrared sensor aboard a U.S. Customs and Border Protection aircraft recorded two indistinct objects near Rafael Hernández Airport in Puerto Rico. The footage became famous because it appeared to show one fast object dividing into two, passing over the airport and then entering and leaving the ocean.

AARO’s 2025 assessment. A reconstruction combining the aircraft’s position, sensor look angles and local wind placed the objects at about 656 feet, drifting in a straight line at approximately 8 mph. The apparent high speed resulted from motion parallax as the aircraft curved around the airport. AARO assessed with high confidence that two objects were present throughout the observation and that they remained over land rather than entering the water.

Why the objects seemed to disappear. The distance between aircraft and objects nearly tripled, cloud cover partly obscured the view, and the observation occurred within a period when thermal crossover could reduce contrast. When an object and its background return similar infrared values, the target can fade or vanish from the display even though it remains present.

Attribution and confidence. AARO assessed with moderate confidence that the objects were a pair of sky lanterns. Their sub-metre size, wind-matched drift and changing thermal signatures were consistent with lanterns, and investigators confirmed that releases were common at local hospitality events. Poor image quality prevented absolute identification, but AARO assessed with high confidence that no anomalous speed or transmedium behaviour occurred.

Read the AARO resolution ↗

Mt. Etna — a balloon 170 kilometres from the plume

The apparent anomaly. In December 2018, a shortwave infrared camera aboard a U.S. military uncrewed aircraft recorded an eruption of Mt. Etna from over the Mediterranean. For roughly four and a half minutes, a round object appeared to move rapidly through the volcano’s superheated ash plume without any visible effect. The operator estimated approximately 345 mph and an altitude of 500 feet.

The reconstructed geometry. AARO combined full-motion video analysis, pixel examination, wind calculations, photogrammetry and three-dimensional modelling. It assessed the object at about 15,000 feet, travelling approximately 24 mph with the wind. Rather than crossing the plume, it was about 170 kilometres from the volcano and approximately 30 kilometres from the sensor. Motion parallax created the impression of high speed.

Sensor environment. Volcanic ash and turbulent air scatter and absorb infrared radiation, producing a noisy scene. The sensor was configured mainly for air-to-ground observation and did not provide active range finding to the airborne target. Contrast stretching and post-processing could also create flickering and luminosity changes that looked like physical responses.

Conclusion. AARO assessed with high confidence that the object showed no anomalous behaviour and with moderate confidence that it was an approximately one-foot spherical balloon. The case is valuable because an apparently spectacular material claim — unaffected passage through volcanic ash — disappeared once the foreground and background were correctly separated.

Read the AARO resolution ↗

Eglin — a probable lighter-than-air object

The report. On 26 January 2023, a military pilot reported four radar objects between 16,000 and 18,000 feet in the Eglin Air Force Base training range. One was seen visually and captured in two electro-optical/infrared still images. The pilot described a roughly twelve-foot grey object with a rounded, cone-like form and orange-red colouring, moving slowly or remaining nearly stationary.

Complicating details. Video equipment was inoperable, and no imagery existed for the other three radar contacts. The aircraft radar failed after the pilot approached the visible object, but maintenance records showed that the same circuit breaker had tripped three times in preceding months. AARO therefore assessed the malfunction as probably coincidental rather than caused by the object.

Resolution. AARO concluded with moderate confidence that the observed object was very likely lighter than air — potentially a large commercial lighting balloon, meteorological balloon or Mylar balloon. Its slow, wind-consistent movement and two-tone infrared appearance matched known balloon characteristics. AARO tested a commercial lighting balloon that reproduced aspects of the reported shape and infrared contrast.

What remains limited. The exact balloon model and the other three radar returns were not identified. Nevertheless, no anomalous flight characteristics were confirmed, and independent intelligence and science partners also assessed the photographed object as ordinary and very likely some form of balloon.

Read the AARO resolution ↗

Unknown distance creates false performance

A camera records angles, not an object’s full three-dimensional position. To convert movement across an image into physical speed, an analyst needs reliable distance, sensor orientation, platform location and the time between frames. When even one variable is missing, dramatically different trajectories may fit the same picture.

GoFast appeared close to the sea but was assessed at approximately 13,000 feet. The Mt. Etna object seemed to cross a volcanic plume but was reconstructed about 170 kilometres from it. The Puerto Rico objects seemed to race across terrain and enter the ocean, yet reconstruction placed them drifting over land at wind speed. In all three cases, the visual impression was sincere and compelling; the inferred geometry was wrong.

This does not mean every fast-looking UAP is slow. It means apparent speed is a hypothesis until range is constrained. A video that lacks dependable distance can document an observation while remaining unable to prove the capability most often attributed to it.

Identification and performance are different questions

Public discussion often treats a case as either completely solved or completely mysterious. Official analysis is more granular. GoFast is the clearest example: AARO could not categorically identify the object, but it could assess with high confidence that the available video did not demonstrate anomalous speed. The identity remained open while the performance claim was resolved.

The reverse distinction also matters. Investigators may assess an object as probably a balloon without naming its manufacturer or recovering it. Moderate-confidence attribution is not absolute proof, but it can still be the best explanation supported by shape, wind, altitude and sensor behaviour. Eglin and Mt. Etna illustrate this level of conclusion.

A useful case file should therefore answer several questions separately: Was a physical object present? Can its motion be reconstructed? Did it display exceptional capability? Can it be attributed to a known category? How confident is each answer? Compressing these into a single word such as “solved” discards information.

Military-grade does not mean interpretation-free

Military sensors are sophisticated instruments designed for operational tasks, not infallible cameras. Their displays may stabilize a moving view, stretch contrast, change magnification, track a selected area and compress data for recording. These functions help an operator, but they can make the displayed outline differ from the physical outline of a distant target.

Gimbal raises the question of whether visible rotation belongs to the object, its glare or the optical system. Atmospheric Wakes shows how image persistence or sensor artefacts can resemble a physical exhaust trail. Al Taqaddum shows how automatic grayscale adjustment can make a balloon cluster fluctuate in appearance. Mt. Etna demonstrates how atmospheric turbulence and an air-to-ground configuration can degrade air-to-air interpretation.

Calling such effects “sensor artefacts” is not a dismissal. It is a testable explanation that must fit the instrument, scene and timing. Equally, invoking a sensor effect without demonstrating the mechanism is not a resolution. The quality of the conclusion depends on calibration information, complete footage and independent corroboration.

Unresolved is not evidence of a preferred theory

Some official files remain unresolved even when AARO assesses their behaviour as unremarkable. The reason may be simple: a short infrared clip can show that something physical was probably present while containing too little detail to distinguish a bird, balloon, drone or distant aircraft. Non-identification is then a property of the data, not necessarily of the object.

This principle cuts in both directions. An unresolved label does not justify declaring an extraterrestrial vehicle, secret weapon or sensor error. It also does not justify pretending that a final identification has been made. The proper conclusion is exactly as narrow as the evidence permits, together with a clear account of what additional information would be needed.

FLIR1 and Gimbal deserve continued attention because they are authenticated and publicly unresolved. They do not deserve certainty that the released material cannot support. Preserving uncertainty is not indecision; it is the discipline that prevents an evidence archive from becoming either advocacy or reflexive dismissal.