Aguadilla: one film, competing tracks

The target against land — Robert M. Powell, Morgan Beall, Larry Cates, Carl Paulson, Richard Hoffman and Daina Chaviano / SCU; original thermal imagery: U.S. CBP
Event image. A full thermal frame selected in SCU’s report, with the target left of the reticle and terrain visible behind it. The displayed screen counter is 5173; this is a still from the same recording, not an independent photograph. Extracted without retouching from the image embedded in SCU’s report.
CC BY 3.0 US
· Original image source · Image licence

A small thermal target appears to cross an airport, separate and disappear. Independent researchers reconstructed a remarkable flight; AARO later reconstructed drifting objects. The disagreement turns on geometry that you can inspect in both complete reports.

World: Earth
Location: Puerto Rico
Case date or period: 25–26 April 2013 · analyses 2015–2026

The recording that started the investigation

A Customs and Border Protection aircraft recorded a short infrared sequence near Rafael Hernández Airport. SCU dates the event to the evening of 25 April 2013; the frame display and AARO’s case heading use 26 April. Keep that date convention visible when comparing records. A dark target moves across a changing background as the aircraft circles. Near the end, two signatures become distinguishable and contrast fades. Those are image observations. Whether the target crossed water, moved rapidly or divided requires a model connecting the camera to the landscape.

Independent researchers assemble a case

Powell and colleagues did substantially more than enlarge a circulating clip. Their report describes obtaining the original recording from an anonymous official source, checking its metadata against FAA-origin radar data obtained through a Freedom of Information Act request, interviewing relevant witnesses and reconstructing the aircraft’s position. They divided the film into individual frames, compared identifiable ground features, considered the camera’s infrared response and estimated the target’s dimensions and motion. The radar work supported their assessment of the recording’s authenticity; it did not automatically establish every interpretation of the small thermal target.

The extraordinary interpretation — and a correction

SCU’s analysis inferred a fast-moving object travelling close to the surface, entering water and apparently separating into two parts. Its report examines alternatives including aircraft, birds and balloons, then argues that they do not explain its reconstruction. The conclusion depends on where the target lies along each camera sightline. Version 8 also preserves an important correction: an appendix about acceleration was removed because it had failed to account for the camera locking onto the background. That published erratum is part of the investigation, not an excuse to silently retain the superseded calculation.

AARO reconstructs something much slower

AARO’s March 2025 report uses Systems Toolkit modelling to place two objects over land, drifting at about eight miles per hour. It attributes apparent rapid motion to parallax from the moving aircraft and apparent water entry to loss of thermal contrast, increasing distance and possible cloud interference. It identifies earlier moments when two signatures can already be distinguished. AARO expresses high confidence that the behaviour was not anomalous, but only moderate confidence that the objects were sky lanterns. Those confidence statements concern different questions and should not be collapsed into certainty about the objects’ identity.

The debate becomes specific

In April 2025, SCU published the technical questions it had sent to AARO: the three-dimensional target path and uncertainties, the cloud information, access to tower or military radar records, and the pixel analysis of an apparent utility-pole crossing. SCU said AARO had acknowledged the inquiry but was not then ready to provide answers. This is a documented request for reproducible details, not a completed rebuttal demonstrating that AARO is wrong. Nor does the existence of an official resolution erase the earlier independent work or the assumptions a reader can examine.

A new team reopens the analysis

At the July 2026 SCU conference, Douglas Buettner presented a revisit with Tejin Cai and Joshua Pierson. The slides say SCU’s board requested researchers outside the original analysis team. Their tools read the on-screen telemetry, align infrared frames with mapped ground features, and estimate motion from changing viewing directions. Several statistical methods produce candidate paths rather than a directly measured target range. The presentation labels its results preliminary and its comparison of explanations unfinished; it does not announce a final identification. This brings an internal reassessment into a debate often presented as two fixed positions.

Testing the apparent water entry

The team also questions treating displayed grey levels as raw temperature measurements. Its slides identify conversion from 14-bit sensor data to 8-bit video and H.264 compression as issues the earlier report did not adequately address. Buettner reports a prototype simulation producing apparent water entry, while ocean calibration and further simulations remain incomplete. The authors call for a comparable camera experiment to test the modelling. These are methodological criticisms and preliminary simulation results, not another recording of the encounter. When replaying the thermal disappearance here, keep the observed loss of contrast separate from a proposed physical crossing.

Read the two models against the same frames

Start with SCU’s eight-frame sequence, then AARO’s earlier separation frames. Open the full reports for the assumptions behind the pictures. SCU supplies extensive calculations and source correspondence; AARO supplies a shorter alternative reconstruction and considers both birds and balloons as competing hypotheses. Neither a dark spot’s shape nor a confident narrative alone yields its distance. The most useful unresolved comparison is whether the different sightlines, occlusions and timing conventions can be reproduced in one common coordinate system with explicit uncertainty. The material here lets you follow that disagreement without choosing a conclusion in advance.