The Aguadilla Airport Incident, Puerto Rico (2013)

On the night of April 26, 2013, a U.S. Customs and Border Protection (CBP) DHC-8 patrol aircraft departing Rafael Hernández Airport (TJAB), Aguadilla, Puerto Rico captured a mid-wave infrared (MWIR) video of a small, light-emitting target transitioning across the airport environment and into coastal airspace. The video, recorded with a Wescam MX-15D EO/IR turret, was later leaked and became one of the most cited UAP clips of the 2010s. A comprehensive technical study by the Scientific Coalition for UAP Studies (SCU) reconstructed geometry and timing from the video frames and airport/radar contexts and argued for anomalous, low-altitude performance, including apparent water entry and object “bi-location” (split) near the end of the clip. Zenodo

In March 2025, the All-domain Anomaly Resolution Office (AARO) released an official case-resolution report concluding, with high confidence, that the footage shows two small objects drifting at wind speed/direction over land; with moderate confidence AARO attributed them to sky lanterns. AARO’s reconstruction using Systems Toolkit (STK) found no trans-medium entry and no anomalous kinematics; the “split” is, in their view, simply two nearby objects that are unresolved in portions of the clip. AARO

This article lays out the data chronologically and technically: sensor parameters, airport geometry, winds, radar references, and the two main explanatory models (SCU’s anomalous interpretation vs. AARO’s sky-lantern resolution), plus earlier skeptical takes (e.g., balloon/lantern hypotheses on Metabunk). We conclude with a transparent claims taxonomy, speculation labels, and practical implications for future UAP imagery analysis.

The record: What we know happened and when

• Platform & sensor. The recording platform was a CBP De Havilland Canada-8 (Dash-8) patrol aircraft. The sensor was a Wescam MX-15D operating in the 3–5 μm MWIR band. SCU’s publication includes a downloadable copy of the video and report; their description states the authenticity was checked against FAA radar data and the video’s embedded metadata. Zenodo
• Time & place. AARO dates the event to April 26, 2013 at ~9:22 pm local over/adjacent to Rafael Hernández Airport (Aguadilla) and the north-west Puerto Rico shoreline. AARO
• Clip length. ~3 minutes of MWIR full-motion video (FMV). SCU’s Zenodo entry links both the report and the source AVI. Zenodo
• Initial detect & track. The DHC-8 crew acquires a hot target near the airport, then orbits while the MX-15D alternates wide/narrow FOVs and polarity (with the usual caveat that in many MX-15D modes black = hot). The target traverses residential/airport environs. Near the end, the clip shows two distinct returns many viewers interpret as a single object splitting and entering seawater with little splash before vanishing. Zenodo
• Public release history. Anonymized sources provided the high-res clip to independent researchers (archived by The Black Vault with an accompanying letter). Over time the video circulated broadly, inspiring multiple technical analyses and media segments. The Black Vault

The airport footage: What it actually shows

Metadata and overlays

The MX-15D embeds time stamps, sensor azimuth/elevation, polarity, and GPS readouts. SCU notes the lat/long rounding and a one-second metadata lag relative to scene content, accounted for in their reconstruction. These details matter because they anchor line-of-sight (LOS), slant-range assumptions, and cross-checks against airport features. Academia

Observables seen in the video

• Thermal intensity: The target’s apparent temperature fluctuates within MWIR; at times it dims to background levels and reappears.
• Morphology: For much of the clip, the object appears as a compact, unresolved hotspot with intermittent shape changes under zoom or angle changes, typical for sub-meter objects near sensor resolution limits.
• Trajectory over terrain: The target’s path crosses road grids, trees, and structures near the airfield perimeter; later it seems to approach the shoreline.
• “Split” & “water entry” moments: In late segments, the target appears to bifurcate into two returns. Viewers also note moments where the return vanishes with ocean background interpreted by some as immersive entry (trans-medium). Zenodo

AARO’s reconstruction from the same pixels

AARO’s STK model fuses aircraft position, sensor look angles, and the changing geometry as the Dash-8 gains altitude and arcs around the field. Their key findings:

• The video always shows two objects (not one splitting), with separations visible at multiple earlier timestamps once you know where to look.
• The objects move in a straight line at ~8 mph wind speed that evening over land, not over water at the moments of “vanish.”
• The “vanish” is attributed to thermal crossover and reduced contrast (sensor distance & clouds).
• Pixel analysis estimates each object < 1 meter across.
• Attribution (moderate confidence): sky lanterns, consistent with local resort customs and the observed flickering thermal return. AARO

Watch the AARO-hosted version & read the resolution: “Puerto Rico Object” on AARO’s Official UAP Imagery page (includes links to the case resolution and STK reconstruction summary). AARO

Radar narratives and external corroboration

SCU’s report asserts that FAA radar data were obtained and correlated with metadata, supporting authenticity and the spatiotemporal context they modeled from the video frames. Their reconstruction relies on airport map alignment, tower/aircraft geometry, and winds to derive ground speeds they argue exceed those expected for balloons/lanterns, particularly when the object moves near tree-top altitude around built-up areas. Zenodo

AARO’s resolution, by contrast, emphasizes a high-fidelity flight-path reconstruction driven by sensor geometry and aircraft kinematics rather than external radar tracks, concluding the objects drift with wind and remain over land. While AARO cites local wind and hotel/resort customs (lantern releases), their report does not lean on FAA radar to quantify object motion; instead, it treats the full-motion video itself as sufficient (once properly modeled) to resolve speed/altitude. AARO

Competing hypotheses: data-driven comparison

SCU’s anomalous-performance model

• Speed & path. SCU’s frame-by-frame analysis estimates speeds approaching ~100 mph at times and asserts controlled flight near obstacles, a profile inconsistent with passive drift. The airport cut-across and near-shore maneuvering, combined with minimal hydrodynamic signature at alleged water entry/exit, are foregrounded as non-conventional. Zenodo
• Split & trans-medium. SCU interprets the late video bifurcation as a single object splitting into two equal parts, then re-entering the water at different times. They argue that two independent lanterns should have been visibly separate earlier and cannot account for the “entry” moments or thermal changes they chart. Zenodo
• Balloon/lantern rejection. In a 2024 update responding to Rubén Lianza’s lantern hypothesis, SCU argues required winds for a balloon/lantern to match lines-of-sight would be too high for sustained buoyancy; further, disappearances behind trees and purported water interactions are said to contradict lantern behavior. The SCU

AARO’s sky-lantern resolution

• Two objects throughout. AARO identifies multiple early moments where the objects are already separated. The later “split” appears more pronounced as the viewing angle steepens with aircraft climb.
• Wind-matched drift. STK modeling sets the objects’ speed near 3.6 m/s (~8 mph), consistent with surface winds (~9.8 mph) from E/NE; thus, no propulsion or anomalous acceleration is required.
• No trans-medium. The vanish events occur as thermal contrast collapses (crossover, sensor range, thin clouds) and the ocean background makes the objects indistinct. AARO concludes the objects never enter the water.
• Size & signature. Pixel analysis yields sub-meter scale; flicker is consistent with combustion in sky lanterns; local vendors reportedly confirm celebratory releases are common in the area. Confidence: high that behavior is non-anomalous, moderate that attribution is sky lanterns. AARO

Earlier skeptical takes and alternative conventional models

Prior to AARO’s report, Metabunk contributors proposed hot-air wedding/sky lanterns as a likely match, noting apparent wind-tracking and the plausibility of two tied lanterns appearing as one at some angles. Others mused about birds or balloons. These discussions anticipated parts of AARO’s 2025 reasoning, though without access to AARO’s STK tools or IC partner reconstructions. Metabunk

Trans-medium claim assessment (water entry/exit)

Claim: The object entered the Atlantic, exited, appeared to “split,” and then re-entered, all with negligible splash and no apparent hydrodynamic penalty. This is the central reason the case is frequently cited as trans-medium.

• SCU reading. SCU points to frame-timed intensity changes, contextual shoreline geometry, and two-object sequence timing, arguing that the simplest reading of the late clip is water interaction with energy management that preserves motion and thermal contrast, followed by a genuine bifurcation. Zenodo
• AARO reading. The STK reconstruction keeps both objects over land for the entire clip; perceived approach to the shoreline coincides with aircraft climb, cloud interference, and range increase. The “submerge” effect is explained by thermal crossover and contrast loss against ocean background, not by an actual interface crossing. AARO

Assessment: Because AARO supplies explicit geometry (aircraft track, sensor boresight projections) and quantified wind matching, its no-trans-medium conclusion is well-constrained. SCU’s counter-argument stresses different grounding assumptions (e.g., earlier radar corroboration and interpretations of LOS occlusion/terrain). As of 2025, the official position is no trans-medium; the anomalous interpretation remains disputed in the research community.

What the winds, speeds, and sizes imply

• Winds & speed. AARO’s 8 mph inferred speed (moving with wind) drastically undercuts claims of ~100 mph dashes. SCU’s estimated higher speeds rest on its own LOS geometry and timing; discrepancies turn on how the camera motion, zoom levels, and target parallax are handled. AARO
• Size. AARO’s pixel analysis suggests < 1 m objects, consistent with lanterns. SCU argues the dynamic behavior and environmental interactions are inconsistent with such small, passive, flame-borne devices. AARO
• Thermal behavior. AARO cites thermal crossover (around 2 hours post-sunset) and cloud interference as key to vanish/reappear moments; this fits known MWIR phenomenology (contrast collapses when target/background temperatures converge). AARO

Context from the public archive

• Primary technical study (SCU). Powell et al. (2015), updated on Zenodo, remains the major public-domain technical deep dive, complete with frame extraction, metadata notes, and a narrative emphasizing anomalous kinematics and trans-medium behavior. Zenodo
• SCU 2024 update. A response to Lianza/IPACO rebuts the lanterns hypothesis on wind, occlusion, and split logic. The SCU
• AARO 2025 resolution. A 7-page DoD case report uses STK to model look angles, flight path, and separation events, concluding two small objects drifting at wind speed, likely sky lanterns. The AARO UAP Imagery hub hosts the video and case resolution (plus a “Digital Systems Toolkit Reconstruction” note). AARO
• Independent archiving. The Black Vault maintains an early high-resolution copy and contemporary correspondence from an anonymous source; useful for case history but not a technical resolution. The Black Vault
• Community skeptics. Metabunk threads (2017→) laid out lantern/balloon ideas and parallax reasoning; these broadly anticipate elements in AARO’s 2025 analysis. Metabunk

Implications for UAP analysis

1. Geometry matters more than vibes. The same pixels can produce opposite narratives unless aircraft kinematics, sensor angles, and parallax are modeled. AARO’s STK formalism demonstrates how to bound speeds and rule out “split” illusions. AARO
2. Thermal phenomenology is tricky. MWIR systems are vulnerable to crossover windows, thin cloud, range, and gain settings, all of which can mimic “vanish” events or changes in object count. AARO
3. Attribution confidence ≠ behavior confidence. AARO is high-confidence on non-anomalous behavior, moderate on lanterns specifically, underscoring that behavioral bounds (speed/altitude) can be firm even when identity is less so. AARO
4. Public-domain science helps. SCU’s publication (and willingness to evaluate counter-claims) raised the bar for transparent, crowd-auditable UAP casework, even if the official answer later diverged. Zenodo

Disputes in focus

Is there any multi-sensor corroboration?

• SCU: indicates radar corroboration for authenticity and timing. Zenodo
• AARO: doesn’t need external radar to close the case; the FMV geometry suffices to bound speed and path and to refute trans-medium claims. AARO

One object that splits vs. two objects all along

• SCU: late “bi-location” is physical split; earlier frames do not obviously show two. Zenodo
• AARO: two are present throughout; separations at 00:29.56, 00:40.76, 00:47.00 become more evident with viewing angle changes and zoom. AARO

Trans-medium entry/exit

• SCU: interprets apparent water penetration with minimal hydrodynamic signature. Zenodo
• AARO: no water entry; the clip remains over land; “vanish” is thermal crossover + distance/clouds reducing contrast. AARO

Lanterns feasible given wind & local custom?

• SCU: contends required winds incompatible with lantern buoyancy, and occlusion/“entry” sequences are inconsistent with lantern dynamics. The SCU
• AARO: wind-matched drift, sub-meter size, flicker, and local resort releases support sky lantern attribution (moderate confidence). AARO

Practical takeaways

1. Always solve the camera before the craft. Movement relative to a moving, climbing sensor plus zoom changes can manufacture apparent accelerations and splits. Formal kinematic reconstructions are essential. AARO
2. Thermal crossover windows (≈2 hours after sunset) can induce vanish/reappear illusions; plan multi-band capture or time-offset revisits to break degeneracies. AARO
3. Attribution tiers help. Distinguish between behavioral bounds (speed/altitude/acceleration) and identity (what it is). Cases frequently permit high confidence on the former, lower on the latter. AARO
4. Public reproducibility. Provide raw frames, sensor parameters, and aircraft tracks so independent teams can replicate STK-like reconstructions. SCU’s open framework plus AARO’s modeling together point the way. Zenodo

Conclusion

The Aguadilla Airport Incident has been a litmus test for how our community handles ambiguous IR video. For years, the clip stood as a prime public candidate for trans-medium behavior and object “split.” The SCU report made the strongest open-source case for that reading, grounded in frame-level analysis, airport mapping, and a stated radar correlation for authenticity. In 2025, AARO entered with a complete camera/aircraft kinematic solution—not dismissing the footage as noise, but re-explaining its observables via two small objects drifting at wind speed, seen under changing look angles, zoom, range, and thermal crossover—and therefore non-anomalous.

The data-first position is clear:

• The footage is real and significant (government-origin, airport environment, good metadata).
• What it shows is disputed: SCU retains an anomalous interpretation; AARO offers a conventional reconstruction with high confidence on non-anomalous behavior and moderate on lantern attribution.
• Trans-medium claims, specifically water entry/exit, are not supported by AARO’s geometry and contrast model and remain unconfirmed.

The case’s value endures not because it “proves” a trans-medium craft, but because it demonstrates how rigorous modeling can turn an apparently extraordinary video into a bounded, testable scenario and how open-source and official analyses can both improve the evidentiary record. Future investigations should treat multi-sensor triangulation, synchronized visible/MWIR, and precisely logged platform state as non-optional if we want to close debates without residual ambiguity.

References

1. AARO – Official UAP Imagery hub: “Puerto Rico Object” (video, case page, and reconstruction links). AARO
2. AARO – Puerto Rico UAP Case Resolution (PDF, Mar 20, 2025): speeds, size, wind match, separation timings, thermal crossover, and attribution. AARO
3. SCU – 2013 Aguadilla Puerto Rico UAP (Powell et al., 2015/updated): report + source video; asserts radar corroboration; anomalous interpretation with trans-medium and split. Zenodo
4. SCU – 2024 update responding to Lianza/IPACO (lantern hypothesis): wind/LOS, occlusion, split critique. The SCU
5. The Black Vault – Archive & background letter: early high-resolution copy and source context for the leak. The Black Vault
6. Metabunk – Lantern/balloon discussion thread: early conventional explanations and parallax arguments. Metabunk

Claims Taxonomy

• Verified
  • CBP Dash-8 recorded MWIR footage over Aguadilla on April 26, 2013; video provenance is established in public archives (SCU) and the U.S. DoD AARO portal. Zenodo
  • AARO published a case-resolution report with STK reconstruction, concluding no anomalous kinematics/trans-medium and assessing two small objects. AARO
• Probable
  • Objects’ characteristic size is on the order of sub-meter; late-clip separation is best modeled as two objects, not a split. (AARO high confidence; aligns with pixel-based size bounds.) AARO
• Disputed
  • Trans-medium behavior: SCU sees water entry/exit; AARO attributes vanish to thermal crossover/contrast loss and holds the track over land. Zenodo
  • Speed/propulsion: SCU infers high-speed segments; AARO bounds motion at ~8 mph and wind-locked, implying no propulsion. Zenodo
• Legend
  • Narrative that the clip definitively shows a single craft splitting and diving into the ocean is not supported by the official reconstruction; it persists in popular retellings and social media. AARO
• Misidentification
  • Lanterns/balloons are plausible conventional stimuli; Metabunk and others proposed this years prior to AARO. AARO’s final attribution is sky lanterns (moderate confidence). Metabunk

Speculation labels

• Hypothesis.
  A heterogeneous scene is possible: two small, independent hot sources (e.g., lanterns) traversing a complex background, with sensor parallax, cloud veil, and crossover conspiring to produce bi-location and vanish illusions or, alternately, two compact engineered devices in loose formation, whose IR emissions and occlusion timing simulated drift. The first option is now the official solution space (AARO), but because attribution rests partly on contextual (lantern custom) rather than direct physical data, a small uncertainty remains. AARO
• Witness interpretation.
  Analysts (and crews) viewing rapidly zooming FMV from a moving platform often over-estimate target speed and under-estimate parallax and look-angle effects. The human visual system readily infers solidity and split/merge when two unresolved points alternately blend and separate. (This is a general caution, not a dismissal.) AARO
• Researcher opinion.
  The decisive factor in this case is not a never-before-seen “capability,” but instrumented reconstruction: line-of-sight, camera kinematics, wind match, and contrast physics. SCU’s legacy contribution remains important, open-publication forensics of government-origin video, but the 2025 official resolution shows how standard aerospace tools (e.g., STK) can reduce ambiguity without dismissing serious inquiry. Zenodo

SEO keywords

Aguadilla UAP 2013, Puerto Rico airport UAP video, CBP Dash-8 MX-15D infrared, trans-medium water entry claim, AARO Puerto Rico Object case resolution, SCU Aguadilla report, sky lantern hypothesis UAP, MWIR thermal crossover UAP, Wescam MX-15D UAP analysis, wind-matched drift parallax, UAP split illusion, FAA radar Aguadilla UAP