Apparent Rotation and Halo Spectral Structure of the "Gimbal" UAP: A Combined Kinematic and Spectral Analysis

The U.S. Navy AN/ASQ-228 ATFLIR recording designated “Gimbal” (DoD release April 2020; original recording January 2015) shows an unidentified infrared-significant object whose visible major axis rotates through three distinct events, at approximately frames 720, 820, and 980, identified by Peings & von Rennenkampff (2023). The interpretation of these rotations is contested: an alternative hypothesis attributes them to roll of the ATFLIR pod about its line-of-sight axis, making the apparent object rotation an artefact of a rotating camera reference frame. Peings & von Rennenkampff argue against this account from the pod’s documented optical design. We provide a complementary direct empirical test. The pod-rotation hypothesis, in its simplest form, predicts that the imaged background must rotate together with the object. Across an extended manoeuvre window (frames 700–1032; ~11.1 s), per-frame principal component analysis of the object’s hot core establishes a continuous apparent-rotation track of −116° ± 1°, cross-validated by two threshold-free methods (confirming the rotation is a property of the imaged object, not the segmentation), while three measurements of background rotation — two using Lucas–Kanade optical flow and one using an independent gradient-orientation technique — all exclude a rigid −116° rotation of the displayed frame. This dissociation rules out a rigid rotation of the entire displayed frame, but it does not, on its own, establish that the rotation is intrinsic to the object: because an ATFLIR-class targeting pod de-rotates its displayed imagery to hold the scene level, a stable background accompanied by a rotating object is the predicted signature of both an object-frame rotation and de-rotation of an optics-fixed feature (such as glare). The background-referenced test is therefore non-discriminating for the very hypothesis it is most often invoked against. We characterise the kinematic profile of the apparent rotation across the three events, and identify the quantities that do discriminate the two interpretations — the rotation’s magnitude and its timing relative to the line-of-sight boresight crossing — which we find consistent with the de-rotation account. We then characterise the halo itself. Per-frame spectral analysis of the parked tracking window recovers a periodic brightness modulation near 1.3 Hz localised to the halo and anti-phase between an upper-right and a lower region in the black-hot window; tested by multitaper magnitude-squared coherence against an analytic threshold, phase-randomised surrogates, an off-object control, and an out-of-sample split-sample design, the modulation reaches significance in the black-hot parked window (0.46 at 1.30 Hz, p = 0.025); the inverted-polarity white-hot segment is non-significant at full window (0.20) and its per-pixel structure is dominated by saturation-induced row banding, providing no independent confirmation, so the modulation is marginal, sensitive to region localisation, and established in black-hot only. The two results combine: if the rotating structure is an optics-fixed glare, the halo is part of that same glare and its modulation is most economically a modulation of the glare rather than a thermal structure on a body. We read the kinematic and spectral findings as two views of one phenomenon — a structured, weakly modulated, optics-fixed glare carried by the pod’s de-rotation — while noting that the data are consistent with this reading rather than establishing it. We make no claim as to the object’s nature.