Incident: 4 October 2025, approximately 04:47–04:49 local, Piton de la Fournaise, Réunion. Outcome: GEIPAN category B, probably an artificial satellite re-entry. The proposed identity was Starlink 2437, with an acknowledged trajectory discrepancy.
A group climbing to watch sunrise saw a luminous appearance cross the sky over several minutes. The agency’s summary describes a steady white-yellow trail and, towards the end, two closely separated lights. Two people supplied six usable photographs.

The walking time is not the sighting time
The group set out at about 03:00. The published questionnaire instead places the observation at about 04:47, with the first photograph at 04:48 and the last at about 04:49. Those times refer to different stages of the outing and should not be conflated.
The questionnaire also supplies an approximate mountain-trail coordinate. It is an observation-site reference, not a satellite location or a prediction of where debris fell. UKPX rounds it and does not treat it as current walking or access advice.
What the orbital enquiry found
The long duration and regular motion led GEIPAN away from a conventional brief meteor. The published analysis considered a satellite re-entry and obtained information from CNES’s operational space-surveillance centre, COSE. Starlink 2437 was identified as the candidate.
Crucially, the agency did not conceal a mismatch: the modelled path did not exactly coincide with the trajectory reconstructed from the witness photographs. Its comparison annex illustrates the two tracks. The conclusion remained category B—probable—rather than an unqualified category A identification.
What probable means here
A candidate orbit and an appearance consistent with re-entry can make an explanation persuasive without establishing every detail. Near final descent, atmospheric drag can complicate a simple orbital prediction. GEIPAN used that consideration in discussing the discrepancy; UKPX has not independently recomputed the orbit.
The photographs can document a luminous track while leaving uncertainty about the precise satellite identity. Conversely, a mismatch in a prediction does not automatically imply an extraordinary craft. The sensible record preserves both the evidence for re-entry and the limit on the specific match.
How the case is indexed
The multiple-lights tag reflects the reported close pair towards the end. It does not imply two independently piloted craft. The incident is indexed under Réunion, with its actual Indian Ocean timezone, rather than being placed in metropolitan France because a French agency investigated it.
The mountain photograph is contextual and is credited separately. Original sky photographs and technical annexes remain linked to GEIPAN; they are not replaced by a reconstruction or presented as UKPX’s own measurements.
Sources and record status
Primary case source · Original witness questionnaire · CNES COSE orbital annex · GEIPAN trajectory-comparison annex. Sources checked 10 November 2026. Reported observations, source assessments and UKPX’s discussion of evidence limits are distinguished above. No independent identification is claimed beyond the stated source finding.
Map reference: Rounded public mountain-trail observation reference from the witness’s published approximate coordinate; not a debris fall, satellite position or access instruction. Coordinate source. Incident year: 2025.

