TAMPON (LE) (974) 04.10.2025

On 4 October 2025 at 03:00, the main witness (T1) began climbing Piton de la Fournaise on Réunion from the volcano refuge with 3 companions, including T2, to see sunrise from the summit. About two-thirds of the way up, under a completely clear sky, T1 briefly moved away from the group and noticed a luminous object. It was visible for 2–3 minutes, moving horizontally from left to right before disappearing at the horizon. Initially white-yellow, it became a deeper yellow without flashing and left a visible trail. T1 judged it faster than an aeroplane but slower than a shooting star. Near the end, he distinguished 2 closely spaced light sources, suggesting a possible separation into 2 elements following the same path. T1 and T2 took 6 usable photographs clearly showing it. GEIPAN assesses the evidence as good. Only T1 reported to GEIPAN, but the photographs provided additional information and a very precise trajectory reconstruction. Its evidential-strength criterion concerns the quantity of reliable, objectively supported information collected for an account. T1 found the event only slightly unusual, thinking it probably a rational, identifiable phenomenon potentially recorded in an astronomical or space database because of its duration and visual extent. Unable to identify it himself, he contacted GEIPAN. The investigation likewise left little residual strangeness. A fireball, described in the source as a meteorite-type phenomenon, is highly improbable because of the duration and apparent speed. Conversely, 2 minutes or more, intermediate speed, a regular path, a white-yellow luminous trail, 2 close sources and a dawn sighting towards sunrise are typical of an artificial atmospheric re-entry visible from Réunion. CNES’s Space Surveillance Operations Centre, COSE, supplied orbital information pointing to the re-entering satellite Starlink® 2437 (Appendix 1). The trajectory reconstructed from those data does not exactly match the photographic reconstruction (Appendix 2). GEIPAN explains that at about 160 km altitude, immediately before final re-entry, strong atmospheric drag can bend a Starlink satellite’s path away from a “Kepler plus average braking” model. Small differences between simulated and actual atmospheric density can produce tens of kilometres of along-track error, corresponding to tens of seconds and a few degrees of apparent displacement from the ground. GEIPAN classifies the case B: probable atmospheric re-entry of a Starlink® satellite.