Two Robust Interstellar Meteor Candidates in the Post-2018 CNEOS Fireball Database

TL;DR

The paper develops a statistically rigorous approach to identify interstellar meteoroids among space-based fireball detections by applying a calibrated low-discrepancy uncertainty model to post-2018 CNEOS velocities and propagating them into heliocentric orbits with Monte Carlo realizations. It identifies two robust interstellar candidates, CNEOS-22 and CNEOS-25, whose heliocentric speeds exceed the Solar escape threshold by several standard deviations and yield zero bound realizations in $10^6$ trials, with $p_{bound} < 3\times10^{-6}$ and $z_\Delta$ of 8.7 and 5.5, respectively. The work compares these objects to known interstellar bodies, discusses robustness to systematics, and outlines practical material-recovery prospects and implications for the interstellar meteoroid flux. The findings provide the strongest calibrated evidence to date for interstellar meteoroids in the terrestrial environment and motivate rapid follow-up and recovery efforts where feasible.

Abstract

We report the identification of two previously unrecognized interstellar meteor candidates in the NASA CNEOS fireball database. Using the empirically calibrated low-discrepancy uncertainty model of Peña-Asensio et al.\ (2025) for post-2018 CNEOS velocity accuracy (1$σ$: speed 0.55~km~s$^{-1}$, right ascension 1.35$^\circ$, declination 0.84$^\circ$), we transform CNEOS velocity vectors to heliocentric orbits and assess interstellar candidacy via $10^{6}$-draw Monte Carlo simulations. Two post-2018 events have heliocentric speeds robustly exceeding the Solar System escape speed. CNEOS-22 (2022-07-28; 6.0$^\circ$S, 86.9$^\circ$W; eastern tropical Pacific) has $v_{\rm hel}=46.98$~km~s$^{-1}$, exceeding escape by $Δ= 5.18 \pm 0.60$~km~s$^{-1}$ ($z_Δ=8.7σ$), with interstellar speed $v_{\infty,\odot}=21.5$~km~s$^{-1}$. CNEOS-25 (2025-02-12; 73.4$^\circ$N, 49.3$^\circ$E; Barents Sea) has $v_{\rm hel}=45.63$~km~s$^{-1}$, exceeding escape by $Δ= 3.22 \pm 0.58$~km~s$^{-1}$ ($z_Δ=5.5σ$), with $v_{\infty,\odot}=16.9$~km~s$^{-1}$. For both events, none of $10^{6}$ realizations yield a gravitationally bound orbit ($p_{\rm bound} < 3\times 10^{-6}$). The adopted error model would need to underestimate the true uncertainties by factors of 5--9 for either candidate to be marginally bound.

Figures (3)

• Figure 1: Heliocentric speed $v_\odot$ versus geocentric speed $v_\mathrm{geo}$ for CNEOS fireballs with complete velocity vectors. The horizontal dashed line marks the bound/unbound boundary at the solar escape speed $v_{\mathrm{esc},\odot} \approx 42~\mathrm{km\,s^{-1}}$. Grey circles are low-discrepancy events with bound nominal orbits. Several events lie above the boundary at their nominal velocities (individual legend entries): pre-2018 events (open squares), including IM1 and IM2, and a post-2018 marginal event (open diamond). CNEOS-22 (2022-07-28) and CNEOS-25 (2025-02-12) (colored stars) are the only post-2018 events that remain robustly unbound across all $10^6$ Monte-Carlo realizations ($z_\Delta > 5$; $p_\mathrm{bound} < 3 \times 10^{-6}$).
• Figure 2: Monte-Carlo distributions of heliocentric speed $v_\odot$ for CNEOS-22 (2022-07-28; left) and CNEOS-25 (2025-02-12; right), based on $10^6$ realizations of the PenaAsensio2025 low-discrepancy uncertainty model. The dashed vertical line marks the solar escape speed $v_{\mathrm{esc},\odot}$ at each event's heliocentric distance. In both cases the entire distribution lies above Solar System escape, with no bound realizations observed. Inserts provide the mean heliocentric speed (in km s$^{-1}$), the mean margin above escape $\langle\Delta\rangle$ (in km s$^{-1}$), and its statistical significance in units of standard deviations, $z_\Delta$.
• Figure 3: Sensitivity of the interstellar classification to systematic underestimation of CNEOS velocity uncertainties. The margin significance $z_\Delta$ is plotted as a function of a multiplicative inflation factor applied uniformly to all three uncertainty components ($\sigma_v$, $\sigma_\mathrm{RA}$, $\sigma_\mathrm{Dec}$). CNEOS-22 (2022-07-28) remains above $3\sigma$ until the errors are inflated by $\sim$3$\times$, and above $1\sigma$ until $\sim$9$\times$. CNEOS-25 (2025-02-12) crosses $3\sigma$ near $\sim$2$\times$ and $1\sigma$ near $\sim$5.5$\times$.