A machine learning approach to meteor classification
Samantha Hemmelgarn, Nicholas Moskovitz, Denis Vida · 2026
Read A machine learning approach to meteor classification by Samantha Hemmelgarn, Nicholas Moskovitz, Denis Vida (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-22986.txt.
Introduction
Hemmelgarn, Moskovitz, and Vida sort meteoroids by machine learning from 13 observed Global Meteor Network parameters, deepening the three-parameter Kb index. A semi-qualitative study of 28,177 meteor events from 2023 tests normalization, dimensionality-reduction, and clustering methods. Factor Analysis with a Gaussian Mixture Model yields clusters closest to old models, with three factors for kinematics, trigger thresholds, and size and shape effects. The activation factor best splits asteroidal and cometary origin, and the hardness scheme extends the older index from the densest iron ones to the softest cometary material.
Catalog record
| CREATOR | Samantha Hemmelgarn, Nicholas Moskovitz, Denis Vida |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 107303 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.22986
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A machine learning approach to meteor classification Samantha Hemmelgarn, Nicholas Moskovitz, Denis Vida Abstract We use machine learning to develop a framework for classifying meteoroids based on 13 directly observed parameters from the Global Meteor Network. This method adds depth to the K b K_{b} parameter, which uses only three parameters. We employ a semi-qualitative approach using 28,177 meteor events observed in 2023 by the Lowell Observatory Cameras for All-Sky Meteor Surveillance (LO-CAMS) network to evaluate multiple normalization, dimensionality-reduction, and clustering algorithms. We find that a combination of Factor Analysis (FA) and a Gaussian Mixture Model (GMM) results in clusters most consistent with traditional models. Three FA-derived factors corresponding to meteoroid kinematics, activation thresholds, and size/geometry effects describe the underlying structure of meteoroid behavior. The activation factor emerged as the most discriminating factor distinguishing whether a meteor is of asteroidal or cometary origin. Resulting 3, 6, and 11 cluster models reveal progressively finer compositional structure, from broad physical regimes to detailed subdivisions within cometary and asteroidal populations. From these results, we introduce a physically motivated hardness classification scheme: H class H_{\mathrm{class}} . H class H_{\mathrm{class}} is a data-driven extension of K b K_{b} which physically interprets clusters in terms of the densest iron meteoroids down to the softest cometary material. Application to nine well-studied meteor showers and analysis of clusters in orbital space aids in the physical interpretation of H class H_{\mathrm{class}} groups. The H class H_{\mathrm{class}} model is supported by an analytical FA–GMM formulation that enables application to future datasets. Our results demonstrate that machine learning methods can extract compositional information from modern optical meteor datasets at scale and offers a new framework for interpreting meteoroid populations. [orcid=0009-0006-1160-3829] [orcid=0000-0001-6765-6336] [orcid=0000-0003-4166-8704] keywords Asteroids, composition ,Comets, composition ,Meteors , † † titlenote: Corresponding Author † † credit: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Software; Validation; Visualization; Writing - original draft; and Writing - review & editing † † credit: Conceptualization; Data curation; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Validation; and Writing - review & editing † † credit: Conceptualization; Data curation; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Validation; and Writing - review & editing 1 Introduction Meteoroids are natural interplanetary particles which ablate when they enter a gaseous atmosphere, producing light and ionization among other things ( 9 ) . Meteoroids typically come from one of two types of bodies: comets or asteroids; thus giving them distinct properties that are indicative of their parent bodies. Cometary meteoroids are considered to be desiccated remnants of low-density cometary material, essentially thermally processed fractal dust aggregates ( 26 ) . Heat from the sun causes volatiles to sublimate from the surfaces of comets, ejecting dust from the nucleus, and creating a stream of particles ( 71 ; 23 ) . Understanding this process is crucial for determining the origins of dust in the solar system. For example, meteoroid streams from Long-Period Comets (LPCs) can be traced back to their sources in the outer Solar System to guide searches for their parent bodies ( 29 ) . On the other hand, asteroidal meteoroid streams are believed to be formed through asteroid collisions ( 34 ) or thermal cracking of the asteroid surface ( 35 ; 46 ) , although the exact mechanism of ejection is an active area of research ( 62 ; 36 ) . Observing and analyzing the light emitted from a meteoroid as it ablates in the Earth’s atmosphere informs its composition and evolutionary history. Spectroscopic analysis of meteors allows compositional classification, however it requires complex data collection, reduction, and modeling ( 6 ; 78 ) . Advanced meteoroid fragmentation models (e.g 7 ; 41 ; 10 ; 74 ; 79 ) can provide insights on the physical parameters of meteors that do not survive their journey through the atmosphere. These models generally work by fitting a complex numerical ablation model to meteor light curve and deceleration data. This approach is computationally intensive, requires accurate manual measurements, and has thus far only been applied to datasets of a few dozen meteors at a time. With the development and wide-spread availability of technologies such as CCD and CMOS sensors, automated camera networks began in the early 2000s to observe the sky on a nightly basis for meteors. Networks such as Cameras for All-Sky Meteor Surveillance ( 31 , CAMS;) , the Global Meteor Network ( 85 ; 76 ; 73 , GMN;) , the Desert Fireball Network ( 20 , DFN;) , the International Meteor Organization Network ( 55 ; 54 , IMO;) , and the Fireball Recovery and InterPlanetary Obsevation Network ( 17 , FRIPON;) have curated the orbits of \sim 3 million meteors ( 31 ; 85 ) since routine observations began in the early 2010’s. As automated meteor camera networks continue to grow, more scalable classification and characterization methods that rely only on what we directly observe are needed. Classifying meteoroids based on observed physical properties is crucial for constraining their evolutionary history and establishing links to parent bodies. Current classification methods typically rely on simplified physical models, such as single-body ablation theory, or utilize a limited set of observable characteristics. A standard metric for smaller meteors is the K_{b} parameter ( 14 ; 15 ) , which estimates material strength based on the meteoroid’s ability to penetrate the atmosphere. The parameter is defined as: K_{b}=\log\rho_{\mathrm{beg}}+\frac{5}{2}\log V_{\mathrm{init}}-\frac{1}{2}\log\cos\theta_{z} (1) Where \rho_{\mathrm{beg}} is the atmospheric density at the beginning of the trajectory (in \mathrm{g\ cm^{-3}} ), V_{\mathrm{init}} is the initial velocity (in \mathrm{cm\ s^{-1}} ), and \theta_{z} is the zenith angle. Empirically, K_{b} defines categories of meteoroid populations by strength: values \geq 8.0 indicate asteroidal material, 7.3\leq K_{b}<8.0 suggest carbonaceous origins, and K_{b}<7.3 point to cometary sources. Harder, and therefore denser, asteroidal material refers to ordinary chondrites with mean bulk densities of 3700 \mathrm{kg\ m^{-3}} , carbonaceous objects average 2000 \mathrm{kg\ m^{-3}} , and the softest cometary meteors are closer to 1000 \mathrm{kg\ m^{-3}} ( 15 ) . However, the K_{b} criterion is limited by its reliance on broad assumptions regarding meteoroid structure and a small number of entry observables. This restricts the physical fidelity of the method, particularly when analyzing the diverse datasets produced by modern automated networks. Alternative strength metrics, such as PE ( 13 ) and P_{f} ( 8 ) , utilize fireball end height or maximum dynamic pressure as proxies for strength, normalizing for velocity, mass, and entry angle. While effective for decimeter-sized fireballs, these methods are ill-suited for the millimeter-to-centimeter scale meteoroids that dominate the data stream from low-light camera networks. At this smaller scale, mass loss is driven by the continuous erosion of surface grains rather than the discrete fragmentation events typical of larger bodies ( 7 ) . Despite the structural assumptions inherent in K_{b} , it remains the preferred metric for smaller meteors as it relies on properties observable at the onset of the event. Machine learning (ML) algorithms present the opportunity to apply modern techniques to advanced meteor datasets, which contain a variety of directly observable features that may be used to indirectly constrain the physical properties of meteoroids. ML has successfully been applied in a number of recent meteor studies, including using ML to identify