SB10
The rapid advancement of observational and modeling techniques has elevated meteor science to one of the primary avenues for investigating the nature and origin of interplanetary matter and its parent bodies. This session aims to serve as a platform for presenting fundamental results and innovative concepts in this field, while also informing the broader planetary science community about the interdisciplinary impact of ongoing and future research efforts.
Session assets
Meteoroid impacts pose a critical threat to spacecraft. Natural objects as small as sub-millimeter ( >0.2 mm) upon impact, can deliver enough kinetic energy to damage or disable satellites[1]. Impact damage is governed by the meteoroid’s velocity, mass and bulk density as given by the ballistic limit equations [2][3]. Slow sporadic meteoroids (velocity < 20 km/s) dominate the meteoroid flux on Earth, but are very difficult to observe by both radar and optical methods, as they produce little ionization and light. Recent work shows that approximately 16% of these slow meteoroids are iron-rich [4], making them especially important to study as their higher bulk density will result in higher impact hazard for satellites in orbit.
In this work, we observe slow and small sporadic meteoroids using high-sensitivity Electron Multiplying CCD (EMCCD) video cameras. These EMCCDs, have a limiting meteor sensitivity of magnitude +8, 50 m/pixel spatial resolution at 100 km, and operate at 32 frames per second (FPS). To complement these measurements we fuse EMCCD data with high resolution imagery from the Canadian Automated Meteor Observatory’s (CAMO) mirror tracking system. CAMO achieves a spatial resolution of 6 m/pixel at 100 km and operates at 100 FPS with limiting detection sensitivity of +7. This provides high-cadence and precision observations of fragmentation and morphology. When merged with higher sensitivity EMCCD data (which captures the onset of ablation earlier) these measurements provide are critical constraints for modelling meteoroid structure.
Our study is based on the analysis of 100 slow sporadic meteors for which we have simultaneous CAMO and EMCCD data. As shown in Fig. 1 our data clearly shows two separate populations of meteoroids : (a) porous cometary particles that fragment and decelerate rapidly at high altitudes, and (b) dense, iron-rich or stony asteroidal meteoroids with minimal deceleration. These findings support past results [5].
Fig. 1. 100 slow meteoroids jointly recorded by EMCCD and CAMO video systems and their total trail length as a function of the energy required to be intercepted through atmospheric molecular collisions to begin erosive fragmentation and F-parameter. The F-parameter is a normalized measure of the location of the peak brightness along the trail with 0 indicating peak at the start and 1 peak brightness at the end of the trail. These two populations are denoted with (a) and (b). Following the work in [4] iron meteoroid candidates are those that have F < 0.31, Trail Length < 11 km and erosion Energy per unit cross section > 4 MJ/m.
To robustly infer the physical properties of these meteoroids based on our observations, we develop a novel method using Dynamic Nested Sampling [6]. This Bayesian inference technique, implemented via the dynesty Python package [7], is specifically designed to handle high-dimensional, degenerate, and multimodal parameter spaces. We use the Borovička et al. (2007) [8] meteoroid ablation and fragmentation to provide model fits to measured brightness and deceleration of meteors. This model assumes meteoroids fragment by continuous ejection of micrometer-sized grains. In combination with Dynamic Nested Sampling, we are able to define statistically significant solutions with credible intervals (CIs) for all the meteoroid physical characteristics. We define a custom log-likelihood function that jointly incorporates measurements of both luminosity and meteoroid dynamics. Unlike traditional forward-modeling approaches [8][9], which are challenged to produce uncertainty estimation and solution degeneracy, our method allows rigorous quantification of posterior distributions, capturing model degeneracies, and assessment of the uniqueness of retrieved solutions.
Our work provides the first probabilistic framework to extract meteoroid mass, bulk density, and fragmentation properties from atmospheric observations of meteoroids with quantified uncertainties. These results offer valuable inputs for space environment models like NASA’s MEM [10] and ESA’s IMEM [11] helping safeguard satellites from an often overlooked impact threat.
References:
[1] Moorhead, A. V. et al. (2019). Planetary and Space Science, 165, 208–218.
[2] Christiansen, E. L. (2001). NASA TP-2001-210788.
[3] Moorhead, A. V. et al. (2020). NASA/TM–20205011017.
[4] Mills, T. M. et al. (2021). Monthly Notices of the Royal Astronomical Society, 506(4), 6012–6024.
[5] Vida, D., Brown, P. G., & Campbell-Brown, M. (2018). Monthly Notices of the Royal Astronomical Society, 479(4), 4307–4319.
[6] Higson, E., Handley, W., Hobson, M., & Lasenby, A. (2019). Statistics and Computing, 29, 891–913.
[7] Speagle, J. S. (2020). Monthly Notices of the Royal Astronomical Society, 493(3), 3132–3158.
[8] Borovička, J. et al. (2007). Astronomy & Astrophysics, 473(2), 661–672.
[9] Buccongello, N., Brown, P. G., Vida, D., & Pinhas, A. (2024). Icarus, 410, 115907.
[10] Moorhead, A. V. (2020) NASA/TM-2020-220555.
[11] Soja, R. H., et al. (2019) Astronomy & Astrophysics, 628 (2019): A109.
How to cite: Vovk, M., Brown, P., and Vida, D.: Characterizing the Population of Small & Slow Meteoroids: New Physical Characterization Method for Satellite Risk Assessment, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-66, https://doi.org/10.5194/epsc-dps2025-66, 2025.
Climate change, driven by increased greenhouse gas concentrations, not only warming the troposphere (∼0–10 km) but also cooling and contracting the atmosphere above, including the stratosphere (∼10–50 km), mesosphere (∼50–85 km), and thermosphere (∼85–500+ km) [1–3]. This contraction is measurable and has been confirmed by multiple independent datasets and models over the past two decades. In particular, meteor radars, which routinely detect the ablation of small impact micrometeoroids at altitudes between 80 and 100 km, have shown that the peak ablation altitude is decreasing at rates ranging from ~200 to 800 m per decade [4–8]. This observed lowering is consistent with expectations from cooling-induced density changes at fixed altitudes. Although the implications of these changes for satellite drag and orbital debris lifetimes are starting to be explored in detail [9,10], little attention has been paid to their possible influence on larger meteoroids that penetrate deeper into the atmosphere and survive as meteorites.
In this study, we investigate whether ongoing climate-driven changes in atmospheric density can significantly affect the atmospheric trajectory and survivability of meteorite dropping fireballs, focusing on the century-scale timescale. To do this, we simulate the atmospheric entry of the Winchcombe meteorite fall, one of the best documented carbonaceous chondrite falls to date, under both atmospheric conditions of 2021 and those projected for the year 2100, using the output of the climate model from the WACCM-X (Whole Atmosphere Community Climate Model - Extended) [11]. The model assumes a moderate emissions scenario (SSP2-4.5) [12], and the density trends are extracted from simulations accounting for solar and geomagnetic activity variations [13].
Winchcombe represents an ideal test case. It was a slow low-altitude fireball (entry velocity: 13.9 km/s) with minimal atmospheric deceleration below 40 km, and produced a carbonaceous CM2 chondrite [14]. Its low strength (onset of fragmentation at ~0.07 MPa) and unusually low peak dynamic pressure (~0.6 MPa) make it highly sensitive to changes in atmospheric density. We modeled its entry using a semi-empirical fragmentation and erosion model [15–17], informed by manual identification of fragmentation points and limited by deceleration and photometry data. The simulations were repeated under a projected 2100 atmospheric density profile, obtained by applying regression-derived trends from WACCM-X output to the location and season of the Winchcombe fall. See Figure 1.
The comparison between the 2021 and 2100 simulations shows only modest differences in trajectory, light curve, and survivability. The luminous trajectory begins 3 km lower in the 2100 case for a typical +3 magnitude detection threshold. The first fragmentation occurs 820 m lower and the catastrophic fragmentation that produces most of the surviving fragments occurs 300 m lower. However, the final luminous point is actually 190 m higher in 2100 because of slightly faster deceleration in the denser lower stratosphere. The peak brightness remains virtually unchanged, although the fireball is ~0.5 mag fainter at altitudes above 120 km, due to lower densities and reduced drag in the mesosphere. The final surviving mass is reduced by just 0.13 g, or 0.037%, from an initial ~13 kg meteoroid. These variations are small compared to daily and seasonal variations in density [18], and are far below the uncertainties in most meteorite recovery campaigns.
Figure 1. Dynamic pressure vs. altitude for the Winchcombe fireball (blue) and its 2100 climate change simulation (red), with eleven fragmentation points marked (crosses). The right panel shows their pressure differences (black).
Acknowledgements
EP-A acknowledges financial support from the LUMIO project funded by the Agenzia Spaziale Italiana (2024-6-HH.0). DV was supported in part by the NASA Meteoroid Environment Office under cooperative agreement 80NSSC24M0060. IC was supported by a Natural Environment Research Council (NERC) Independent Research Fellowship (NE/R015651/1). EF acknowledges the funding received by the Grant DeepCFD (Project No. PID2022-137899OB-I00) funded by MICIU/AEI/10.13039/501100011033 and by ERDF, EU.
References
[1] Roble, R. G., & Dickinson, R. E. (1989). Geophysical Research Letters, 16(12), 1441–1444.
[2] Cnossen, I., Emmert, J. T., Garcia, R. R., Elias, A. G., Mlynczak, M. G., & Zhang, S.-R. (2024). Advances in Space Research, 74(11), 5991–6011.
[3] Emmert, J. T. (2015). Journal of Geophysical Research: Space Physics, 120(4), 2940–2950.
[4] Clemesha, B., & Batista, P. (2006). Journal of Atmospheric and Solar-Terrestrial Physics, 68(17), 1934–1939.
[5] Jacobi, C. (2014). Advances in Radio Science, 12, 161–165.
[6] Lima, L. M., Araújo, L. R., Alves, E. O., Batista, P. P., & Clemesha, B. R. (2015). Journal of Atmospheric and Solar-Terrestrial Physics, 133, 139–144.
[7] Dawkins, E. C. M., Stober, G., Janches, D., et al. (2023). Geophysical Research Letters, 50(2).
[8] Venkat Ratnam, M., Teja, A., Pramitha, M., et al. (2024). Advances in Space Research.
[9] Brown, M. K., Lewis, H. G., Kavanagh, A. J., & Cnossen, I. (2021). Journal of Geophysical Research: Atmospheres, 126(8).
[10] Brown, M., Lewis, H., Kavanagh, A., Cnossen, I., & Elvidge, S. (2024). Journal of Geophysical Research: Space Physics.
[11] Cnossen, I. (2022). Geophysical Research Letters, 49(19).
[12] O’Neill, B. C., Tebaldi, C., van Vuuren, D. P., et al. (2016). Geoscientific Model Development, 9(9), 3461–3482.
[13] Matthes, K., Funke, B., Andersson, M. E., et al. (2017). Geoscientific Model Development, 10(6), 2247–2302.
[14] McMullan, S., Vida, D., Devillepoix, H. A. R., et al. (2024). Meteoritics & Planetary Science, 59(5), 927–947.
[15] Borovička, J., Tóth, J., Igaz, A., et al. (2013). Meteoritics & Planetary Science, 48(10), 1757–1779.
[16] Borovička, J., Spurný, P., & Shrbený, L. (2020). The Astronomical Journal, 160(1), 42.
[17] Vida, D., Brown, P. G., Devillepoix, H. A. R., et al. (2023). Nature Astronomy, 7, 318–329.
[18] Vida, D., Brown, P. G., Campbell-Brown, M., et al. (2021). Icarus, 354, 114097.
How to cite: Peña-Asensio, E., Vida, D., Cnossen, I., and Ferrer, E.: Effect of climate change on meteorite dropping fireballs, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-594, https://doi.org/10.5194/epsc-dps2025-594, 2025.
Introduction
Classifying meteoroids based on their physical properties is important for understanding their origins, how they evolve, and what parent bodies they may be linked to. Current classification methods typically rely on simplified physical models like single-body ablation theory or use only a small number of observable characteristics. A commonly used approach is the Kb parameter, which estimates a meteoroid’s material strength based on how it penetrates the atmosphere. But using Kb requires several derived quantities and depends on assumptions about the meteoroid’s structure and behavior. This limits how broadly and objectively the method can be applied, especially when dealing with the large datasets produced by modern automated meteor camera networks.
More advanced fragmentation models can provide better insights, but they’re computationally intensive and have only been applied to relatively small datasets, typically just a few dozen meteors at a time. As these networks continue to grow, we need more scalable and objective methods that rely only on what we can directly observe.
Here, we’re developing a machine learning approach to classify meteoroids based purely on observed characteristics. We use 13 features that can be consistently measured from low-light video cameras, such as energy received before ablation, atmospheric density, and trail length. These inputs are analyzed using dimensionality reduction and clustering to find natural groupings in the data. The goal is to create a reliable, scalable way to classify meteoroids that can keep up with the size and complexity of current and future meteor datasets.
Methods
This project uses data collected in 2023 from two low-light meteor camera networks: the Croatian Meteor Network (CMN) and the Lowell Observatory Cameras for All-sky Meteor Surveillance (LOCAMS). The CMN network features cameras with 16mm lenses, allowing for fainter meteor detections and better orbit fits. LOCAMS operates across the state of Arizona and records high-resolution trajectory data for hundreds of meteors per night. Both networks contribute their observations to the Global Meteor Network (GMN), which publishes publicly accessible datasets that include physically meaningful features.
We focused on directly observable parameters to ensure that our analysis is interpretable, scalable, and transferable across datasets. These include trail length, energy received, deceleration, peak brightness height, atmospheric density at multiple trajectory points (beginning, peak, end), mass and velocity in logarithmic form.
Before applying machine learning methods, the dataset was standardized using Python-based scikit-learn’s StandardScaler to ensure equal weighting across all features. This step is necessary for algorithms that rely on
To reduce dimensionality and highlight key patterns in the data, we applied Principal Component Analysis (PCA) to the normalized feature set. PCA identifies the principal axes along which the data varies most and projects the dataset onto these axes. The first two principal components capture approximately 95% of the total variance.
The PCA matrix indicates that the first principal component (PC-1) is driven primarily by energy received (0.773) and negatively by trail length (−0.608), with smaller contributions from mass and atmospheric density. This axis appears to reflect a combination of total energy and material penetration behavior. The second principal component (PC-2) is shaped by atmospheric density at the beginning, peak, and end of the trajectory (all 0.446), along with trail length (0.350) and mass (0.249), suggesting a link to fragmentation behavior and how a meteoroid interacts with varying atmospheric conditions during entry.
After dimensionality reduction, we applied a Median Absolute Deviation (MAD) filter to further identify and remove features that contribute minimal variability to the overall dataset. For each feature, MAD was calculated as the median of the absolute deviations from that feature's median value. Features with MAD values less than 50% of the median of all MAD values were excluded. This step refined the input to include only those features contributing the most physical diversity to the dataset, leaving eight features: energy received, deceleration, beginning, peak, and end atmospheric densities, trail length, peak magnitude, and mass in kg.
To identify potential groupings within the meteor population, we applied the scikit-learn’s K-Means clustering algorithm to the PCA-transformed dataset. The elbow method was used to determine the ideal number of clusters by plotting the within-cluster sum of squares against the number of clusters and identifying the point where additional clusters no longer significantly reduce variance. Based on this analysis, three clusters were identified.
Results
Unsupervised clustering applied to the PCA-transformed dataset revealed three groups of meteoroids, which we interpret as representing carbonaceous, asteroidal, and cometary materials. The first principal component is most strongly influenced by energy received and trail length, suggesting a relationship with material strength and penetration depth. The second component highlights variation in atmospheric density and mass, which likely reflects differences in deceleration and fragmentation behavior. These clusters appear well-separated in principal component space (Figure 1), indicating physically meaningful differences in meteoroid behavior.
The carbonaceous-type cluster shows intermediate values across most features. The asteroidal cluster includes high-energy meteors with low deceleration and deep atmospheric penetration, traits consistent with dense, rocky material. A subset of events within this cluster may represent iron-rich meteoroids, given their unusually high energy and deep penetration. The cometary cluster contains low-mass meteors with high deceleration and higher entry altitudes, aligning with expectations for fragile, porous cometary sources. These interpretations are supported by differences in initial velocity and begin height (Figure 2), with cometary meteors tending to enter faster and at higher altitudes. Figure 3 separates the clusters into individual subplots, illustrating how the density distributions vary across each classification.
Unlike traditional methods that rely on derived quantities and model assumptions, this classification is based entirely on directly observed features. Future work will focus on linking these clusters to known meteor showers to further compare to this model.
Figure 1: PCA scores plotted as a scatter plot and color coded by physical interpretation on the left. The same scatter plot transformed into a contour plot using kernel density estimation (KDE) on the right.
Figure 2: Initial velocity vs. beginning height color coded by their physical interpretation.
Figure 3: Figure 2 transformed into individual contour plots using KDE and titled by physical interpretation.
How to cite: Hemmelgarn, S., Moskovitz, N., and Vida, D.: A Machine Learning Application to Meteor Classification, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-442, https://doi.org/10.5194/epsc-dps2025-442, 2025.
This study presents a significant advancement in reconstructing meteoroid trajectories and speeds using the Belgian RAdio Meteor Stations (BRAMS) forward scatter radio network. We introduce an improved method based on a novel extension of the pre-t0 phase technique, initially developed for backscatter radars, and adapt it for continuous wave forward scatter systems. This approach leverages phase information recorded before the meteoroid reaches the specular reflection point t0 to enhance speed estimations. Furthermore, we combine this newly determined pre-t0 speed with time of flight measurements to reduce uncertainties in the reconstructed meteoroid paths and velocities. The robustness of our method is assessed using Markov Chain Monte Carlo techniques and validated against optical observations from the CAMS-BeNeLux network.
Measurement uncertainties
A critical aspect of reliable trajectory reconstruction is the accurate characterization of measurement uncertainties, particularly for the times of flight (Δt) between receiving stations. The uncertainty σΔt is closely tied to the uncertainty in determining the specular timing t0 at each station. We developed a method to determine the uncertainty σt0 as a function of two parameters: the rise time of the meteor amplitude curve (trise) and the signal-to-noise ratio (SNR).
To derive this relationship, we performed a series of Direct Monte Carlo (DMC) simulations. For each combination of trise and SNR, ideal meteor echoes were generated using the Cornu Spiral model, including some diffusion. For each SNR value, a large number of noisy clones of the ideal echo were created by adding Gaussian noise. The t0 values were extracted from these noisy echoes using the same post-processing chain as real observations, and the statistical spread σt0 was computed.
Solver improvement
Building on the measured uncertainties, we integrate them directly into the trajectory reconstruction process by redefining the cost function used by the solver:
where w is a weight parameter balancing the influence of time of flight (Ltof) and pre-t0 speed (Lpt0) measurements:
Optimizing this cost function across different values of w leads to the creation of a Pareto front representing the trade-off between minimizing the two components. The optimal solution is chosen at the "knee" of the curve, corresponding to the maximum curvature point.
Validation against optical observations
The reconstructed trajectories and speeds are compared to CAMS-BeNeLux optical data, which shows good agreement when a combination of time of flight and pre-t0 information is used. The differences are of the order of 5 % on the speed and 2-4° on the inclination.
Uncertainty propagation
To accurately quantify uncertainties on the reconstructed parameters, we employ a Markov Chain Monte Carlo approach. Assuming independent, Gaussian-distributed errors and uniform priors, the cost function L is proportional to the logarithm of the posterior probability. Thus, minimizing L is equivalent to maximizing the posterior. We use a Single Component Adaptive Metropolis-Hastings algorithm to efficiently explore the parameter space as well as to determine uncertainties and correlations.
How to cite: Balis, J., Lamy, H., Anciaux, M., Jehin, E., De Keyser, J., Kastinen, D., and Brown, P. G.: Improved meteoroid trajectory and speed reconstruction with BRAMS: pre-t0 phase technique and uncertainty quantification, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-1336, https://doi.org/10.5194/epsc-dps2025-1336, 2025.
Fossil micrometeorites (MMs) are mostly I-type cosmic spherules (CSs): iron rich cosmic dust particles that have survived atmospheric entry and later preserved in sedimentary rock [1] [2]. The quantity and composition of fossil MMs can advance our understanding of dust producing bodies in the solar system throughout Earth’s history, providing ground-truth empirical evidence of previous asteroid break-up events and cometary showers that would have otherwise remain undiscovered [3] [4] [5]. Their abundances and estimates of local sedimentation rates have been used to reconstruct the past flux of extraterrestrial dust [6] [7], but the influence of terrestrial or sampling processes ultimately affecting these calculations were ignored (e.g. diagenesis [8] or separation techniques). The cause for difference in CS concentration found throughout a Cenomanian succession in Dorset, England is discussed here, focusing on possible explanatory terrestrial and solar system processes utilising spherule quantification, petrography and geochemistry. Additionally, numerical models simulating changes in micrometeoroid size during entry heating are also used to study the possible entry parameters, sources, and solar system events acting as causes for changes in flux .
To determine changes in CS concentration, 2-10kg of chalk/marl from 5 horizons in the South England Chalk Group in Lulworth Cove, Dorset were analysed. Each sample was processed by a rock crusher and then placed into an electric vibrating sieve to separate the dust fractions (<250µm) which were then magnetically separated three times and optically examined to find CSs following protocols described in [1]. The CSs external petrographic textures and qualitative geochemistry were acquired using a desktop Hitachi TM4000Plus SEM and JEOL JXA-8530F EPMA.
The numerical models used closely resemble those discussed in [9]. Atmospheric deceleration was obtained using a simulation based on the model of [10] where deceleration is calculated from the momentum loss of spherical particles due to collisions with atmospheric gas molecules in its flight path. The final radius of particles is calculated using a balance of mass gained by oxidation (assuming every encountered oxygen molecule reacts and is incorporated with the molten iron) and evaporative mass loss.
All sediment samples returned numerous CSs, totalling 481 – the majority being I-types, but some are characterised as potential altered S-types (silicate rich CSs) or G-types (intermediate composition CSs). A minority presented unusual features such as cracks, mottled exteriors, and skeletal magnetite crystals (Figure 1). Most I-types showed similar chemical spectra with high Fe peaks, minor Ti and Mn implantation and variable amount of Si and Al (mostly on the exterior). Sample “Bed 8” yielded the highest number of spherules (>3x the others) normalised to 142 CS/kg of dust. The diameters ranged from 12µm to 130µm, and the mean values were between 24µm and 36µm which is comparable to other fossil MM studies [7] [8].
Almost all particles in the simulated conditions reach peak temperatures above the solidus for iron oxides (1809oK), causing them to melt and become CSs. To create the smallest sized I-types in this study (<25µm), ~25µm-sized micrometeoroids must enter at >14km/s at ~90o. At 18km/s <25µm I-types can be generated in a wider range of settings, encompassing micrometeoroids with initial sizes of up to ~350µm entering at >~75o. At higher velocities, evaporative loss and peak temperatures are too high for dust to survive in most set conditions.
The retention of characteristic quenched dendritic textures, preservation of intact metal beads, detectable Ni and Cr in some samples, and the sub-mm size and spherical nature of the particles indicate these Fe-oxides are extraterrestrial I-types. Since the sampled beds are geologically equivalent, contiguous, have similar accumulation periods and were processed using identical protocols, the high CS/kg concentration of Bed 8 likely represents a significant change in flux rates during the Cenomanian, rather than terrestrial or sampling influences. The enhancement in I-type accumulation rate (given in m-2.yr-1) was calculated by integrating power lines fitted on flux vs diameter plots for each bed and modern-day values (using [11] as the background; Figure 2). The ratios of the integrals reveal Bed 8 had an enhancement factor of >20x for I-types between 30 and 90µm in size, presenting a clear spike in flux.
The results of the numerical simulations show high entry velocities and angles are required to produce the small sizes of I-types discovered in this study, implying the dominating source of dust had an elliptical orbit such as from a cometary flyby. However, the lower power law exponents of the beds compared to those of modern day could indicate that most dust had lower velocities, which may suggest an asteroidal origin [10]. Further work will refine the possibilities of solar system events leading to this CS enrichment by discussing influences on these calculations and source assumptions.
[1] M. J. Genge, et al. (2008) Meteorit. Planet. Sci., 43(3): 497–515. [2] M. Genge, et al. (2020) Planet. Space Sci., 187: 104900. [3] B. Schmitz, et al. (1997) Science, 287: 88-90. [4] P. R. Heck, et al. (2008) Meteorit. Planet. Sci., 43: 517-528. [5] G.G. Voldman, et al. (2013) Geol. J., 48: 222–235 [6] S. Taylor and D. E. Brownlee (1991) Meteorit. Planet. Sci., 26(3): 203-211. [7] T. Onoue, et al. (2011) Geology, 39(6): 567-570. [8] M. D. Suttle and M. J. Genge (2017) Earth Planet. Sci. Lett., 476: 132-142. [10] S. G. Love & D. E. Brownlee (1991) Icarus, 89: 26-43. [11] J. Rojas, et al. (2021) Earth Planet. Sci. Lett., 560: 116794. [12] M. D. Suttle & L. Folco (2020) J. Geophys. Res. Planets, 125(2).
Figure 1. SEM images of fossil I-type cosmic spherules (CSs).
Figure 2 The cumulative I-type flux (m-2.yr-1) plotted against diameter of spherules. Dashed lines represent power lines fitted to the main population of spherules. The blue dashed lines are extrapolated fitted power lines for [11] and [12]. The shaded regions show errors in the sedimentation rate used. The flux assumes <5% of their collection consists of I-types. The grey dotted lines show the regions selected to represent medium-sized, and large spherules in the study (d = 30 to 60µm, 60 to 90µm).
How to cite: Mattia, I., Genge, M., Suttle, M., and Wong, A.: An Estimate of Cenomanian Cosmic Dust Flux: Implications on Their Identification, Sources, and Mass Contribution Calculations, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-1793, https://doi.org/10.5194/epsc-dps2025-1793, 2025.
Ursa Astronomical Association (https://www.ursa.fi/english.html) (founded 1921) is the biggest astronomical association in Finland. By membership count (over 19000 members) it is also one of the largest scientific associations in the world. Ursa runs an observational database called “Skywarden” (or “Taivaanvahti” in Finnish). Citizen observations of many types of celestial phenomena are collected including fireballs. The system is accessible both in English and Finnish. Ursa Astronomical Association also provides an umbrella organization for Finnish Fireball Network connecting scientists from different universities, camera station owners and citizen scientists.
The database contains about 20000 fireball observations in May 2025 (the system was established in 2011). Quite many of those observations are linked to the same fireball event even though there are also sporadic cases. The authors have developed an algorithm to group the fireballs belonging to a certain event. The algorithm runs automatically in the background. Altogether there are a few hundred cases of fireball events.
The Skywarden system had a crucial role in finding the Annama meteorite (Trigo-Rodriguez et al.). The video images posted to the database were used to determine the strewn field. In addition, a meteor on a hyperbolical track was identified from the user submitted images and the orbit of the meteor could be determined (Peña-Asensio et al.).
However, most of the observations in the database are without video images since they come from laymen. Usually, special hardware is required to retrieve images of meteors but nowadays surveillance cameras and dashcams do provide some material. In some cases, the citizen observations can provide auxiliary information to video images like if sound phenomena were observed. Daytime fireballs are rarely recorded since the special cameras are too sensitive during daylight. The authors wanted to investigate how well the citizen observations without any photos can be used to estimate the luminous track of the fireball.
The database collects information of the observer, date, location, duration of the phenomenon, bearing of the fireball, the apparent altitude, and the angle it came down. More information can be provided such as images, information on the sound phenomena etc. If the location is provided as a name, the system polls coordinate from Google Maps API. Skywarden provides data through its own API. It is possible to search for fireball events or displays and then fetch all the observations easily.
To analyze the data the data is fetched from Skywarden API. A Python Pandas based script preprocesses the data. All information that contains angular data (bearing, altitude, inclination angle) is added to the table. The data can then be exported in many formats like json, csv etc.
The authors developed two methods to analyze the data. The first one gives a crude estimate of the geographical location of the end of the luminous track only. For each observation pair the cross points of a major circle using bearings are calculated. Only those over Finland are preserved. The vector average of the point cloud is taken as the estimate. This is system is fast and is implemented as a web service that can calculate the estimate on the fly and visualize it on a map.
The authors developed also a second method that guesses a large set of the coordinates of the luminous track, then projects it to all observers and using least squares the best fit is then selected for the next round. This continues until the estimate isn’t improved significantly anymore. This method is slower and doesn’t give near real time results. However, it is possible to run in the background and thus provide an estimate for every fireball case automatically.
The Finnish Fireball Network does the estimation of the meteor tracks using video images. Some prominent fireballs with potential meteorite were used to compare the results obtained from the citizen observations only. The worst-case distance with the fast method was 73 km whereas with the least squares method the distance was 45 km. In average the quick method was off 44 km and the least squares method 26 km. A prominent fireball was seen in Vätsäri, Finnish Lapland on Nov 16, 2017. In Fig. 1 the results are presented on a map.
Figure 1 Results of Vätsäri case (Nov 16, 2017) on a map (Map data copyrighted OpenStreetMap contributors and available from https://www.openstreetmap.org). Red arrow denotes the flight path estimated from video observations, green path is the least squares estimated flight path, blue is obtained using the quick algorithm. The ellipsoid is the video based estimated strewn field.
The results were surprisingly good even though alone they are not sufficient to justify a field trip. The least squares algorithm is rather easy to use with mixed set of citizen observations and video-based estimates. In some cases, there is only one video observation and augmenting with citizen data would give at least some idea of the luminous track.
The spherical Earth model was used so integrating ellipsoid model and DEM would probably be a good next step of development. Also, the bearings are presented in quite a coarse manner. Maybe a mobile phone app would be appropriate to mitigate this. It would also help to improve the accuracy of grouping of observations.
References:
Trigo-Rodríguez, J.M., Lyytinen, E., Gritsevich, M., Moreno-Ibáñez, M., Bottke, W. F., Williams, I., Lupovka, V., Dmitriev, V., Kohout, T. and Grokhovsky, V. 2015, Orbit and dynamic origin of the recently recovered Annama's H5 chondrite, Monthly Notices of the Royal Astronomical Society, vol. 449, iss. 2, May 11, pp. 2119–2127, doi: 10.1093/mnras/stv378.
Peña-Asensio, E., Visuri, J., Trigo-Rodríguez, J. M., et al. 2024, Oort cloud perturbations as a source of hyperbolic Earth impactors, Icarus, Vol. 408, doi: 10.1016/j.icarus.2023.115844.
How to cite: Takala, M., Bruus, E., Moilanen, J., Mäkelä, V., and Pekkola, M.: Estimating fireball luminous tracks using citizen observations, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-305, https://doi.org/10.5194/epsc-dps2025-305, 2025.
Accurate classification of meteoroid streams is essential for understanding their evolution, dynamics, and source bodies. Conventional methods such as traditional orbital similarity metrics or look-up tables face limitations, as they are not mathematically consistent or rely on subjective criteria. In this work, we evaluate the performance of the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) algorithm [1] for unsupervised stream identification using data from the CAMS Meteoroid Orbit Database v3.0 [2], which contains 471,582 meteoroid entries observed from 2010 to 2016. After applying standard quality filters—convergence angle ≥15°, velocity uncertainty ≤10%, eccentricity e ≤ 1, and perihelion distance q ≤ 1 au—a total of 316,235 meteoroids are retained, of which ~70% are classified as sporadic by CAMS.
To characterize each meteoroid, we define three feature vectors. The LUTAB vector comprises the solar longitude (λ☉), the Sun-centered ecliptic radiant coordinates (α☉ and β☉), and the geocentric velocity (Vg), following the parameters used in the CAMS classification look-up table. The ORBIT vector consists of the heliocentric orbital elements: perihelion distance (q), eccentricity (e), inclination (i), argument of perihelion (ω), and the longitude of ascending node (Ω), similar to [3]. The GEO vector encodes the solar longitude, the geocentric radiant via sin(λg − λ☉)·cos(βg), cos(λg − λ☉)·cos(βg), and sin(βg), along with the geocentric velocity, following [4]. In all cases, normalization is applied: each feature is standardized to zero mean and unit variance to prevent dominance of any single variable during clustering, while angular parameters are represented in their sine and cosine components to address periodicity.
HDBSCAN is executed with a range of minimum cluster sizes (10–1000) and with two cluster selection strategies: ‘excess of mass’ (EOM) and ‘leaf’. Clustering performance is assessed using the Silhouette score (for internal compactness), Normalized Mutual Information (NMI) with respect to CAMS classifications, and F1 score following cluster-label assignment via the Hungarian algorithm. Sporadic meteors are treated as noise by HDBSCAN.
The ORBIT vector yields the highest Silhouette scores, indicating the most internally consistent clusters (see Figure 1). However, the LUTAB vector provides the best agreement with CAMS classifications (see Figure 2), achieving NMI = 0.76 and 25 out of 45 HDBSCAN-confirmed streams matching the CAMS labels with F1 > 0.8. Using ORBIT, only 7 out of 25 clusters reach F1 > 0.8. The GEO vector results are intermediate. The EOM cluster selection method consistently outperforms ‘leaf’, producing more stable and coherent cluster structures. Nevertheless, increasing the minimum cluster size improves compactness but reduces agreement with CAMS due to excessive merging of smaller showers.
We observe that HDBSCAN naturally groups streams with similar orbital geometry even when CAMS classifies them separately. For example, the Orionids (8/ORI) and Eta Aquariids (31/ETA), both from 1P/Halley, are clustered together when using the ORBIT vector. Likewise, HDBSCAN merges multiple minor showers (see Figure 3), reducing classification granularity but improving statistical coherence.
These findings show that HDBSCAN, when applied to normalized orbital or geocentric vectors, can provide a robust, parameter-efficient framework for meteoroid stream classification. Compared to CAMS look-up table, HDBSCAN yields more statistically consistent clusters. However, their physical validity remains to be verified.
Figure 1. Silhouette score vs. minimum cluster size for ORBIT (blue), GEO (red), and LUTAB (green); solid, dashed, and dash-dotted lines indicate each vector. Dark/light shades: EOM/leaf methods. Markers at size 100 with reference lines.
Figure 2. NMI vs. minimum cluster size, comparing HDBSCAN to CAMS for ORBIT (solid blue), GEO (dashed red), and LUTAB (dash-dotted green). Darker shades: eom; lighter: leaf.
Figure 3. X-axis: number of clusters; Y-axis: cluster size. Clusters sorted by size (largest on the left). Black: CAMS. Cyan: HDBSCAN with max NMI. Red: HDBSCAN with max Silhouette at min size 100. Yellow: HDBSCAN with overall max Silhouette.
Acknowledgements
The scientific activities of LUMIO are supported by the Italian Space Agency (ASI), whereas the Phase B engineering work has been conducted under ESA Contract No. 4000139301/22/NL/AS within the General Support Technology Programme (GSTP) through the support of the national delegations of Italy (ASI) and Norway (NOSA).
References
[1] Campello, R. J., Moulavi, D., & Sander, J. (2013, April). Density-based clustering based on hierarchical density estimates. In Pacific-Asia conference on knowledge discovery and data mining (pp. 160-172). Berlin, Heidelberg: Springer Berlin Heidelberg.
[2] Jenniskens, P., Baggaley, J., Crumpton, I., Aldous, P., Pokorny, P., Janches, D., ... & Ganju, S. (2018). Planetary and Space Science, 154, 21-29.
[3] Peña-Asensio, E., & Sánchez-Lozano, J. M. (2024). Advances in Space Research, 74(2), 1073-1089.
[4] Sugar, G., Moorhead, A., Brown, P., & Cooke, W. (2017). Meteoritics & Planetary Science, 52(6), 1048-1059.
How to cite: Peña-Asensio, E. and Ferrari, F.: Unsupervised Meteoroid Stream Identification Using HDBSCAN, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-606, https://doi.org/10.5194/epsc-dps2025-606, 2025.
Introduction: Cosmic objects impact Earth's atmosphere on a daily basis, but due to their their small size, they often go unnoticed before atmospheric interaction. To better constrain impactor size and characteristics, we require calibrated multi-detector observations of meteoroid impacts. In a previous study, Anghel et al. (2021) [1] explored techniques for measuring pre-atmospheric mass of meteoroids with well-known trajectories at the source of ton TNT-scale impacts. This resulted in an empirical correspondence relation: log(E) = 0.72 · log(Eo) + (0.6 ± 0.5), where E represents the impact energy and Eo is the optical energy.
Methods: For the current analysis, we focused on an additional set of eight well-documented atmospheric impacts with meteorite recovery. We explore the full range of published results for each impact, including the data for Winchcombe (UK) [2], Saint-Pierre-le-Viger (France) [3], and Ribbeck (Germany) [4]. The bolides did not have their total radiated energy estimated, hence, this was obtained by digitizing the published light curve, and converting it into TNT equivalent. Next, their kinetic energy was computed based on the published estimates of velocity and mass.
How to cite: Anghel, S.: Revisiting the Impact Energy Estimation of Well-Known Atmospheric Impacts, EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-1593, https://doi.org/10.5194/epsc-dps2025-1593, 2025.
Introduction: Brachinites are olivine-rich primitive achondrites which record planetary differentiation onset on asteroidal bodies, displaying equilibrated textures and homogenous mineral phases with oxidised mineral chemistry (Fa26-36, Fig. 1) [1]. Ungrouped brachinite-like achondrite meteorites (UBAs) share many similarities with brachinites [2], whose asteroidal counterpart, as well as for brachinites, has not been identified to date. Their close compositional and textural affinities pose questions about the possible genetic link among these two meteorite groups, i.e. whether they derive from a single heterogeneous or more compositionally affine parent body(ies) [3]. Studying the oxidation state of these meteorites helps to understand the nature and evolution of their chondrite parent body(ies). This information was previously only available indirectly through thermodynamic calculations [4]. Few works tried to directly constrain oxygen fugacities of brachinites source material, by measuring Cr valence in olivines of some brachinites and brachinite-likes [5], while V valences in chromites from different meteorite groups, including brachinites, were investigated in [6].
Our purpose is to obtain a large and complete cation valence dataset on chromites, low/high-Ca pyroxenes and olivines from brachinites and brachinite-like achondrites by means of X-ray absorption spectroscopy (XAS), to directly constrain oxygen fugacity of source material and possibly better define both the genetic relationship between them and which kind of material formed the(ir) parental body(ies).
Fig. 1. Fayalite mol% versus FeO-MnO ratio for brachinite and UBA studied olivines.
Samples and Method: A whole set of XAS measurements at V, Fe and Ti K-edges has been performed on 3 brachinites, NWA 4969, NWA 12733, NWA 13489, and 4 UBAs, AlH 010, MIL 090206, NWA 5400, NWA 6112 (Fig. 1).
Micro-XAS measurements were carried out at ID21 beamline, ESRF, using Si(311) monochromator crystals. Beam spot size on the sample surface was approximately 1000*500 nm2. Spectra were measured in fluorescence mode using a silicon drift detector. For each meteorite, we collected multiple spectra for each metal edge on several mineral grains after XRF elemental maps inspection (for minerals identification) of several meteorite areas.
Preliminary Results: Preliminary results from XANES on chromite crystals show that Fe is mainly present as Fe2+ in tetrahedral coordination while V is present as V2+. Quantitative EXAFS fit performed at V K-edge shows that divalent V is hosted in octahedra with a V-O distance of 2.00(1) Å and II and III shell distances (V-Cr=2.99(2) Å; V-Fe=3.49(2) Å in excellent agreement with V positioned at the 16d position.
V divalent state has been rarely described in natural samples [7], thus indicating extremely reduced oxygen fugacity values for brachinites and UBAs source material. We will expand XAS analysis to pyroxenes and olivines of all meteorite dataset, comparing valence states of Fe, V and Ti.
References: [1] Keil K. (2014) Chemie der Erde 74, 311–329. [2] Day J.M.D. et al., (2012) Nature Geosc. 5, 614-617. [3] Cuppone et al., in publication [4] Gardner-Vandy K.G. et al., (2013) Geochimica et Cosmochimica Acta 122, 36-57. [5] Goodrich C.A. et al., (2014) Geochimica et Cosmochimica Acta 135, 126-169. [6] Righter K. et al. (2016) American Mineralogist 101, 1928-1942. [7] Camara et al. (2019) Minerals 9, 4.
Acknowledgements: This work is supported by the ASI-INAF agreement n.2018-16-HH.0, Ol-BODIES project. We thank the Natural History Museum Bern and the National Institute for Polar Research for the meteorite samples (AlH010 and MIL 090206, respectively). We acknowledge the European Synchrotron Radiation Facility (ESRF) for provision of synchrotron radiation facilities (Exp. ES-1400) and we thank ID21 staff for the support during the measurements. This study was carried out within the Space It Up project funded by the Italian Space Agency, ASI, and the Ministry of University and Research, MUR, under contract n. 2024-5-E.0 - CUP n. I53D24000060005.
How to cite: De Santis, V., Cuppone, T., Lepore, G. O., Hole, C., Pratesi, G., and Giuli, G.: Fe and V oxidation state in chromites of brachinites and brachinite-like ungrouped achondrites. , EPSC-DPS Joint Meeting 2025, Helsinki, Finland, 7–13 Sep 2025, EPSC-DPS2025-1710, https://doi.org/10.5194/epsc-dps2025-1710, 2025.
Additional speaker
- Patrick Shober, NASA Johnson Space Center, United States of America