CoolFace
Datasetpublic

juliensimon/global-meteor-network

Global Meteor Network Trajectory Data Credit: NASA/ESA Part of a dataset collection on Hugging Face. Dataset description Individual meteor trajectory solutions from the Global Meteor Network (GMN), a worldwide network of 500+ all-sky cameras operated by volunteer astronomers. Each row is one detected meteor with orbital elements derived from multi-station triangulation. The GMN was founded in 2018 and has grown to cover all longitudes from Europe, the… See the full description on the dataset page: https://huggingface.co/datasets/juliensimon/global-meteor-network.

sourceHugging Facecc-by-4.0updated 2h agoView on Hugging Face
0likes509downloads
Dataset Card

Global Meteor Network Trajectory Data

<div align="center"> <img src="banner.jpg" alt="Rosetta spacecraft approaching Comet 67P/Churyumov-Gerasimenko" width="400"> <p><em>Credit: NASA/ESA</em></p> </div>

Part of a [dataset collection](https://huggingface.co/collections/juliensimon/orbital-mechanics-datasets-69c24caca4ab3934c9856994) on Hugging Face.

Dataset description

Individual meteor trajectory solutions from the Global Meteor Network (GMN), a worldwide network of 500+ all-sky cameras operated by volunteer astronomers. Each row is one detected meteor with orbital elements derived from multi-station triangulation.

The GMN was founded in 2018 and has grown to cover all longitudes from Europe, the Americas, Australia, and beyond. When at least two cameras simultaneously detect a meteor, the geometry of their positions allows triangulation of the atmospheric trajectory. Combined with timing, this yields the meteoroid's velocity at the top of the atmosphere and — after correction for Earth's gravitational attraction — the heliocentric orbit before encounter. The result is a complete set of Keplerian elements (a, e, i, ω, Ω, q) that places each detected meteoroid in the Solar System context.

Unlike the IAU Meteor Shower Database which catalogs mean radiant/orbit solutions per shower, this dataset contains individual meteor detections with full orbital parameters. The majority of detections are sporadic meteors (showercode = '...') with no known parent body; shower members are identified by matching with the IAU shower list. The nstations column provides a quality indicator: two-station solutions are the minimum for a valid trajectory, while higher counts improve the accuracy of both the radiant and the orbital elements.

This dataset is valuable for: identifying new meteor streams, studying the dynamical evolution of meteoroid trails, searching for meteoroids of potential interstellar origin (high eccentricity or retrograde orbits), correlating meteor detections with asteroid/comet close approaches, and building ML models for meteor source classification.

This dataset is suitable for tabular classification, tabular regression tasks.

Schema

ColumnTypeDescriptionSampleNull %
datetime_utcdatetime64[ns]Peak brightness datetime in UTC (format: YYYY-MM-DD HH:MM:SS.ff); used as the primary deduplication key; typically accurate to ±0.1 s2018-12-10 01:06:56.5881510.0%
shower_codestrIAU three-letter code for the meteor shower (e.g. 'PER' = Perseids, 'GEM' = Geminids); '...' = sporadic (not associated with any known stream)...0.0%
radiant_ra_degfloat64Geocentric radiant right ascension J2000.0 (degrees, 0–360); the apparent sky point from which meteors of this stream diverge, corrected for Earth's orbital velocity321.769120.0%
radiant_dec_degfloat64Geocentric radiant declination J2000.0 (degrees, -90 to +90); together with ra defines the meteor's approach direction in inertial space57.702210.0%
v_g_kmsfloat64Geocentric velocity at the top of the atmosphere before deceleration (km/s); range ~11 km/s (Earth-grazing) to ~72 km/s (retrograde head-on); determines meteor brightness and persistent train likelihood18.182620.0%
a_aufloat64Orbital semi-major axis of the meteoroid's heliocentric orbit (AU); NaN/inf for hyperbolic trajectories; Jupiter-family comets: 3-5 AU; Halley-type: 10-50 AU7.952210.0%
efloat64Orbital eccentricity (0 = circular, 1 = parabolic, >1 = hyperbolic); most shower meteoroids: 0.7–0.99; sporadic: wider range0.8766160.0%
i_degfloat64Orbital inclination to the ecliptic plane (degrees, 0–180); <90° = prograde (same direction as planets); >90° = retrograde; Perseids: ~113°, Leonids: ~162°23.6957530.0%
peri_degfloat64Argument of perihelion of the meteoroid orbit (degrees, 0–360); combined with node_deg locates the perihelion direction187.2721690.0%
node_degfloat64Longitude of the ascending node (degrees, 0–360); for Earth-crossing orbits, approximately equals the solar longitude at the shower's peak activity257.6490260.0%
q_aufloat64Perihelion distance of the meteoroid's orbit (AU); must be ≤ ~1.01 AU for Earth-crossing; values close to 1.0 AU indicate recent parent-comet ejection0.9811740.0%
peak_abs_magnitudefloat64Absolute magnitude at peak brightness (normalized to 100 km range); lower values = brighter; scale: -4 (fireball) to +7 (faint); used for mass and flux estimation0.050.0%
peak_height_kmfloat64Altitude above sea level at peak brightness (km); typical range 80–110 km; slower meteors peak higher; depends on velocity and meteoroid composition78.09390.0%
duration_secfloat64Total duration of the visible meteor trail in seconds; from first detection to last; fast meteors: 0.1-0.5 s; fireballs: up to 5-10 s1.290.0%
n_stationsInt64Number of GMN cameras that simultaneously detected this meteor; ≥2 required for trajectory solution; higher values indicate better geometry and orbital accuracy20.0%

Quick stats

  • 3,498,969 meteor trajectories (2018-12-10 to 2026-09-21)
  • 953,612 shower meteors (27%) and 2,545,357 sporadics (73%)
  • Top 5 showers by count: PER (174,287), GEM (104,735), ORI (52,467), SDA (44,684), ETA (33,443)
  • Median geocentric velocity: 41.4 km/s; fastest detected: 84.4 km/s

Usage

python
from datasets import load_dataset

ds = load_dataset("juliensimon/global-meteor-network", split="train")
df = ds.to_pandas()
python
from datasets import load_dataset
import pandas as pd

ds = load_dataset("juliensimon/global-meteor-network", split="train")
df = ds.to_pandas()

# Shower vs sporadic breakdown
print(df["shower_code"].value_counts().head(10))

# Velocity distribution by shower
import matplotlib.pyplot as plt
showers = df[df["shower_code"] != "..."]
top = showers["shower_code"].value_counts().head(6).index
showers[showers["shower_code"].isin(top)].boxplot(
    column="v_g_kms", by="shower_code", figsize=(10, 5)
)
plt.suptitle("")
plt.title("Geocentric Velocity Distribution by Meteor Shower")
plt.ylabel("v_g (km/s)")
plt.show()

# Radiant sky map
fig, ax = plt.subplots(figsize=(12, 6))
scatter = ax.scatter(
    df["radiant_ra_deg"], df["radiant_dec_deg"],
    c=df["v_g_kms"], s=0.5, cmap="plasma", alpha=0.3
)
plt.colorbar(scatter, label="v_g (km/s)")
ax.set_xlabel("RA (degrees)")
ax.set_ylabel("Dec (degrees)")
ax.set_title("GMN Meteor Radiants on the Sky")
plt.show()

Data source

https://globalmeteornetwork.org/data/

Update schedule

Daily at 10:00 UTC

Related datasets

If you find this dataset useful, please consider giving it a like on Hugging Face. It helps others discover it.

About the author

Created by Julien Simon — AI Operating Partner at Fortino Capital. Part of the Space Datasets collection.

Citation

bibtex
@dataset{global_meteor_network,
  title = {Global Meteor Network Trajectory Data},
  author = {Simon, Julien},
  year = {2026},
  url = {https://huggingface.co/datasets/juliensimon/global-meteor-network},
  note = {Derived from Global Meteor Network, https://globalmeteornetwork.org/data/},
  publisher = {Hugging Face}
}

License

CC-BY-4.0