juliensimon/neo-close-approaches

Near-Earth Object Close Approaches

Credit: NASA/Johns Hopkins APL

Part of a dataset collection on Hugging Face.

Dataset description

All close approaches of Near-Earth Objects (asteroids and comets) to Earth within 0.05 AU (~7.5 million km), spanning 1900 to 2100, from NASA JPL CNEOS. Updated daily.

This dataset contains every known close approach of a near-Earth object (NEO) to Earth, computed by NASA's Center for Near-Earth Object Studies (CNEOS) at the Jet Propulsion Laboratory. The data is recomputed continuously as new observations refine orbit estimates and new asteroids are discovered.

Each record includes the closest-approach distance (with 3-sigma uncertainty bounds), relative velocity, absolute magnitude, and -- where available -- measured diameter. For objects without a measured diameter, estimates derived from absolute magnitude using standard albedo assumptions are included.

Near-Earth objects are asteroids and comets whose orbits bring them within 1.3 AU of the Sun, placing them on trajectories that can intersect Earth's path. Close approaches within 0.05 AU (~19.5 lunar distances) are of particular interest for planetary defense. At these distances, gravitational perturbations from Earth can significantly alter an object's future orbit.

The distinction between past and predicted future approaches is scientifically important. Historical approaches are constrained by astrometric observations and have well-determined parameters. Future predictions depend on orbit propagation and degrade in accuracy over time, especially for small objects with short observation arcs or those subject to non-gravitational forces like the Yarkovsky effect.

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

Schema

Column	Type	Description	Sample	Null %
designation	str	Primary designation of the NEO (e.g. '433', '2024 YR4'); assigned by the Minor Planet Center; numbered objects use their permanent number, unnumbered use provisional designation	509352	0.0%
orbit_id	str	Orbit solution ID used for the close-approach computation; identifies which JPL orbit fit was used; changes when new astrometric observations refine the orbit	64	0.0%
close_approach_jd	float64	Close-approach time in Julian Date (TDB timescale); precise epoch of closest geometric approach to Earth; used for orbital mechanics calculations	2415024.433813976	0.0%
close_approach_date	datetime64[us]	Close-approach date and time in UTC; derived from the Julian Date for human-readable reference	1900-01-04 22:25:00	0.0%
distance_au	float64	Nominal closest-approach distance to Earth in astronomical units (AU); 1 AU = 149.6 million km; computed from the best-fit orbit solution	0.009632	0.0%
distance_min_au	float64	Minimum possible approach distance in AU at the 3-sigma confidence level; lower bound accounting for orbital uncertainty; tighter for well-observed objects	0.009625	0.0%
distance_max_au	float64	Maximum possible approach distance in AU at the 3-sigma confidence level; upper bound accounting for orbital uncertainty	0.009639	0.0%
velocity_relative_kms	float64	Relative velocity of the NEO with respect to Earth at closest approach in km/s; determines kinetic energy of a potential impact; typical range 5-30 km/s	8.68671	0.0%
velocity_infinity_kms	float64	Hyperbolic excess velocity (v-infinity) in km/s; the NEO's velocity relative to Earth at infinite distance; determines deflection mission requirements	8.654806	0.1%
time_uncertainty	str	3-sigma uncertainty in the close-approach time (e.g. '< 00:01' or '4_15:23' for days_hours:minutes); reflects how well the orbit is determined; large values indicate poorly constrained predictions	00:02	0.0%
absolute_magnitude	float64	Absolute magnitude H of the NEO; brightness at 1 AU from both Sun and observer at zero phase angle; proxy for size: H=18 ~ 1 km, H=22 ~ 140 m, H=25 ~ 40 m (size depends on unknown albedo)	20.15	0.0%
diameter_km	float64	Measured physical diameter in km from thermal IR (WISE/NEOWISE), radar, or occultation; null for the vast majority of NEOs; range from sub-km to tens of km	0.33	97.6%
diameter_sigma_km	float64	1-sigma uncertainty on the measured diameter in km; null when diameter itself is null	0.05	97.8%
full_name	str	Full formatted name/designation including permanent number and name where assigned (e.g. '433 Eros (1898 DQ)'); provides the most complete identification string	509352 (2007 AG)	0.0%
distance_ld	float64	Nominal closest-approach distance in Lunar Distances (1 LD = 384,400 km); more intuitive scale for close approaches; the Moon orbits at 1.0 LD	3.7485	0.0%
estimated_diameter_min_m	float64	Estimated minimum diameter in meters assuming a bright (albedo=0.25) S-type surface; computed from absolute magnitude when no measured diameter is available; null when measured diameter exists	248.1	2.5%
estimated_diameter_max_m	float64	Estimated maximum diameter in meters assuming a dark (albedo=0.05) C-type surface; computed from absolute magnitude when no measured diameter is available; null when measured diameter exists	554.7	2.5%
is_pha	bool	Potentially Hazardous Asteroid flag: True if absolute magnitude H <= 22 (roughly >= 140 m diameter) AND minimum approach distance <= 0.05 AU; these objects warrant continued monitoring	True	0.0%

Quick stats

• 42,031 close approaches (1900--2099)
• 32,110 past, 9,921 future predictions
• 4,249 involving Potentially Hazardous Asteroids
• 1,029 objects with measured diameters
• Closest recorded approach: (2025 UC11) at 0.02 LD (0.000044 AU)

Usage

from datasets import load_dataset

ds = load_dataset("juliensimon/neo-close-approaches", split="train")
df = ds.to_pandas()

from datasets import load_dataset

ds = load_dataset("juliensimon/neo-close-approaches", split="train")
df = ds.to_pandas()

# Upcoming close approaches sorted by distance
upcoming = df[df["close_approach_date"] > "2025-01-01"].sort_values("distance_au")

# Potentially hazardous approaches
pha = df[df["is_pha"] == True].sort_values("distance_au")

# Large objects (estimated > 100m) passing within 10 Lunar Distances
big_close = df[
    (df["estimated_diameter_max_m"] > 100) &
    (df["distance_ld"] < 10)
]

# Approaches per decade
import matplotlib.pyplot as plt
df["decade"] = (df["close_approach_date"].dt.year // 10) * 10
by_decade = df.groupby("decade").size()
by_decade.plot(kind="bar")
plt.xlabel("Decade")
plt.ylabel("Number of close approaches")
plt.title("NEO Close Approaches per Decade")
plt.show()

Data source

https://ssd-api.jpl.nasa.gov/doc/cad.html

Related datasets

• juliensimon/sentry-impact-risk

• juliensimon/fireball-bolide-events

• juliensimon/jpl-small-body-database

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

@dataset{neo_close_approaches,
  title = {Near-Earth Object Close Approaches},
  author = {juliensimon},
  year = {2026},
  url = {https://huggingface.co/datasets/juliensimon/neo-close-approaches},
  publisher = {Hugging Face}
}

License

CC-BY-4.0