juliensimon/reentry-events

Reentry Events

Credit: NASA

Part of a dataset collection on Hugging Face.

Dataset description

Catalog of objects that have reentered Earth's atmosphere, derived from the NORAD Satellite Catalog (SATCAT) via CelesTrak. Includes launch and decay dates, orbital parameters, and time in orbit.

Every object launched into Earth orbit eventually returns -- whether through natural orbital decay, controlled deorbits, or breakup. This dataset catalogs every object in the NORAD SATCAT that has a recorded decay date, providing a comprehensive history of atmospheric reentries including payloads, rocket bodies, and debris.

Atmospheric reentry is governed by the interplay between altitude, ballistic coefficient, and solar activity. Objects in low orbits (below ~400 km) experience enough atmospheric drag to decay within months or years, while those at 800+ km can persist for centuries. Solar maxima heat and expand the upper atmosphere, dramatically increasing drag on LEO objects. Controlled reentries of large objects target uninhabited ocean areas (the 'spacecraft cemetery' in the South Pacific), while uncontrolled reentries carry a small but nonzero risk of surviving debris reaching populated areas.

This dataset is suitable for tabular classification tasks.

Schema

Column	Type	Description	Sample	Null %
norad_id	int32	NORAD catalog number -- sequential integer assigned by the 18th Space Defense Squadron at launch; primary key for cross-referencing with TLE databases and the SATCAT	47374	0.0%
object_name	str	Official name as listed in the NORAD catalog (e.g. 'COSMOS 1234 DEB'); rocket bodies typically include 'R/B' and debris fragments include 'DEB' in the name	STARLINK-2093	0.0%
object_type	str	Object classification: 'PAY' (payload/spacecraft), 'R/B' (rocket body or upper stage), 'DEB' (fragmentation debris), 'UNK' (unknown/unclassified)	PAY	0.0%
country_code	str	Two- or three-letter country or organization code (e.g. 'US', 'CIS', 'PRC', 'ISS' for ISS-associated objects) identifying the launch owner as assigned in the SATCAT	US	0.0%
launch_date	datetime64[us]	Date the object was launched into orbit (UTC); null for a small number of objects with incomplete catalog entries	2021-01-20 00:00:00	0.0%
decay_date	datetime64[us]	Date the object reentered Earth's atmosphere (UTC); for uncontrolled reentries this is the date radar tracking was lost; for controlled reentries it is the planned impact date	2026-07-03 00:00:00	0.0%
period_min	float64	Last recorded orbital period in minutes before reentry; LEO objects typically 88-128 min; null if no orbital period was recorded in the final catalog entry	88.27	4.0%
inclination_deg	float64	Last recorded orbital inclination in degrees (0-180); the angle between the orbital plane and the equatorial plane; polar orbits ~90 deg, equatorial ~0 deg	53.02	4.0%
apogee_km	float64	Last recorded apogee (highest point) altitude above Earth's surface in km; null if not recorded; typically low and declining for objects near reentry	191.0	4.0%
perigee_km	float64	Last recorded perigee (lowest point) altitude above Earth's surface in km; null if not recorded; objects with perigee below ~200 km reenter within days to weeks	187.0	4.0%
rcs	float64	Radar cross-section in m-squared; proxy for object size used by space surveillance radars; null for many objects where RCS was not published or was too small to measure reliably	0.056	48.3%
days_in_orbit	int64	Number of days between launch and reentry (decay_date minus launch_date); null if either date is missing; ranges from 0 (immediate reentry) to tens of thousands of days	1990	0.0%
decay_year	Int32	Calendar year of reentry; derived from decay_date for efficient grouping and time-series analysis	2026	0.0%

Quick stats

• 35,316 reentered objects
• 7,469 payloads, 23,285 debris fragments, 4,453 rocket bodies
• Median time in orbit: 939 days
• Reentries span 1957 to 2026
• Top countries: CIS (18,516), US (10,566), PRC (3,180), FR (837), JPN (590)

Usage

from datasets import load_dataset

ds = load_dataset("juliensimon/reentry-events", split="train")
df = ds.to_pandas()

from datasets import load_dataset

ds = load_dataset("juliensimon/reentry-events", split="train")
df = ds.to_pandas()

# Reentries per year
reentries_by_year = df.groupby("decay_year")["norad_id"].count()
print(reentries_by_year.tail(10))

# Longest-lived objects
longest = df.nlargest(10, "days_in_orbit")[["object_name", "days_in_orbit", "launch_date", "decay_date"]]
print(longest)

# Reentry trend over recent decades
import matplotlib.pyplot as plt
yearly = df[df["decay_year"] >= 1970].groupby("decay_year")["norad_id"].count()
yearly.plot(kind="bar", figsize=(14, 5), edgecolor="black", width=0.8)
plt.xlabel("Year")
plt.ylabel("Reentries")
plt.title("Annual Atmospheric Reentries")
plt.tight_layout()
plt.show()

Data source

https://celestrak.org/pub/satcat.csv

Related datasets

• juliensimon/space-track-satcat

• juliensimon/space-launch-log

• juliensimon/space-track-tle-history

• juliensimon/constellation-census

• juliensimon/gcat

• juliensimon/gcat-satcat

• juliensimon/launch-cost

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{reentry_events,
  title = {Reentry Events},
  author = {juliensimon},
  year = {2026},
  url = {https://huggingface.co/datasets/juliensimon/reentry-events},
  publisher = {Hugging Face}
}

License

celestrak-usage-policy