Hugging Face's logo

Datasets:
juliensimon
/
lunar-eclipse-catalog

Tasks:
Tabular Classification
Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Size:
10K - 100K
Tags:
space
lunar-eclipse
eclipse
moon
astronomy
nasa
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
lunar-eclipse-catalog / README.md
juliensimon's picture
juliensimon
Upload lunar eclipse catalog: 12,064 eclipses
0541230 verified
preview code
|
raw
history blame contribute delete
8.04 kB
metadata 
license: cc-by-4.0
pretty_name: Five Millennium Catalog of Lunar Eclipses
language:
  - en
description: >-
  Complete catalog of lunar eclipses spanning five millennia (-1999 to +3000),
  computed by Fred Espenak as part of NASA's Five Millennium Canon of Lunar
  Eclipses.  A lunar eclipse occurs when the Moon p
task_categories:
  - tabular-classification
tags:
  - space
  - lunar-eclipse
  - eclipse
  - moon
  - astronomy
  - nasa
  - planetary-science
  - open-data
  - tabular-data
  - parquet
size_categories:
  - 10K<n<100K
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/lunar_eclipses.parquet
    default: true

Five Millennium Catalog of Lunar Eclipses

The Moon seen from Apollo 8, showing craters and surface detail

Credit: NASA/Apollo 8

Part of a dataset collection on Hugging Face.

Dataset description

Complete catalog of lunar eclipses spanning five millennia (-1999 to +3000), computed by Fred Espenak as part of NASA's Five Millennium Canon of Lunar Eclipses.

A lunar eclipse occurs when the Moon passes through Earth's shadow. The Moon can enter the faint penumbral shadow (producing a subtle darkening), the darker umbral shadow (producing a clearly visible partial eclipse), or become fully immersed in the umbra (a total lunar eclipse, often dramatically colored red by sunlight refracted through Earth's atmosphere). Unlike solar eclipses, lunar eclipses are visible from the entire night side of Earth simultaneously, making them historically important for synchronizing calendars across civilizations.

This catalog is derived from Fred Espenak's Five Millennium Canon of Lunar Eclipses, using the same computational foundation as the companion solar eclipse catalog: Besselian elements with polynomial expressions from Chapront, Chapront-Touze, and Francou for lunar and solar coordinates, with corrections for the secular acceleration of the Moon and variable Earth rotation via Delta-T.

The gamma parameter measures the signed minimum distance of the Moon's center from Earth's shadow axis. When gamma is between -0.9972 and +0.9972, the Moon is fully immersed in the umbra (total eclipse); between ±1.0260 it is partially in the umbra (partial eclipse); and between ±1.0620 only the penumbra is intersected. The umbral_magnitude gives the fraction of the Moon's diameter immersed in the umbra — values above 1.0 indicate a total eclipse. Red coloring during totality depends on Earth's atmospheric transparency at the time, described qualitatively by the Danjon scale (0 = very dark, 4 = bright copper-red).

This dataset is suitable for tabular classification tasks.

Schema

Column Type Description Sample Null %
catalog_number Int64 Sequential catalog number (1 to ~12,000), monotonically increasing with time 1 0.0%
date object Date of greatest eclipse in ISO format (YYYY-MM-DD); negative year values indicate BCE dates -1999-06-26 0.0%
year Int64 Calendar year; negative for BCE (e.g., -500 = 500 BCE); range -1999 to +3000 -1999 0.0%
td_greatest_eclipse object Time of greatest eclipse in Terrestrial Dynamical Time (TDT), format HH:MM:SS 14:13:28 0.0%
delta_t float64 Difference between Terrestrial Dynamical Time and Universal Time (TDT - UT) in seconds; large and uncertain for ancient dates (>10,000 s before 1000 CE) 46437.0 0.0%
luna_number Int64 Lunation count in Brown's series (consecutive full-moon numbering since 1923-01-17) -49456 0.0%
saros_number Int64 Saros series number; each series repeats every 18 years 11 days 8 hours and produces ~70-80 eclipses over ~1,300 years 17 0.0%
eclipse_type object Eclipse type code: T (total), P (partial), N (penumbral); may include + or - suffix for exceptional duration, or x/b/e suffix for subtype N 0.0%
eclipse_type_name object Full English name of eclipse type corresponding to the eclipse_type code Penumbral 0.0%
is_total bool True if the eclipse is a total lunar eclipse (Moon fully within Earth's umbra) False 0.0%
gamma float64 Signed minimum distance of the Moon's center from Earth's shadow axis in Earth equatorial radii; |gamma| < 0.9972 for total, < 1.0260 for partial, < 1.0620 for penumbral -1.0981 0.0%
umbral_magnitude float64 Eclipse magnitude in the umbral shadow: >1.0 for total, 0 to 1.0 for partial, negative for penumbral-only eclipses -0.1922 0.0%
penumbral_magnitude float64 Eclipse magnitude in the penumbral shadow; always >= umbral_magnitude; >1.0 triggers easily visible penumbral shading 0.8791 0.0%
partial_duration float64 Duration of partial phases (first/last umbral contact) in minutes; null for penumbral-only eclipses 102.7 36.3%
total_duration float64 Duration of totality (both umbral contacts) in minutes; null for partial and penumbral eclipses 98.2 71.2%
penumbral_duration float64 Total duration of penumbral contact in minutes from first to last penumbral contact 268.8 0.0%
latitude float64 Geographic latitude at the point of greatest eclipse in decimal degrees (+ = North, - = South) -24.0 0.0%
longitude float64 Geographic longitude at the point of greatest eclipse in decimal degrees (+ = East, - = West) -22.0 0.0%
century Int64 Derived century computed as year // 100; useful for aggregating eclipses by historical period -20 0.0%

Quick stats

  • 12,064 lunar eclipses (-1999 to 3000)
  • 3,479 total (28.8%), 4,207 partial (34.9%), 4,378 penumbral (36.3%)
  • Average totality duration for total eclipses: 80.8 min
  • Longest total eclipse in catalog: 106.6 min (~1.78 h)

Usage 

from datasets import load_dataset

ds = load_dataset("juliensimon/lunar-eclipse-catalog", split="train")
df = ds.to_pandas()

from datasets import load_dataset

ds = load_dataset("juliensimon/lunar-eclipse-catalog", split="train")
df = ds.to_pandas()

# Filter by eclipse type
total = df[df["is_total"] == True]
print(f"{len(total):,} total lunar eclipses across 5 millennia")

# Longest total eclipses (Blood Moons)
longest = total.nlargest(5, "total_duration")[["date", "total_duration", "gamma"]]
print(longest)

# Eclipses per century
import matplotlib.pyplot as plt
by_century = df.groupby("century")["eclipse_type_name"].value_counts().unstack(fill_value=0)
by_century.plot.bar(stacked=True, figsize=(14, 5), colormap="Set2")
plt.xlabel("Century")
plt.ylabel("Number of Eclipses")
plt.title("Lunar Eclipse Types by Century")
plt.tight_layout()
plt.show()

Data source

https://eclipse.gsfc.nasa.gov/LEcat5/LEcatalog.html

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 

@dataset{lunar_eclipse_catalog,
  title = {Five Millennium Catalog of Lunar Eclipses},
  author = {juliensimon},
  year = {2026},
  url = {https://huggingface.co/datasets/juliensimon/lunar-eclipse-catalog},
  publisher = {Hugging Face}
}

License

CC-BY-4.0