juliensimon/neutron-monitor-cosmic-rays

Neutron Monitor Cosmic Ray Intensity (Hourly)

Credit: NASA

Part of a dataset collection on Hugging Face.

Dataset description

Hourly cosmic ray intensity measurements from the Neutron Monitor Database (NMDB) -- the worldwide network of ground-based cosmic ray detectors. Neutron monitors detect secondary neutrons produced when galactic cosmic rays interact with Earth's atmosphere.

The count rate is a proxy for cosmic ray intensity at Earth and is modulated by solar activity (11-year cycle), transient solar events (Forbush decreases), and geomagnetic conditions. Higher-latitude and higher-altitude stations have lower geomagnetic cutoff rigidity, making them more sensitive to lower-energy cosmic rays.

Galactic cosmic rays (GCRs) are relativistic charged particles -- predominantly protons and heavier nuclei -- accelerated to GeV-TeV energies by supernova remnant shocks. As they enter the heliosphere, their flux is modulated by the solar wind's outward-convecting magnetic field: during solar maximum, enhanced magnetic turbulence reduces the count rate at Earth by 15-25% compared to solar minimum. On shorter timescales, neutron monitors are the primary ground-based detectors for Forbush decreases -- sudden drops in cosmic ray intensity (typically 3-15% over hours) caused by the passage of interplanetary CMEs. Ground-level enhancements (GLEs) -- rare but dramatic increases in count rate -- signal the arrival of GeV-energy solar energetic particles from the most powerful solar flares.

This dataset is suitable for time-series forecasting, tabular regression tasks.

Schema

Column	Type	Description	Sample	Null %
datetime	datetime64[ns]	Observation timestamp (UTC, hourly averages from 1-minute raw counts)	2005-01-01 00:00:00	0.0%
station	str	NMDB station code (e.g. 'OULU' for Oulu Finland, 'JUNG' for Jungfraujoch Switzerland); location determines geomagnetic cutoff rigidity and therefore the cosmic ray energy threshold the station is sensitive to	"-//W3C//DTD	0.0%
count_rate	float64	Pressure-corrected and efficiency-corrected cosmic ray neutron count rate in counts per second; decreases during Forbush decreases (CME-driven CR depressions of 1-30%) and spikes during Ground Level Enhancements (GLEs, solar energetic particle events); null when station data are unavailable	101.758	4.2%
station_name	str	Human-readable station name and country (e.g. 'Oulu, Finland'); polar stations detect lower-energy cosmic rays (0.2 GV cutoff) while equatorial stations have higher cutoffs (15 GV)	"-//W3C//DTD	0.0%
daily_mean_count_rate	float64	24-hour mean count rate for this station on the same calendar day, in counts per second; used as the baseline for pct_deviation calculation	102.157	3.7%
pct_deviation	float64	Hourly count rate deviation from the daily mean in percent, i.e. (count_rate - daily_mean) / daily_mean x 100; negative values indicate suppressed cosmic ray flux (Forbush decrease); large positive values (>1%) may indicate GLE onset	-0.391	4.2%

Quick stats

• 473,600 hourly readings (2005-01-01 to 2026-04-22)
• 5 stations: "-//W3C//DTD, 4.01//EN", <!DOCTYPE, HTML, PUBLIC
• Mean count rate: 128.4
• Minimum count rate: 29.6 (<!DOCTYPE, 2018-04-05 02:00)
• Missing values: 4.2%

Usage

from datasets import load_dataset

ds = load_dataset("juliensimon/neutron-monitor-cosmic-rays", split="train")
df = ds.to_pandas()

from datasets import load_dataset

ds = load_dataset("juliensimon/neutron-monitor-cosmic-rays", split="train")
df = ds.to_pandas()

# Compare stations
import matplotlib.pyplot as plt

pivot = df.pivot_table(index="datetime", columns="station", values="count_rate")
pivot.resample("1M").mean().plot(figsize=(12, 5))
plt.title("Monthly Mean Cosmic Ray Intensity by Station")
plt.ylabel("Count Rate")
plt.tight_layout()
plt.show()

# Detect Forbush decreases (sudden drops in cosmic ray intensity)
oulu = df[df["station"] == "OULU"].set_index("datetime")["count_rate"]
daily = oulu.resample("1D").mean()
forbush = daily[daily.pct_change() < -0.03]  # >3% daily drop

# Solar cycle modulation
df["year"] = df["datetime"].dt.year
yearly = df.groupby(["year", "station"])["count_rate"].mean().unstack()
yearly.plot(figsize=(10, 5))
plt.title("Annual Mean Count Rate (Solar Cycle Modulation)")
plt.show()

Data source

https://www.nmdb.eu/

Update schedule

Daily at 14:30 UTC

Related datasets

• juliensimon/dst-index

• juliensimon/space-weather-indices

• juliensimon/solar-flare-events

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{neutron_monitor_cosmic_rays,
  title = {Neutron Monitor Cosmic Ray Intensity (Hourly)},
  author = {juliensimon},
  year = {2026},
  url = {https://huggingface.co/datasets/juliensimon/neutron-monitor-cosmic-rays},
  publisher = {Hugging Face}
}

License

CC-BY-4.0