juliensimon/space-weather-indices

---
license: cc-by-4.0
pretty_name: "Space Weather Indices (Kp, Ap, F10.7)"
language:
- en
description: "Daily geomagnetic and solar activity indices since 1957 from NOAA SWPC via CelesTrak — Kp, Ap, F10.7 flux, sunspot numbers."
task_categories:
- tabular-regression
- time-series-forecasting
tags:
- space
- space-weather
- geomagnetic
- solar
- noaa
- celestrak
- open-data
- kp-index
- f10.7
- sunspot
- solar-cycle
- swpc
- tabular-data
- parquet
size_categories:
- 10K<n<100K
configs:
- config_name: default
data_files:
- split: train
path: data/space_weather_indices.parquet
default: true
---

# Space Weather Indices

*Part of the [Space Weather Datasets](https://huggingface.co/collections/juliensimon/space-weather-datasets-69c24cae98f1666f2101ca70) collection on Hugging Face.*


Daily geomagnetic and solar indices from [CelesTrak](https://celestrak.org/SpaceData/),
which mirrors NOAA Space Weather Prediction Center data. Covers **1957-10-01** to present
with **25,015** observed days plus **230** days of predictions/forecasts.

## Dataset description

This dataset contains the fundamental indices used to characterize space weather conditions:

- **Kp/Ap indices** — Planetary geomagnetic activity (3-hourly and daily). Higher values indicate
stronger geomagnetic storms that cause satellite drag, GPS errors, and power grid disturbances.
- **F10.7 solar radio flux** — Proxy for solar EUV radiation that heats the upper atmosphere,
directly affecting satellite drag and orbital decay rates.
- **International Sunspot Number** — Long-running indicator of solar activity cycle.

## Schema

| Column | Type | Description |
|--------|------|-------------|
| `date` | datetime | Observation date |
| `bartels_rotation` | int | Bartels Solar Rotation Number (27-day cycle since 1832) |
| `bartels_day` | int | Day within Bartels cycle (1-27) |
| `kp_0000`–`kp_2100` | float | 3-hourly Kp index for each UT interval |
| `kp_sum` | float | Sum of eight daily Kp values |
| `ap_0000`–`ap_2100` | int | 3-hourly Ap index for each UT interval |
| `ap_avg` | float | Daily average Ap index |
| `cp` | float | Daily Character Figure (0-2.5) |
| `c9` | int | Converted Cp (0-9 scale) |
| `sunspot_number` | int | International Sunspot Number |
| `f107_obs` | float | Observed 10.7cm solar radio flux (sfu) |
| `f107_adj` | float | F10.7 adjusted to 1 AU |
| `f107_data_type` | string | Source: OBS (observed), INT (interpolated), PRD/PRM (predicted) |
| `f107_obs_center81` | float | 81-day centered average (observed) |
| `f107_obs_last81` | float | 81-day trailing average (observed) |
| `f107_adj_center81` | float | 81-day centered average (adjusted) |
| `f107_adj_last81` | float | 81-day trailing average (adjusted) |
| `is_storm` | bool | Geomagnetic storm flag (daily Ap >= 50) |
| `storm_level` | string | NOAA G-scale: G1 (minor) to G5 (extreme), based on max Kp |
| `data_type` | string | "observed" or "predicted" |

## Quick stats

- **25,015** observed days (1957-10-01 to 2026-03-27)
- **613** geomagnetic storm days (Ap >= 50)
- **24,896** severe storms (G3+)
- Strongest storm: Ap=280 on 1960-11-13

## Usage

```python
from datasets import load_dataset

ds = load_dataset("juliensimon/space-weather-indices", split="train")
df = ds.to_pandas()

# Only observed data (exclude predictions)
observed = df[df["data_type"] == "observed"]

# Geomagnetic storms
storms = df[df["is_storm"] == True].sort_values("ap_avg", ascending=False)

# Solar cycle visualization
df["year"] = df["date"].dt.year
yearly_ssn = df.groupby("year")["sunspot_number"].mean()

# F10.7 flux trend (drives atmospheric drag)
df.set_index("date")[["f107_adj"]].rolling(81).mean().plot()

# Storm frequency by solar cycle phase
df["cycle_phase"] = df["sunspot_number"].rolling(365).mean()
```

## Data source

[CelesTrak Space Weather Data](https://celestrak.org/SpaceData/), maintained by Dr. T.S. Kelso,
mirroring NOAA SWPC and GFZ Potsdam geomagnetic indices. The original indices are produced by the
International Service of Geomagnetic Indices (ISGI) and NOAA Space Weather Prediction Center.

## Update schedule

Daily at 11:00 UTC via [GitHub Actions](https://github.com/juliensimon/space-datasets).

## Related datasets

- [solar-flare-events](https://huggingface.co/datasets/juliensimon/solar-flare-events) — Individual solar flare detections from GOES
- [space-track-satcat](https://huggingface.co/datasets/juliensimon/space-track-satcat) — Full NORAD satellite catalog
- [space-track-tle-history](https://huggingface.co/datasets/juliensimon/space-track-tle-history) — 232M historical TLE records
- [neo-close-approaches](https://huggingface.co/datasets/juliensimon/neo-close-approaches) — Near-Earth object approaches

## Pipeline

Source code: [juliensimon/space-datasets](https://github.com/juliensimon/space-datasets)

## Support

If you find this dataset useful, please give it a ❤️ on the [dataset page](https://huggingface.co/datasets/juliensimon/space-weather-indices) and share feedback in the Community tab! Also consider giving a ⭐️ to the [space-datasets](https://github.com/juliensimon/space-datasets) repo.

## Citation

```bibtex
@dataset{space_weather,
author = {Simon, Julien},
title = {Space Weather Indices},
year = {2026},
publisher = {Hugging Face},
url = {https://huggingface.co/datasets/juliensimon/space-weather-indices},
note = {Based on NOAA/SWPC space weather data}
}
```

## License

[CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)