README.md

metadata

license: mit
tags:
  - heliophysics
  - solar_active_regions
  - AR_emergence
  - space_weather
  - machine_learning
size_categories:
  - n<1K
configs:
  - config_name: default
    default: true
    data_files:
      - split: train
        path: train.csv
      - split: validation
        path: validation.csv
      - split: test
        path: test.csv
pretty_name: Active Region Emergenc

Active Region Emergence Dataset

Dataset Summary

The Active Region Emergence Dataset is designed to support research on the early detection of solar Active Regions (ARs) and the development of predictive models for space weather. By characterizing the evolution of ARs before, during, and after their emergence, the dataset enables studies of pre-emergence signatures and early warning methods.

This dataset is derived from NASA’s Solar Dynamics Observatory (SDO) using measurements from the Helioseismic and Magnetic Imager (HMI). It includes timeline data of:

• Acoustic power (from Doppler velocity maps)
• Photospheric magnetic field
• Continuum intensity

for 59 large ARs that emerged on the visible solar disk between 2010 and 2023. Each AR is tracked within a 30° × 30° patch over multiple days.

These data products have already been applied successfully in machine learning models for AR emergence forecasting in [1].

Supported Tasks and Applications

• Spatio-temporal Forecasting: Forecasting the spatially resolved emergence characteristics of active regions.

Data Structure

Data Files

The repository contains csv files containing paths and compressed data files for emergence characteristics. The dataset is split for consistent training, validation and testing.

• train.csv — training split (37 ARs)

• valid.csv — validation split (9 ARs)

• test.csv — test split (13 ARs)

• README.md — dataset description

• data.zip — compressed folder containing all Active Region (AR) subfolders:

  • AR{noaa_num}/
    • mean_int{noaa_num}_flat.npz → continuum intensity timeline
    • mean_mag{noaa_num}_flat.npz → magnetic field timeline
    • mean_pmdop{noaa_num}_flat.npz → 4 acoustic power timeline (2–3, 3–4, 4–5, 5–6 mHz).
      • Index 0: 2-3 mHz
      • Index 1: 3-4 mHz
      • Index 2: 4-5 mHz
      • Index 3: 5-6 mHz

where noaa_num is the NOAA Active Region Number for the active regions in this dataset.

Features

CSV files include following features:

• AR: NOAA Active Region number
• t_start: Start time of tracked patch
• t_end: End time of tracked patch
• mean_int_path: Path to continuum intensity .npz file
• mean_mag_path: Path to magnetic field .npz file
• mean_pmdop_path: Path to acoustic power .npz file

Dataset Details

Field	Description
Temporal Coverage	2010 – 2023
Data Format	CSV for metadata / NPZ for rasters
Data Size	Total 59 instances
Total File Size	~23.1 MB
Data Shape	(6, 240, 9, 9) per instance
Cadence	1 hour

Example Usage

import numpy as np
import pandas as pd

# Load CSV metadata
df = pd.read_csv("train.csv")
print(df.head())

# Load one AR’s power map dopplergram data
# for intensity and magnetic flux use mean_int_path and mean_mag_path column names on df
sample_path = df.iloc[0]["mean_pmdop_path"]

# Update to local path after unzipping data.zip
sample_path = sample_path.replace("/data", "data")  

data = np.load(sample_path)
print("Keys in npz file:", data.files)
print("Data shape:", data[data.files[0]].shape)

Contact

[1] Spyros Kasapis [email protected]

References

[1] Kasapis, S., Kitiashvili, I. N., Kosovichev, A. G. & Stefan, J. T. Prediction of intensity variations associated with emerging active regions using helioseismic power maps and machine learning. The Astrophys. J. Suppl. Ser. 10.3847/1538-4365/adfbe2 (2025)