SuryaBench ARPIL: Hourly SDO/HMI Binary Maps of Active Regions and Polarity Inversion Lines (2010-2024)

Provider
NASA-IBM AI4Science (Surya / SuryaBench team); Jinsu Hong, Kang Yang, Berkay Aydin (Georgia State University)
License
CC-BY-4.0 (source)
Size
7 files
Format
CSV (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/nasa-ibm-ai4science/surya-bench-ar-segmentation)
Access
Free and public. No account with the data provider is needed.
Category
Astronomy & Space
Source
https://github.com/NASA-IMPACT/Surya

About this dataset

Hourly binary segmentation masks of solar active regions (ARs) and polarity inversion lines (PILs) derived from SDO Helioseismic and Magnetic Imager (HMI) line-of-sight magnetograms, released by Hong, Yang and Aydin (Georgia State University) as the segmentation task of SuryaBench, the benchmark suite for the NASA-IBM Surya heliophysics foundation model. Each HDF5 file holds a (2, 4096, 4096) full-disk array with two keys: intersection, the binary map of strong-gradient PILs, and union_with_intersect, the map of active regions that contain those PILs. There are 128,352 hourly instances from 13 May 2010 to 31 December 2024, about 53 kB each and 1.2 GB in total. Splits follow the shared SuryaBench calendar: train.csv (15 February to 31 December of 2010 to 2019), validation.csv (15 to 31 January), leaky_validation.csv (1 to 14 January and 1 to 14 February) and test.csv (all of 2020 to 2024), each indexing timestamps and file paths so masks can be paired with the Surya core SDO stack (s3://nasa-surya-bench). Intended for magnetogram segmentation, flare-precursor feature extraction and AR tracking. Read with h5py. CC-BY-4.0 as stated in the card.

How to load this dataset in Python

import pandas as pd
df = pd.read_csv("hf://datasets/nasa-ibm-ai4science/surya-bench-ar-segmentation/leaky_validation.csv")
df.head()

Data structure

File types in sampled files: CSV

Files

Example file: leaky_validation.csv

Analyze this dataset without downloading it

On Scigantic, SuryaBench ARPIL: Hourly SDO/HMI Binary Maps of Active Regions and Polarity Inversion Lines (2010-2024) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Astronomy & Space datasets