# SuryaBench Full-Disk Solar Flare Forecasting Labels: Hourly 24-Hour GOES X-ray Flare Windows (2010-2024)

## Facts

- Provider: NASA-IBM AI4Science (Surya / SuryaBench team)
- License: CC-BY-4.0 (https://huggingface.co/datasets/nasa-ibm-ai4science/surya-bench-flare-forecasting)
- Size: 7 files
- Format: CSV (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/nasa-ibm-ai4science/surya-bench-flare-forecasting
- Access: public, readable without an account
- Category: Astronomy & Space
- Source: https://github.com/NASA-IMPACT/Surya
- Page: https://scigantic.com/archive/f978e621-2f10-5b44-977a-64e07f967739
- Keywords: Astronomy & Space, heliophysics, benchmark, CSV, Hugging Face

## About

Hourly labels for full-disk solar flare forecasting from NOAA GOES X-ray flare event lists, the flare task of SuryaBench, the benchmark suite for the NASA-IBM Surya heliophysics foundation model. Each timestamp from 13 May 2010 to 31 December 2024 opens a 24-hour rolling prediction window annotated with the maximum GOES class (peak 1-8 Angstrom X-ray flux, or FQ for flare-quiet), a cumulative flare index summing C, M and X subclasses with weights 1, 10 and 100, and two binary targets: label_max (window contains a flare of M1.0 or above) and label_cum (cumulative index of 10 or more). The labels support binary, ordinal (flare class) and regression (cumulative index) formulations and are meant to be paired with full-disk SDO AIA and HMI inputs such as the Surya core stack on s3://nasa-surya-bench. CSV splits follow the shared SuryaBench calendar: train (15 February to 31 December of 2010 to 2019), validation (15 to 31 January), leaky_validation (1 to 14 January and 1 to 14 February) and test (2020 to 2024). Small (a few MB); load with pandas or Hugging Face datasets. CC-BY-4.0 as stated in the card.

## Load the data

```python
import pandas as pd
df = pd.read_csv("hf://datasets/nasa-ibm-ai4science/surya-bench-flare-forecasting/data.csv")
df.head()
```

## Data structure

File types in sampled files: CSV

Example file: `data.csv`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/f978e621-2f10-5b44-977a-64e07f967739
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/f978e621-2f10-5b44-977a-64e07f967739
- Over MCP: `get_schema_card` with id `f978e621-2f10-5b44-977a-64e07f967739` at https://api.scigantic.com/mcp

## Related topics

- [CSV datasets](https://scigantic.com/topics/csv)
- [Benchmark datasets](https://scigantic.com/topics/benchmark)
- [Astronomy & Space datasets](https://scigantic.com/topics/astronomy-and-space)
- [Heliophysics datasets](https://scigantic.com/topics/heliophysics)
