SEP-PRISM Data: A multi-source dataset for solar energetic particle forecasting

A curated, multi-source dataset for 24-hour-ahead forecasting of operational solar energetic particle (SEP) events, defined as GOES integral primary >10 MeV proton flux ≥ 10 pfu (NOAA operational threshold). The database integrates five solar-driver families—active-region magnetograms (SMHARP), flares, CME catalogues (CDAWDONKI), GOES soft X-ray flux (XRS-B), and GOES >10 MeV proton flux—with ground-truth SEP labels from the CLEAR Benchmark / FetchSEP catalogue. All records are harmonised into fixed rolling 24-hour windows suitable for supervised machine learning, benchmark comparison, and feature analysis. Coverage: 3 February 1986 – 10 September 2025 (UTC) for analysed rolling-window tablesCompanion publication: Yu et al., Scientific Reports Data Descriptor — [arXiv:2607.16160] Highlights Item Detail Primary ML table Analyzed Data/rolling_combinded_seq_24hours.csv — 14,464 rows × 274 columns (~33 MB) Operational SEP positives 650 / 14,464 rows (4.5%) with Future_OSEP_label = 1 General SEP positives 2,122 / 14,464 rows (14.7%) with Future_GSEP_label = 1 Predictor families SMHARP (full + flare-matched subset), flares, CDAWDONKI CME, CDAW CME, proton flux, XRS-B Extended archives SMHARP (SHARP extended to 1996 via SMARP); CDAWDONKI (DONKI-like CME features extended via CDAW) Reproducibility Full Python acquisition + R preprocessing/aggregation pipeline included Repository structure SEP-Prediction-Database/ ├── Raw Data/ # Upstream downloads and SEP event catalogue │ ├── GOES_integral_PRIMARY.*_sep_events.csv │ ├── flare_raw.csv, sharp_raw.csv, smarp_raw.csv │ ├── cdaw_cme_raw.csv, donki_cme_raw.csv, goes_10mev_flux_raw.csv │ ├── GOES-ProtonFlux/ # Per-satellite integral flux + status tables │ └── GOES-XRS/ # Per-satellite soft X-ray flux (GOES-06–18) ├── Processed Data/ # Fused, gap-filled feature tables │ ├── SMHARP.csv # ~5.31M rows, ~1.9 GB │ ├── Flare.csv, CDAWDONKI_CME.csv, CDAW_CME.csv │ ├── GOESHAPI_ProtonFlux.csv, Merged_XRays.csv │ └── All_features.RData # Bundled workspace for aggregation ├── Analyzed Data/ # Model-ready rolling-window tables │ ├── rolling_combinded_seq_24hours.csv # Primary benchmark table (~33 MB) │ └── rolling_combinded_seq_1hours.csv # 1-hour step, overlapping (~803 MB) └── Code/ ├── fetch_solar_data.ipynb # Python: download raw data from external archives ├── All_features_RData.R # Build R workspace from processed CSVs ├── Data_Aggregation.R # Build rolling-window ML tables ├── Dataset_Preprocessing/ # Fusion and feature-extraction scripts └── Plot/ # Figure-generation scripts ├── dataspan.R ├── sharp_smarp_plot.R └── Figures/