Solar Situation Awareness and Flare Forecasting Dataset for Operational Applications

The dataset was developed to support solar situational awareness and flare forecasting applications by integrating multi-source solar observations with structured data products and region-level physical parameters. The released dataset consists of four main components: standardized solar observation images, situation awareness data products, solar flare forecasting data, and accompanying data-processing code resources.All data products are organized using a unified temporal indexing scheme, enabling direct association among observations, structural annotations, and derived parameters at the same observation time.Standardized solar observation imagesThe standardized solar observation image collection provides harmonized multi-instrument observations intended for multimodal solar activity analysis. This component includes magnetograms, continuum images, coronal extreme ultraviolet (EUV) observations, and H-alpha images.Line-of-sight magnetograms and continuum intensity images are derived from the SDO/HMI instrument and are processed through a unified spatial standardization pipeline with solar-disk centering and consistent image scaling. Coronal EUV observations are obtained from SDO/AIA at four wavelengths (131 Å, 193 Å, 211 Å, and 304 Å). All wavelength channels are spatially co-registered to enable pixel-level alignment across modalities. In addition, ground-based H-alpha observations are included to support filament identification.Solar situation awareness datasetsThe situation awareness component provides annotated representations of key solar structures, including ARs (AR), coronal holes (CH), and filaments (FL). For each structure type, both observation images and corresponding segmentation masks are provided. In AR folder, a correspondence file between mask values and NOAA AR is also provided.Segmentation masks are stored as NumPy arrays using integer encoding, where a value of 0 represents background pixels and values of other than zero correspond to individual structural instances. This representation allows direct use in region extraction, statistical analysis, and machine-learning workflows.AR, coronal hole, and filament datasets are stored as independent submodules to preserve semantic separation among structure types and to facilitate selective use depending on research objectives.Solar flare forecasting datasetsThe flare forecasting component is constructed at the active-region level and includes prediction labels together with region-based physical parameters. Flare label masks (folder name: mask_flare_label_M_1024) describe flare occurrence within a predefined 24 hours prediction window, while the flare index file (folder name: mask_flare_index_1024) is used to record the corresponding flare index information for different ARs.The active-region parameter table (folder name: AR_parameters) contains quantitative descriptors characterizing magnetic-field properties and structural complexity, including total unsigned magnetic flux, active-region area, R value, mean magnetic-field gradient, and polarity-inversion-line–weighted gradient. Hale magnetic classifications and McIntosh sunspot classes are also included to provide standardized morphological descriptors commonly used in operational forecasting studies.Code resourcesTo support reproducibility and data reuse, the dataset includes example processing scripts demonstrating the primary workflow used to generate standardized data products from raw observations. These scripts cover preprocessing procedures for magnetograms, continuum images, multi-wavelength AIA observations, and H-alpha data.