2026-10-14T15:23:16Z https://www.scidb.cn/oai

10.57760/sciencedb.space.02978 2026-02-11T10:32:58Z

2026-02-11 Solar Activity AI Forecaster Database doi:10.57760/sciencedb.space.02978 en SA-AI Forecaster Dataset Structure==================================Data used in SA-AI Forecaster modeling is stored in the "dataset" folder, which consists of four subfolders: - "Observation_image"- "SPNet_data"- "IATools_data"- "FPNet_data"1. Observation_image--------------------Observational image data (JPG format) used in modeling, categorized into:• Magnetograms: Solar photospheric magnetic field observations - SDO_HMI - SOHO_MDI - ASO-S_FMG• Coronal EUV images: Solar coronal atmospheric observations - SDO_AIA - SOHO_EIT - GOES-16_SUVI (Images of different wavelengths stored in separate subfolders)• Hα: Solar chromospheric Hα observations - BBSO - KSO - HSOS2. SPNet_data-------------Files for SPNet module input/output:• Labeling_data: Manually annotated ground truth• output_data: Module recognition outputBoth contain three feature subfolders: - AR (Active Region) - CH (Coronal Hole) - FL (Filament)3. IATools_data---------------TXT output from IATools module (Jan 2023 - May 2024):• AR_parameters: Active Region parameters - Daily_flare_events: Identified/located flare events - flare_index: Calculated flare indices - parameters: Other AR parameters• CH_parameters: Coronal Hole parameters• FL_parameters: Filament parameters4. FPNet_data-------------Training/testing data for FPNet module:• F10.7_index: Daily solar radio flux observations• FPNet-independent_data: Data from SWPC daily reports ▪ flare_index: Flare indices (Formula 1) ▪ label_M_class: M-class+ flare labels (24hr forecast) ▪ mag_type: Mount Wilson magnetic classification ▪ AR_parameters: Active region parameters (R-value, location, etc.)• Output: The output of the FPNet and FPNet* ▪ Full-disk_FPNet-independent: Full-disk forecasting result of FPNet* ▪ AR_FPNet-independent: AR forecasting result of FPNet* ▪ Full-disk_FPNet: Full-disk forecasting result of FPNet ▪ AR_FPNet: AR forecasting result of FPNetThe data used in the SA-AI forecaster modeling process is stored in the dataset folder, which consists of four subfolders: "Observation_image", "SPNet_data", "IATools_data", and "FPNet_data"."Observation_image" contains observational image data in JPG format used for modeling, categorized into three types: "magnetograms", "Coronal EUV images", and "Hα". The "magnetograms" subfolder stores solar photospheric magnetic field observation images from different sources: "SDO-HMI", "SOHO-MDI", and "ASO-S-FMG". The "Coronal EUV images" subfolder contains solar coronal atmospheric observation images from various sources: "SDO-AIA", "SOHO-EIT", and "GOES-16-SUVI", with images of different wavelengths stored in separate subfolders. The "Hα" subfolder preserves solar chromospheric Hα observation images from different institutions: "BBSO", "KSO", and "HSOS".The "SPNet_data" folder contains input and output files for building the SPNet module, with two subfolders: manually annotated ground truth data "Labeling data" and module recognition output data "output_data". Each subfolder is further divided into three feature-based subfolders: "AR", "CH", and "FL".The "IATools_data" folder stores output data (TXT format) from the IATools module between January 2023 and May 2024. The "AR_parameters" subfolder contains three subfolders: daily identified and located flare events "Daily_flare_events", flare indices calculated using formulas "flare_index", and other AR parameters "parameters". The "CH_parameters" and "FL_parameters" subfolders store CH and FL parameters respectively.The "FPNet_data" folder contains training and testing data for building the FPNet module. It includes "F10.7_index" (daily F10.7 index observation data required for modeling), "FPNet_data" (data obtained from daily reports provided by SWPC for building the flare prediction module FPNet), and "Output"(forecasting results of the FPNet and FPNet*). The "FPNet_data*" contains four subfolders: "flare_index" (flare indices calculated using formula 1 based on SWPC event list), "label_M_class" (M-class flare eruption labels for the next 24 hours derived from SWPC event lists), "mag_type" (AR Mount Wilson magnetic classifications obtained from SWPC solar active region summaries), and "AR_parameters" (AR parameters including R-values and locations, obtained from SWPC and manual annotations). The "Output" contains four files in .xlsx format, which are forecasting results of the FPNet and FPNet* on the full-disk and AR, respectively. Solar activity; Large model; AI forecaster; Space weather; Artificial intelligence Jingjing Wang Pengyu Liang Tingyu Wang Ming Li Yanmei Cui Siwei Liu Xin Huang Yunshi Zeng Zhu Cao Xiang Li Jiekang Feng Minghui Zhang Qinghua Hu Bingxian Luo Bing Cao PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank