CAIG: AI-Driven Generation of Vector Magnetograms in the Chromosphere and Photosphere with Application to Explainable Solar Eruption Predictions

Solar eruptive events, such as flares and coronal mass ejections (CMEs), are major sources of space weather that can impact the near Earth environment adversely. They are known to cause malfunction on critical technology infrastructure such as satellite and power distribution networks. Therefore, understanding and forecasting solar eruptions is critical for national security and for the economy. It is well known that the structure and evolution of solar magnetic fields, especially in the corona (the Sun's upper atmosphere), are determining factors in storing energy and triggering harmful events. This project will develop a generative AI framework for quantifying risks of solar eruptions. The AI-generated products will provide valuable data on the Sun's magnetic field in the chromosphere (the Sun's lower atmosphere), in which observations are difficult and rarely obtained. This research will significantly advance understanding of the onset of solar eruptions and their predictions. All AI methods, data, and relevant scientific results will be distributed to a wide research community. The project will foster a collaboration between AI experts and solar physicists, integrating research and education through interdisciplinary learning and training activities.This project will develop advanced computing capabilities to characterize solar active regions (ARs) and apply machine learning tools to predict solar eruptions. With a new generative AI framework, named SolarDM, the following tasks will be carried out: (1) generate chromospheric vector magnetograms with an archive combining SDO/HMI photospheric magnetograms, H-alpha images and SOLIS chromospheric magnetograms; (2) create consistent high-resolution, high-cadence non-gapped time series photospheric vector magnetograms for Solar Cycle 23 based on previously AI-generated SOHO/MDI vector magnetograms; (3) with the synthetic data of high spatial-temporal resolution from both the chromosphere and photosphere, develop explainable AI (XAI) methods to make deterministic and probabilistic predictions of solar eruptions, and address two key science questions: (i) What are the essential roles of certain physical parameters, including dynamic and "true" 3-D magnetic parameters, in determining the AR productivity of eruptive events? (ii) Will these physical parameters improve the accuracy of forecasting flares and CMEs?This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.