Authors: Sadykov, V.M. (GSU), Kosovichev, A.G. (NJIT), Kitiashvili, I.N. (NASA ARC), Oria, V. (NJIT), Nita, G.M. (NJIT), O'Keefe, P. (NJIT), Francis, F. (NJIT), Chong, C.J. (NJIT), Ali, A. (GSU), Marroquin, R. (GSU)

Robust prediction of Solar Energetic Particle (SEP) events and their absence (“all-clear” forecast) are among the key priorities of the space weather community. This presentation reports our progress on the project “Machine Learning Tools for Predicting Solar Energetic Particle Hazards,” supported by NASA’s Early Stage Innovation program. First, we highlight our progress in developing an online-accessible database that integrates various solar and heliospheric data, metadata, and descriptors related to SPEs. Specifically, we discuss the current data sources collected in the database, how to request the data via the Application Programming Interface (API), and the search and visualization capabilities of the web front-end. Second, we focus on the problem of “all-clear” forecasts of Solar Proton Events (SPEs) and our contribution to this problem conducted within the project. Specifically, we discuss: (1) the primary outcomes of our study of the daily probabilistic prediction of > 10 MeV > 10 pfu proton events for solar cycle 24 (2010-2020) and its comparison with operational forecasts, (2) an approach to extending this study to solar cycles 22-23 based on the statistical properties of soft X-ray and proton fluxes and the McIntosh and Hale classes of solar active regions, and (3) the development of flare-driven SPE forecasts based on the GOES soft X-ray measurements during solar flares.