Improved Probabilistic Forecasts of Solar Energetic Particles with MagPy

NASA's strategy for returning to the Moon is part of its broader Moon-to-Mars exploration, of which the Artemis program is a critical component. A crucial aspect of this concerns the safety of the crew and their hardware, of which solar energetic particles (SEPs) play a major role. MagPy (based on MAG4) is a probabilistic forecasting tool designed to predict solar events such as flares, (fast) CMEs, and SEPs using SDO/HMI SHARP magnetogram data. It forecasts event probabilities within 24 hours by analysing magnetic parameters like strong-field neutral line length, magnetic flux areas over 100 Gauss, and magnetic field shear and gradient. MagPy is the Python-based evolution of MAG4 and is now maintained and developed by NASA's Space Radiation Analysis Group (SRAG). It offers operational outputs, including graphical active region representations and a "Threat Gauge" for event probabilities, making it a core tool for predicting solar activity. We propose to refine and substantially improve MagPy's capabilities. During Phase I, we will focus on developing a robust implementation of the MagPy tool, incorporating field line connectivity, investigating other refinements, and demonstrating a successful prototype tool in preparation for Phase II. During Phase II, we will add a number of other refinements, including, but not limited to (1) improved datasets incorporating Solar Orbiter's PHI images off the Sun-Earth line; (2) a flux transport model to capture the evolution of ARs better as they propagate across the solar disk; (3) a complementary approach for calculating the likelihood of a flare/CME; (4) an alternate PIL detection routine; and (4) robust statistical and machine learning (ML) methodologies for both input data selection and model validation and verification. Additionally, we will work with NASA personnel to develop and refine robust metrics and skill scores for forecast verification and validation.