University of Michigan researchers deployed 2 solar storm forecasting models for operational testing during the Artemis II transit on 1 April 1. The first is a machine-learning model that uses imagery from SDO (Solar Dynamics Observatory) and SOHO (Solar and Heliospheric Observatory) to generate daily storm probability estimates. The second is a physics-based simulation offering up to 24 hours' advance warning, requiring 3,000 processing units on a NASA supercomputer.
The G2 storm now active provides a live test environment that the research team could not have guaranteed. Apollo had no equivalent forecasting capability. The August 1972 solar particle event that fell between Apollo 16 and 17 arrived without warning. Had a crew been in transit, the consequences could have been severe.
Only the top 5% of solar particle events produce nausea-level radiation exposure 2. The current storm does not appear to approach that threshold. But the value of these models lies not in the storm that does not harm, but in the warning that arrives before the one that could.
If the models issue actionable warnings that align with observed conditions over the next 48 hours, they validate a prediction capability for all future crewed deep-space operations. If they miss, the gap between forecast and reality will itself be instructive data.
