Fujitsu tool boosts forecasts of solar radiation events

Fujitsu and Japanese research institutions have developed a technology that automatically optimises simulations used to predict the arrival of solar energetic particles. The group said it improves forecast accuracy for radiation events that can disrupt satellites and threaten astronauts.

The technology was developed with Nagoya University, the Japan Agency for Marine-Earth Science and Technology, and the National Institute of Information and Communications Technology, along with research partners in Japan and Portugal. It uses observational data from five spacecraft to set the parameters for physical simulations that calculate when and where solar energetic particles will arrive.

Solar energetic particles are high-energy protons and electrons released by solar flares and coronal mass ejections. They can damage satellite electronics, interrupt communications, and increase radiation exposure for crews in space.

Until now, researchers have largely relied on manual trial and error to adjust the inputs used in these simulations. That process depended heavily on expert judgement, making it difficult to justify parameter choices objectively or reproduce results consistently.

The new system automates that work by setting 19 parameters in a physical simulation that calculates particle intensities and arrival times. It aims to minimise the gap between simulated results and observations collected from five spacecraft positioned at different points in space.

At the centre of the work is SEITEN, short for Solar Energetic Particle Intelligent Tuning and Evaluation Network, developed by Fujitsu. The platform links SUSANOO, a solar wind simulation created by Nagoya University and operated by NICT, with SEPIF, a particle acceleration and transport model developed by JAMSTEC.

The system uses black-box optimisation to automate the search for suitable parameter values without manual intervention. The work also drew on interplanetary scintillation observational data from Nagoya University to select settings for the magnetohydrodynamic simulation.

Simulation challenge

Predicting these events is difficult because the particles are electrically charged and scatter as they travel along magnetic field lines extending from the Sun. Small changes in input values can significantly alter predicted particle intensities, limiting the reliability of earlier approaches.

The 19 parameters in the study included seven mean free path values linked to scattering strength and 12 injection efficiency values linked to the share of particles entering the acceleration process at a shock wave. Testing combinations manually would be impractical, especially when trying to match observations from several distant locations at once.

The automated method carried out 1,480 trials in total, with 1,000 broad trials followed by 480 more focused trials in the most promising areas.

Applied to a solar energetic particle event observed by five spacecraft, the method broadly reproduced the onset, rise, and peak of particle intensity across all five. The same overall trend was confirmed in independent repetitions using a procedure and evaluation criteria defined in advance.

Wider importance

The work comes as governments and space agencies place greater emphasis on sustained activity beyond Earth orbit and on the Moon. More accurate forecasts of solar radiation events could help satellite operators assess risks earlier and support planning for human operations in space.

The study also has implications for sectors on Earth that depend on uninterrupted satellite services, including communications, maritime logistics, agriculture, and environmental monitoring. In markets such as Australia and New Zealand, those systems support services ranging from bushfire and flood monitoring to remote connectivity.

Mahesh Krishnan, Chief Technology Officer for Oceania at Fujitsu, said: "Predicting space weather has always been a massive roadblock for the space sector. It's an extremely volatile environment where a single solar event can disrupt the critical satellite data our everyday life relies on. This technology will play a key role in preventing critical satellite outages, protecting the everyday communication networks we rely on, and safely building the space industries for tomorrow."

The research was published in Scientific Reports and focused on protons released during a solar event observed by BepiColombo, Solar Orbiter, Parker Solar Probe, STEREO-A, and SOHO.