AI Predicts Solar Storm Strength, Here Are the Dire Effects on Earth

Solar storms can have serious impacts on various critical infrastructures on Earth, ranging from satellites, communication networks, and navigation systems to electricity grids. Therefore, the ability to predict the strength of a solar storm is crucial for reducing the risks it poses. A researcher from the National Research and Innovation Agency’s (BRIN) Space Research Centre, Tiar Dani, explained that the danger level of a solar storm is heavily influenced by the direction of the interplanetary magnetic field it carries. In space physics, this component is known as Bz. According to him, when the interplanetary magnetic field points southward for several hours, its interaction with Earth’s magnetic field becomes much stronger. This condition makes Earth’s magnetic shield more easily penetrated by particles and energy from the Sun, thus triggering a geomagnetic storm. “Just like facing a storm on land, we also need to know how strong a solar storm is so we can prepare appropriate mitigation measures,” Tiar said. He explained that geomagnetic storms have strength levels ranging from mild (G1) to extreme (G5). At a mild level, the impacts are generally limited to minor fluctuations in the power grid and the appearance of auroras. “On a mild scale, the impact may only be small fluctuations in the electricity network and the appearance of auroras. However, if the Bz value is very negative and persists for a long time, the storm can penetrate the extreme scale (G5),” he stated. Tiar noted that at the highest level, the threats posed are far greater. Very strong induced electrical currents can damage transformers and trigger widespread power outages, such as the one that occurred in Quebec in 1989. Furthermore, radio communications and aviation navigation systems could be disrupted globally. The density of the upper atmosphere can also increase, putting dozens of satellites at risk of losing their orbits. To anticipate these threats, BRIN is developing an artificial intelligence (AI) system based on multi-modal deep learning called Bz4SWx. This system is designed to predict the minimum Bz value and the time of its occurrence up to 96 hours, or four days, after a coronal mass ejection (CME) from the Sun. In the process, the AI not only analyses the speed and direction of the CME but also studies images of the Sun’s magnetic field, or magnetograms. By combining these various data types, the system can predict the magnetic field pattern of a solar storm before it reaches Earth. “By combining these various types of data, the AI gains a more comprehensive understanding of the magnetic field characteristics being formed, enabling it to predict the Bz pattern that will occur from the CME up to four days later,” he explained. Tiar added that the system also utilises attention mechanism technology, which allows the AI to focus its analysis on areas of the Sun most likely to trigger geomagnetic storms. “The way it works is similar to the human eye, which automatically focuses on the most striking or potentially dangerous objects in its surroundings,” he said. Test results show that the AI system can predict the intensity of a solar storm’s magnetic field with a promising degree of accuracy, while also providing an early warning up to four days before the storm reaches Earth. “With faster and more accurate information, mitigation measures can be taken earlier to minimise the risk of disruption caused by extreme space weather,” Tiar stressed.