Chinese researchers have developed an artificial intelligence system capable of automatically detecting space hurricanes in Earth’s upper atmosphere. Scientists at the National Space Science Center under the Chinese Academy of Sciences announced the breakthrough on Tuesday. The system achieves a detection accuracy of nearly 97.9 percent, significantly outperforming all previously existing identification methods. Furthermore, researchers say the tool could directly support polar communications safety and navigation systems worldwide.
Space hurricanes are enormous, funnel-shaped rotating auroral structures that form near Earth’s magnetic poles. Scientists named them for their strong visual resemblance to typhoons and tropical cyclones occurring in ocean regions. These storms develop inside the ionosphere and magnetosphere, causing navigation errors and degrading over-the-horizon radar performance considerably. Until now, however, researchers relied entirely on slow, subjective manual inspection of satellite imagery to find them.
To build the detection system, researchers trained it using 300,000 auroral images collected across both hemispheres between 2005 and 2021. The team deliberately included large numbers of ordinary aurora images that closely resemble space hurricanes to sharpen the model’s accuracy. Additionally, researchers selected over 500 confirmed storm events specifically as core training samples for the deep-learning algorithms. The resulting platform includes a full visual interface designed to streamline scientific workflows efficiently.
Lead researcher Zhang Qinghe emphasized that the system could shift space weather monitoring from passive response toward proactive early warning. The team also confirmed the AI tool integrates directly with data from the recently launched SMILE satellite, a joint mission between the Chinese Academy of Sciences and the European Space Agency. Consequently, SMILE’s continuous high-resolution auroral imaging will help researchers further refine the model over time.
Looking ahead, the team plans to develop short-term forecasting capabilities and establish an integrated space-air-ground monitoring network. Nevertheless, Zhang acknowledged that small-sample modeling challenges and limited understanding of storm formation mechanisms still present significant hurdles. Overcoming those obstacles remains essential before researchers can deliver reliable, real-time operational applications at a practical scale.
