The ionosphere, located between 60 and 1000 kilometers above Earth, significantly influences communication and navigation systems, particularly global navigation satellite systems (GNSS), by disrupting radio signal transmission due to its ionized particles. Recent advancements in neural networks have led to improved models for reconstructing the plasma density in this region, thereby enhancing the accuracy of satellite navigation. Researchers Artem Smirnov and Yuri Shprits from the GFZ German Research Centre introduced a new model based on machine learning, utilizing 19 years of satellite data to more precisely analyze the topside ionosphere. This innovative model, known as the NET model, significantly outperforms the existing International Reference Ionosphere Model (IRI) in terms of accuracy and comprehensive coverage. It effectively reconstructs electron density across various heights and conditions, accommodating fluctuations in solar activity and seasonal variations. The NET model has far-reaching applications in various fields, including wave propagation studies, long-term ionospheric monitoring, and space weather analysis. Its architecture allows for the inclusion of new data, marking a significant step forward in ionospheric research and enhancing the reliability of communication and navigation systems that depend on GNSS.
