Global Ionospheric Total Electron Contents Modeling with Air-borne Observables and GAN Frameworks

The ionosphere services as an important medium for radio wave propagation, and the refraction phenomenon occurs for trans-ionosphere signals, resulting in propagation delay.  When the distribution of electron density in the ionosphere is uneven, amplitude attenuation and phase jitter will occur in trans-ionospheric signals, which are particularly prominent for global satellite navigation signals.  This coupling mechanism makes it possible to use radio waves for ionosphere remote sensing. Currently, international GNSS services use nearly 500 observation stations globally to derive the spatial and temporal feature of global ionospheric total electron content. The resulting product, global ionospheric map has become the major data source for ionosphere study. Moreover, the University Corporation for Atmospheric Research in America took the lead in the use of Low Earth Orbit occultation observations to measure the vertical profile of ionospheric electron density and scintillation effects in the ionosphere. Although existing ionosphere remote sensing techniques have been developed, the fundamental problem is still difficult to solve. For ground-based sensing, there are blind zones that ground stations cannot reach. For instance, it is difficult to deploy ground receivers in deserts and oceans. For space-borne techniques, such as radio occultation, it is mainly applied in middle to low latitude areas, and the observations cannot be continuously in both spatial and temporal domains.   To address the problem, it is helpful to get more sources of ionospheric observations. It is clear that with dual frequency measurements, the accurate ionospheric total electron content can be solved at the pierce point. Therefore, with comprehensive satellite signals, the development of multiple constellations in satellite navigation can make benefits for accurate ionosphere sensing and modeling. In this work, we focus on the application and contribution of air-borne observations in ionosphere sensing and modeling. Supposing the air-borne dual-frequency GNSS receivers can derive accurate total electron content, more ionospheric pierce points are obtained with global air flights. The feasibility of air-borne ionospheric sensing and modeling are further investigated. We apply Generative Adversarial Networks to reconstruct the regional ionospheric TEC map with plentiful air-borne ionospheric observables from airlines equipping dual frequency GNSS receivers. The proposed ionospheric TEC maps are verified by comparison with the IGS global ionospheric map. Several GAN-based deep learning networks are applied, such as DCGAN and SNP-GAN. This study is benefit for exploring the feasibility of air-borne ionospheric sensing and modeling, as supplement for ground and space-borne ionospheric sensing systems.