Characterizing high-latitude ionospheric scintillation signatures with Machine Learning tools

Global Navigation Satellite Systems (GNSS) and other radio signals are affected by ionospheric irregularity structures that generate disruptions known as scintillation. In high latitudes, the plasma structures causing these rapid fluctuations in signal phase and amplitude are produced by various irregularity mechanisms, such as auroral precipitation, density gradients, and velocity shears. Ongoing research connects scintillation signatures to sources of irregularities in selected case studies through detailed physics and radio propagation models. In contrast, this study focuses on data-driven and Machine Learning (ML) methods to explore the link between high-latitude irregularity sources and their respective scintillation signatures. A database has been created containing high-rate (50Hz) scintillation signatures from the polar cap and auroral oval, major high-latitude source regions, during five geomagnetic storms. The classification ML model results for temporal scintillation signatures indicate that polar cap signatures can be distinguished from those originating in the auroral oval. We discuss the role of filtering in developing an ML-ready database of high-latitude 50Hz scintillation signatures, which adapts to specific irregularity scale sizes and dynamics, as well as its limitations.