Machine Learning Classification of Ionosphere and RFI Disturbances in Spaceborne GNSS Radio Occultation Measurements

Low earth orbiting, GNSS radio occultation small satellite constellations provide a unique platform to profile the atmosphere. However, such a platform is susceptible to radio interference, orbit instability and other processing artifacts less common to large scale, high expense and/or higher orbital altitude missions. These disturbances are challenging to differentiate using traditional phase or amplitude disturbance triggers, leading to costly false alerts in hazard monitoring or missed opportunities for science. In this work we apply a combination of physics-based feature engineering with data-driven supervised machine learning to improve classification of low earth orbit Spire Global GNSS radio occultation disturbances for refining earth insight from these data.