Space weather-driven ionospheric perturbations cause rapid amplitude and phase fluctuations in the radio signals that propagate through it. Such ‘scintillation’ events often compromise the fidelity and availability of Global Navigation Satellite System (GNSS) signals at ground based receivers, affecting services that rely on positioning and timing information provided by the system. Increasing reliance on GNSS and GNSS-based services has resulted in the need for accurate forecasts of space weather driven service disruptions. This is particularly true at high latitudes, where our understanding of space-weather effects on the ionosphere remains incomplete.
To tackle this challenge, a data-driven approach is employed to develop machine learning based models that utilize available historical datasets to produce a model capable of forecasting GNSS disturbances. Using a Sun-to-Earth system approach, the machine learning pipeline utilizes data from both space- and ground- based observations of solar activity, geomagnetic variations, and ionospheric conditions derived from GNSS signal data. Preliminary results for predicting ionospheric scintillations using various machine learning algorithms is presented along with relevant metrics that evaluate the utility of these techniques.