Probabilistic Geomagnetic Storm Forecasting via Deep Learning

Geomagnetic storms, which are governed by the plasma magnetohydrodynamics of solar-interplanetary-magnetosphere system, entail a formidable challenge for physical forward modeling. Yet, abundance high-quality observational data has been amenable to application data-hungry neural networks geomagnetic storm forecasting. Almost all applications forecasting have utilized solar wind observations from Earth-Sun first Lagrangian point (L1) or closer and generated deterministic output without uncertainty estimates. Furthermore, work focused on indices that also sensitive induced internal magnetic fields, complicating problem with another layer non-linearity. We address these points, presenting trained both disk L1 point. Our architecture generates reliable probabilistic forecasts over Est, external component disturbance time index, showing can gauge confidence in their output.