Investigating the Influence of Inner Magnetosphere Data on a Regional Geomagnetically Induced Current Forecasting Model

The solar wind's interaction with the geomagnetic field causes geomagnetically induced currents (GICs) at ground level, which can be hazardous to power and communications infrastructure. Fast and accurate forecasts of perturbations in the horizontal component of the ground magnetic field (dB/dt) allow providers to best mitigate damage to their networks.

In this work, we examine a machine learning model that uses the method of spherical elementary current systems (SECS) to provide regional forecasts of dB/dt. The model is initially trained on solar wind input features; however, here we study the effect of incorporating additional training data from sources further within the Earth's magnetosphere, such as from the Defense Meteorological Satellite Program (DMSP). We examine the impact these new input features have on the overall performance of the model as well as their influence on the model's ability to characterize localized patterns in the spatial distribution of dB/dt.