Forecasting space weather risks on power grids
AI Summary: A machine learning pipeline was developed to generate location-specific risk estimates for 66,935 substations in the continental United States, combining forecast-time solar-wind information with local latitude, geology, and ground conductivity. The system uses forecasts of the Auroral Electrojet (AE) and Disturbance Storm Time (Dst) indices, along with physics-informed constraints, to produce risk estimates 30-60 minutes ahead of potential impact. The pipeline detected nearly 80% of major space-weather events during the evaluation period and can warn grid operators of specific risks. The system was built using public data sources and a gradient-boosting model.