Forecasting space weather risks on power grids
TL;DR
Extreme space-weather events can damage power systems on Earth and degrade GPS accuracy and satellite operations. A new machine learning system can predict where damage is likely to occur 30-60 minutes before a storm arrives. The post Forecasting space weather risks on power grids appeared first on Microsoft Research.
Nauti's Take
For small teams, the first test is data quality: check whether the system provides local measurements, clear alert levels, and usable uncertainty estimates before connecting it to operational decisions. A 30 to 60 minute warning window matters only if false alarms, regional coverage, and transferability to your own infrastructure are documented under real-world conditions.
Sources