In Bologna, Italy, supercomputers inside a former tobacco factory crunch weather data daily to help energy traders make informed decisions. However, a new AI model developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) is changing the game.
This AI model not only uses real-time data but also incorporates historical information, resulting in more accurate predictions of temperature, precipitation, wind, and tropical cyclones. The model consumes less computing energy and provides forecasts much faster than traditional methods. Energy traders benefit from these advancements by responding quickly to weather changes, minimizing energy surpluses and shortages. The AI model’s two-week forecasts help companies and policymakers make faster decisions, such as canceling rail services or dispatching trucks for road safety.
The AI-driven approach marks a significant shift from conventional methods, leveraging vast amounts of climate data for improved accuracy. Despite the rapid progress, experts believe a hybrid system combining AI and traditional forecasts will be the most effective. ECMWF’s next step involves integrating AI models with satellite and weather station data and exploring new sources of weather information. These advances promise to increase forecast update frequency and improve performance, ultimately benefiting the energy market.