AI Weather Model Reshapes Energy Trading

A new AI-driven weather model developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) is beginning to influence how energy traders operate in increasingly volatile markets, as first reported by Bloomberg.

The model, which incorporates historical data alongside conventional meteorological measurements, is designed to enhance the accuracy and speed of weather predictions.

Every night in Bologna, Italy, supercomputers at the ECMWF analyze massive atmospheric data to create weather forecasts. Released six hours later, these predictions have long guided power and gas traders in managing energy supply and demand, according to Bloomberg.

The introduction of AI forecasting, however, marks a shift in approach, offering results in a fraction of the time.

“We can update our information set more often than we are used to,” said Daniel Borup, chief executive officer of Danish trading firm InCommodities A/S, as reported by Bloomberg. “That obviously leads to improvements in our predictions. It allows us to improve our job and distribute energy better,” he added.

Bloomberg says that the AI model, released last month, is the first of its kind developed by a major forecasting institution.

Unlike traditional simulations that rely solely on satellite and sensor data, the new system incorporates patterns from past weather events, allowing for more precise predictions of temperature, precipitation, wind conditions, and even tropical cyclones.

Initial testing suggests the AI-generated forecasts perform at least as well as conventional models while requiring less computing power, as reported by Bloomberg.

Traders and policymakers are particularly interested in its potential applications for managing renewable energy output, where sudden shifts in wind and solar generation can affect electricity prices, as noted by Bloomberg.

This has been evident in recent months. Earlier this year, a prolonged period of low wind and solar generation—known as a Dunkelflaute—led to price surges in Germany. More recently, an oversupply of solar power resulted in negative electricity prices, as reported by Bloomberg.

The forecasting model’s faster output could improve market responsiveness. “It’s what the markets move on most,” said Dan Harding, a meteorologist at European weather analytics firm MetDesk, reports Bloomberg.

The AI system’s ability to generate forecasts in three minutes, compared to the 30 minutes required by traditional computing models, could allow for quicker decisions in energy trading and grid management.

Bloomberg notes that despite its advantages, experts note that AI forecasting has limitations. It remains less reliable than traditional models in predicting cloud cover, dust, and certain extreme weather patterns. ECMWF forecasters anticipate a gradual transition toward hybrid forecasting methods, combining AI with conventional approaches.

“AI weather models have the potential to increase the frequency of forecast updates and improve performance,” said Edoardo Simioni, head of trading at Copenhagen-based electricity supplier Reel ApS, as reported by Bloomberg. “The advances in technology are ultimately good for the market,” he added.