AI revolutionizes weather forecasts: More precise forecasts for extreme weather!

AI revolutionizes weather forecasts: More precise forecasts for extreme weather!

Winter in Germany is showing its capricious side this year: snow, ice, rain and sunshine alternate in rapid succession. Such changeable conditions often lead to criticism of weather apps and their forecasts. But the head of the forecast and advisory center of the German Weather Service, Franz Molé, emphasizes that the accuracy rate of the forecasts is quite high. The basis for this is the analysis of large amounts of data from lightning data, satellites and precipitation radars. The atmosphere is scanned in three dimensions at an impressive rate every five minutes, enabling precise short-term predictions.

The nowcasting technology makes it possible to monitor thunderstorms in real time, as can be seen in the German Weather Service's warning weather app. The forecast remains particularly challenging for small-scale or short-lasting phenomena, such as black ice or spontaneous thunderstorms. High-resolution models are in demand here, but often fail due to the available computing capacity. In fact, an increase in extreme events due to climate change is becoming increasingly clear, meaning that record temperatures in Germany could reach up to 45 degrees in the future.

Artificial intelligence in the weather system

One hope for improving weather forecasts is artificial intelligence (AI). This can operate at a speed that is 10,000 times faster than conventional numerical models. However, it does not require a physical image of the processes. The desired combination of physics-based models and AI should significantly improve predictions. A notable advance is the new AI weather model from the European Center for Medium-Range Weather Forecasts (ECMWF), which exceeds previous forecasts. Weather services around the world can use this method to predict extremes such as heavy rain or hurricanes more precisely.

The AI ​​model, known as AIFS (Artificial Intelligence Forecasting System), delivers results that are up to 20 percent more accurate on average. In certain scenarios, the improvement can even be up to 40 percent. Interestingly, the model reduces energy consumption by 1,000 times and calculates predictions ten times faster. In the future, weather services will be able to use new data for their forecasts and warnings, which will significantly benefit the general public.

A look into the future of climate research

Technologies such as Google DeepMind and OpenAI are developing the latest AI models that predict extreme weather events more precisely and detect CO₂ patterns. These models continuously learn from new data sets and can identify unexpected increases in CO₂ emissions. A European early warning system has already demonstrated success in predicting weather disasters with over 90 percent accuracy. It is clear that AI has the potential to fundamentally revolutionize climate research.

However, to realize the full potential of these new technologies, some ethical issues need to be addressed. Access to climate data and control over AI models must be made transparent to avoid inequalities in climate research. The development of clear rules for the use of AI in climate research is seen as urgent, especially since these models could also influence political decisions.

The progress in weather forecasting and climate research seems promising, but the challenge is to use these technologies with the necessary sense of responsibility and to include all social groups. Particular attention must also be paid to the traceability of decisions and data distortions to ensure fair and accurate application.