AI weather forecasts capture path of storm

However, these models tended to underestimate the storm’s maximum wind speeds.

A recent study demonstrates that artificial intelligence (AI) has the capability to swiftly and accurately forecast the trajectory and intensity of major storms.

The study was conducted based on an analysis of Storm Ciaran in November 2023, the research indicates that weather predictions utilizing machine learning can match the precision of traditional forecasts while being faster, more cost-effective, and requiring less computational resources.

Published in npj Climate and Atmospheric Science by the University of Reading, the study underscores the rapid advancements and transformative potential of AI in weather forecasting. Professor Andrew Charlton-Perez, leading the study, remarks on the profound impact AI is having on weather prediction, noting the evolution from limited application of modern machine learning techniques to now having multiple models capable of generating global forecasts in a matter of minutes.

The study highlights both the promise and the challenges of AI-based weather forecasting. By comparing AI and physics-based forecasts of Storm Ciaran, which wreaked havoc in northern and central Europe, claiming lives and causing extensive damage, researchers found that AI models accurately predicted the storm’s rapid intensification and path up to 48 hours in advance.

However, these models tended to underestimate the storm’s maximum wind speeds. While the AI systems accurately captured large-scale atmospheric conditions influencing the storm’s development, certain features such as temperature contrasts near the storm center were not well-predicted.

To enhance the protection of communities from extreme weather events like Storm Ciaran, further investigation into the use of AI in weather prediction is imperative. Continued development of machine learning models could lead to the routine integration of artificial intelligence in weather forecasting, ultimately saving time and resources for forecasters.

-SOURCE – Eurekalert