AI-Powered Predictions Bring a New Dimension to Tropical Cyclone Forecasts

A new study led by University of Miami researchers compared traditional weather forecasting models with the European Centre for Medium-Range Weather Forecasts' Artificial Intelligence/Machine Learning Forecasting System. The research showed that artificial intelligence models can provide experts with significant additional insights into tropical cyclone development processes in the Atlantic Ocean.

Scope and Methodology of the Research

Professor Sharan Majumdar and his colleagues from the University of Miami Rosenstiel School of Marine, Atmospheric, and Earth Science examined African-easterly waves and tropical cyclone developments in the Atlantic basin between 2020 and 2024.

Comparison of AI and Traditional Systems

The European Centre for Medium-Range Weather Forecasts' traditional Integrated Forecasting System was compared in detail with the AI-based forecasting system. The analyses revealed that the artificial intelligence system offers alternative perspectives at different time horizons.

Particularly for stronger tropical waves, the AI ensemble frequently produced higher development probabilities at lead times compared to the traditional system.

Advantages in Location Forecasts

The artificial intelligence system also demonstrated distinct advantages in predicting the location of developing tropical systems. The average track error of the AI ensemble mean was measured as generally lower compared to other models.

Future Weather Forecasts

The findings do not suggest that artificial intelligence should completely replace traditional numerical weather prediction. Experts state that the two systems working together can provide meteorologists with a more comprehensive picture.