Tropical cyclones (TCs) are among the most devastating weather phenomena on Earth, frequently causing numerous fatalities and leading to extensive and costly damage. Effective evacuation planning and the implementation of precautionary measures depend on forecasts that are both highly accurate and available with sufficient lead time. Furthermore, TCs are of particular interest to operational forecasting centres, as they are characterized by extreme deviations from the mean atmospheric state. As such, they pose a significant challenge to numerical weather prediction systems and serve as a rigorous stress test for any forecasting model, making their evaluation a key focus area.

While speculation persisted until not long ago that the predictability limit for TC track forecasts had been reached, dedicated research efforts were launched to pool resources and advance intensity forecasts in physics-based models. With the advent of data-driven weather forecasting, however, previously assumed forecast limits have been exceeded in certain aspects of TC prediction, while others have deteriorated. Current model developments aim to fuse both worlds into a hybrid modelling approach, which has shown promising results.

In this presentation, I will give an overview of the latest research and developments in TC forecasting with ECWMF’s physics-based and data-driven models.