Improving Mediterranean CYCLOnes Predictions in Seasonal forecasts with artificial intelligence

Cyclones form frequently over the Mediterranean Sea. The most intense systems cause extensive damage in the region and beyond. The ability to make climate predictions several months in advance of such extreme events has a large number of crucial socio-economic applications for disaster risk reduction.

State-of the-art Seasonal Prediction Systems exhibit a good skill in predicting anomalies in the seasonal mean of meteorological fields such as temperature and precipitation. The models’ ability to reproduce variations in the occurrence of extreme events tends to be however much lower. This is particularly true in regions such as the Mediterranean, where extreme events are often driven by small scale processes that are not well reproduced at the resolution at which SPS typically run.

The use of artificial intelligence techniques such as machine learning in the study of climate has gained great traction in recent years. The power of those techniques lies in the ability to detect patterns in large datasets without having to make explicit statistical assumptions, allowing to build predictive model whose skill continuously improves as the volume of data on which the models are trained increases. One promising field of application is to use artificial intelligence to improve the prediction of extremes in climate models. In this setting a machine learning model is trained to find relationships between large-scale climate variables (for which dynamical models have a good predictive skill) and the occurrence of extremes.

The aim of this project is to improve the predictability of intense Mediterranean cyclones combining advanced artificial intelligence techniques and a state-of-the-art dynamical seasonal prediction system. The project will not only contribute to the knowledge of climate extremes predictability, but also lead to the implementation of a pre-operational dynamical-statistical prediction system with the potential to be extended to include different extremes.