Reshaping the link between science, society, and the natural world

The interaction between humans and the natural environment is key to understanding both our impact on the planet and how a changing climate will shape our lives and economic activities. Science and innovation play a crucial role in this process, helping to build resilience and strengthen early warning systems.

From forecasting El Niño months in advance to anticipating cyclones and streamflow extremes, artificial intelligence is reshaping the way we understand and respond to climate risks. In this issue, we dive into cutting-edge research and projects where CMCC scientists are using AI to close the gap between knowledge and action — transforming early warning systems, enhancing climate resilience, and ultimately strengthening the link between science, society, and the natural world.

Predicting El Niño events with AI

With recent developments in AI we can now predict El Niño events up to 3-4 months in advance, which could prove fundamental in setting up effective early warning systems, revolutionizing the way marine ecosystems are managed and securing the long-term sustainability of their resources.

As outlined in new research led by CMCC researcher Marie-Lou Bachèlery, AI is used to predict extreme Atlantic Niño and Benguela Niño events, which have substantial impacts on the tropical Atlantic region, with far-reaching consequences on local marine ecosystems, African climates, and the El Niño Southern Oscillation.

This is a notable development as even state-of-the-art dynamic forecasting systems face significant challenges when attempting to predict tropical Atlantic climate events, highlighting AI’s potential to advance our understanding and forecasting.

The study not only accurately predicted the strong 2021 Atlantic/Benguela Niño events 4 months before they occurred, when traditional systems missed them entirely, but also provides a promising first step in the application of AI methods for a wide range of climate phenomena and other important variables such as oxygen levels and fishery productivity.

AI for enhancing seasonal predictions of Mediterranean cyclones

Intense cyclones in the Mediterranean can lead to severe impacts — from flash floods and storm surges to wind damage — threatening lives, infrastructure, and economies not only across southern Europe but even as far as Central Europe. While short-term forecasts of these events have improved over the years, predicting them at seasonal scale remains a challenge due to current model limitations.

In a recent CMCC webinar, researcher Leone Cavicchia explores how Artificial Intelligence is being used to boost the accuracy of seasonal predictions for these high-impact weather events.

At the core of this work is CYCLOPS, a project coordinated by CMCC, that applies machine learning to improve the performance of a state-of-the-art seasonal prediction system for predicting Mediterranean cyclones. By training AI models on historical climate data and applying them to CMCC’s operational seasonal forecasts, the team has demonstrated encouraging improvements in predictive skill.

Why does this matter? Better seasonal forecasts of extreme events can significantly support disaster risk management, climate resilience planning, as well as strengthening key sectors like insurance and reinsurance, ultimately helping society stay one step ahead of climate-related risks.

AI meets hydrology: Smarter forecasts for a climate-ready future

Accurate streamflow predictions are essential for managing water resources, preparing for floods and droughts, and protecting ecosystems. However, traditional large-scale hydrological models often fall short when it comes to capturing local complexities — from shifting weather patterns to human impacts on water systems.

In a recent CLINT–MedEWSa webinar, Yiheng Du (Swedish Meteorological and Hydrological Institute), moderated by Antonello Squintu (CMCC), explores how AI and Machine Learning are revolutionizing the way we predict streamflow across Europe. By combining traditional hydrological modeling with ML techniques, researchers are improving prediction accuracy even in data-scarce or ungauged regions.

These advancements are part of the broader CLINT project, where CMCC and partners are using AI to improve the detection, understanding, and prediction of extreme events such as tropical cyclones and heatwaves. From refining global indices for cyclone activity to enhancing forecasts of extreme precipitation, CLINT is helping make climate intelligence smarter, faster, and more actionable.

In parallel, the MedEWSa project is tackling natural hazards on a global scale. By integrating and upgrading early warning systems across eight pilot sites in Europe, Africa, and the Mediterranean, MedEWSa is creating a powerful, impact-based multi-hazard early warning system. From wildfires in Greece and Ethiopia to coastal flooding in Venice and Alexandria, this initiative connects regions facing similar risks to promote shared solutions and stronger resilience.

Together, these efforts show how AI-driven climate science is paving the way for more accurate predictions, better preparedness, and a safer future.