Artificial intelligence will be key to mitigating the impacts of climate change
An international research team, co-led by the University of Valencia (UV), has just published an article in Nature Communications reviewing the use of artificial intelligence (AI) to deepen the understanding of extreme climate phenomena - floods, droughts, wildfires and heatwaves - with the aim of developing more reliable and precise prediction systems. The ultimate goal of the study is to improve responses to such disasters and enhance the effectiveness of risk communication.
The devastating floods that struck Valencia last October, the wildfires that ravaged Los Angeles in January or the unprecedented heatwaves and droughts in Europe are just a few recent examples of how extreme weather events are becoming increasingly frequent and intense, as highlighted in the latest report by the United Nations' Intergovernmental Panel on Climate Change (IPCC).
In response to this global challenge, a study published in Nature Communications, led by the UV professor Gustau Camps-Valls, proposes the use of AI as a key tool not only for predicting but also for understanding and mitigating the impact of extreme climate events on nature and people's lives.
The article examines how AI is increasing the progress of weather prediction systems, early detection of extreme events and impact forecasting. It is also enabling the development of transparent and comprehensible explanations that inspire confidence in decision-making when facing such climate phenomena. "These studies show that AI is not just a tool for improving weather predictions but also an ally in estimating their impacts, understanding extreme climate events and effectively communicating risks", explains Gustau Camps-Valls, physicist, professor of Electronic Engineering at the UV and researcher at the Image & Signal Processing (ISP) group within the Image Processing Laboratory (IPL) of the University of Valencia.
The study demonstrates how AI has reached a certain level of maturity in weather forecasting, the detection and modelling of extreme events and risk communication through advanced chatbots, largely thanks to deep learning, statistical modelling and processing of satellite data, in-situ measurements and climate simulations.
The increasing frequency and intensity of these phenomena due to climate change are providing more data to train AI systems, which in turn enhances their ability to anticipate future events and adapt predictions to the planet's evolving conditions. "Thanks to AI, we are not only improving the detection of extreme weather events but also gaining a better understanding of their causes and consequences. This allows us to make more responsible and informed decisions", adds Camps-Valls.
An interdisciplinary approach to a global challenge
The study emphasises that, for AI to fulfil its role in climate risk management, certain technical, ethical and operational barriers must be overcome. Greater transparency and explainability in AI models, better integration of AI into early warning systems and more effective risk communication are some of the key challenges.
To address these issues, the article calls for interdisciplinary collaboration between climate scientists, AI experts, policymakers and local stakeholders, asserting that only through joint efforts can accessible, ethical and context-specific solutions be designed. "We are laying the foundations for a technology that not only analyses data but also has the potential to change lives and protect our planet. However, it is crucial that these tools are inclusive, ethical and accessible to all communities", concludes Camps-Valls.
Finally, the study highlights the importance of expert knowledge from both the scientific community and policymakers and vulnerable groups, to correctly interpret results and implement actions in the right context, ensuring the use of technology as a tool at the service of society and the planet.
Entitled "AI for Extreme Events Modeling and Understanding: Methodologies and Challenges", this study, published in Nature Communications, is the result of a workshop held in Valencia in 2023, which brought together leading experts in artificial intelligence and climate sciences as part of the ELLIS Programme on Machine Learning for Earth and Climate Sciences.
The research team is currently working on a new dimension of early warning systems, integrating AI at all stages, which could be crucial for the planning of resilient infrastructure, risk communication and the development of long-term climate adaptation policies.