From 1–5 August 2025, the medieval hilltop town of Erice, Sicily, hosted the second UNDERPIN workshop, a meeting organised within the Nonlinear Processes in Geosciences community and dedicated to advancing the science of weather extremes. The event brought together a truly diverse group of researchers, spanning climate dynamics, attribution science, socio-economic impacts, statistical physics, and mathematical modelling. This breadth reflected the complexity of extremes themselves, which demand insights from both theory and application, as well as the translation of research into societal benefit.
The program opened with a keynote by Rupert Stuart-Smith, who explored how attribution science can underpin climate litigation, showing how rigorous statistical analysis of extreme events can provide the evidence base for legal accountability. This was followed by Suzana Camargo, who presented a global perspective on tropical cyclone activity in both the present and future climate, with a focus on the implications for risk management and adaptation strategies. Over the following days, the keynote sessions continued to move between the applied and the theoretical. Tommaso Alberti examined how climate change influences aviation weather hazards, from turbulence to severe convective events. Erika Coppola discussed the ways in which urbanisation modifies the diurnal cycle of precipitation and temperature, with clear implications for local extremes. Gianmarco Mengaldo addressed the potential of nonlinear dynamics combined with explainable artificial intelligence for climate science, highlighting the dual challenge of harnessing AI for predictive skill while ensuring that its outputs remain physically interpretable. Vera Melinda Galfi closed the series with a theoretical perspective on the statistical mechanics underpinning weather and climate extremes.
Contributed talks reflected the same methodological diversity. Olivia Vashti Ayim introduced a novel method to quantify long-term trends in extreme events using medium-range forecasts, while Emma Holmberg presented an approach to forecasting heat-related mortality during the extraordinary European summer of 2022. Pablo Fernandez de Arroyabe discussed the development of biometeorological profiles for health-related early warnings, and Roberto Micciulla and Maria Calanni Macchio reported on public perception and awareness of climate change in Italy. The session also included Anupama Kuttikkat Xavier’s analogue-based analysis of North Pacific blocking predictability, Guido Occhipinti’s exploration of alternative dynamic regimes in marine biogeochemical cycles under extreme conditions, and Davide Valenti’s theoretical examination of whether complex system models can function without environmental noise.
The poster session fostered detailed discussions and interdisciplinary exchanges. Topics ranged from real-time detection of weather extremes to attribution methods for climate law, classification techniques for cyclones and anticyclones, and studies of precipitation variability in North Africa. My own poster addressed the predictability of cold–wet–windy Pan-Atlantic compound extremes, aiming to improve forecasts for multi-hazard events that simultaneously affect Europe and North America. A particularly memorable aspect of the workshop was its commitment to accessibility: sign language interpreters were present throughout, ensuring full participation for a deaf colleague and illustrating how inclusivity can strengthen scientific dialogue.
To wrap up the workshop, Mireia Ginesta led a lively closing discussion that brought together three key themes: where the biggest gaps in data and knowledge remain, how artificial intelligence can help (and the challenges and ethics it raises), and how science can better connect with policy, especially through attribution science. From this exchange, several research priorities crystallized — for example: integrating physical insight with statistical and AI tools to improve predictions of compound events; strengthening the bridge between science and policy to translate advances in attribution into real-world impact; expanding observational networks and open data to fill critical gaps; and fostering interdisciplinary collaboration to link climate science with societal impacts.
With its combination of keynote insights, methodological innovation, and open interdisciplinary exchange, UNDERPIN#2 was more than a sequence of talks. As part of the Nonlinear Processes in Geosciences community, it provided a collaborative forum where ideas moved freely between theory, computation, and application. The connections forged in Erice will continue to shape research directions, driving forward our collective effort to better understand, predict, and respond to the extremes of a rapidly changing climate.
