CoStErAS · Coastal Storm Erosion Alert Systems

Coastal storm-induced erosion is a pressing global issue threatening coastal communities, infrastructure, and ecosystems, increasingly exacerbated by climate change and human pressures. Yet, many regions—particularly the Atlantic coast of Europe—still lack operational Early Warning Systems (EWSs) able to predict storm-induced erosion, a key component for storm risk mitigation. The CoStErAS project addresses this gap by developing a predictive framework for storm-driven beach erosion that is both robust and transferable, designed for integration into EWSs. A comprehensive storm impact dataset will be built by combining existing long-term records (Australia, USA, Japan, France) with new European field data, satellite-derived shorelines, and XBeach simulations. This dataset will underpin the development of a new modelling framework that couples physics-informed feature engineering with advanced AI approaches (e.g., Bayesian Networks for explanatory/causal analysis, XGBoost for predictive skill optimization). This strategy avoids “black-box” pitfalls, ensuring models remain interpretable, transparent, and applicable across diverse coastal settings and tidal regimes. In data-rich areas, the model will provide reliable real-time predictions; in data-limited regions, it will deliver clear guidelines and strategies for operational implementation. Through this MSCA fellowship, I will be trained by world-leading experts at UNSW-WRL (outgoing phase) and EPOC Bordeaux (return phase), acquiring advanced skills in remote sensing, AI, and large-scale coastal modelling. This experience will also deepen my expertise in Atlantic European systems, strengthening my career goal of securing a permanent research position in France. Ultimately, CoStErAS will advance predictive modelling of storm-induced beach erosion, provide valuable insights for the scientific community, and deliver practical tools to improve the resilience of European—and global—coastal societies.