Towards a complete and quantitative description of the hazard at Campi Flegrei caldera with implications for the emergency planning

Hazard and risk management in densely populated volcanic areas requires the development of quantitative assessments. The Campi Flegrei caldera represents one of the highest-risk volcanic areas on the planet, due to its population density, and is currently in a phase of unrest characterized by ground deformations, seismicity, and gas emissions. Hazard assessment at Campi Flegrei is intrinsically complex, due to the difficulty in interpreting pre-eruptive patterns, and furthermore the caldera shows a great variability in vent locations and eruption sizes, not presenting a simple pattern that can be easily extrapolated to future activity. Here we develop a hazard model based on a Bayesian Event Tree (BET) that integrates eruption forecasting, scenario forecasting and impact forecasting by means of conditional probability rules. In particular: i) the model gives sense of heuristic pre-eruptive information through an entropy-based method and incorporates model heterogeneity through experts' elicitation; ii) takes into account vent and size variability through continuous probability distributions, overcoming the limitations of scenario-based approaches; iii) evaluates the impact of individual hazard phenomena by integrating different computational models. By propagating uncertainties across the BET nodes this approach allows for a transparent and consistent assessment of all possible outcomes in near real time, thus providing a tool that significantly facilitates risk management and decision-making in the Phlegraean area.