Aso volcano: Estimating the probabilistic likelihood of a future Aso4-scale eruption from stochastic uncertainty analysis of volcanological evidence using importance sampling

The Aso4 explosive eruption on Kyushu, Japan, 89,500 years ago was one of the biggest global eruptions in the last one hundred millennia, with a magnitude of approximately M8. Its effects were widespread throughout Japan; a similar scale event now would have huge societal impact. Today, 6.5 million persons live within 100 km of Aso. The key question for disaster preparedness and mitigation is: what is the likelihood of another M8 eruption from Aso-san? We estimate the probability of such an event within the next 100 years so that the scenario and its threats can be compared to other potential natural disasters. To evaluate this probability, we performed a comprehensive stochastic uncertainty analysis using advanced computational Bayes Net (BN) software. Our BN eruption process model is informed by multiple strands of evidence from volcanology, petrology, geochemistry and geophysics, together with estimates of epistemic (knowledge) uncertainty, adduced from reviews of published data, modelling and from expert judgement elicitation. Several lines of scientific evidence characterise the likely structure, magma composition and eruptibility state of the present-day volcano, which has had numerous smaller eruptions since Aso4. An initial analysis indicated that another Aso4-scale event has an extremely low likelihood, being less than 1 - in - 10 million in the next 100 years (i.e. < 10-7 probability). To further constrain this probability, we implemented probabilistic ‘importance sampling’ in our BN to allow even smaller probabilities to be enumerated. We find that the chance of an Aso4-scale eruption (characterised by mean volume 500 km3 DRE, 90% credible interval [370 .. 685] km3 DRE) is less than 1 – in – 1 billion in the next 100 years (i.e. <10-9 probability). We doubt that this conclusion, based on current understanding and evidence, could be different by more than an order of magnitude.